2023-03-07 10:14:36,164 INFO [train2.py:879] (0/4) Training started 2023-03-07 10:14:36,164 INFO [train2.py:880] (0/4) {'frame_shift_ms': 10.0, 'allowed_excess_duration_ratio': 0.1, 'best_train_loss': inf, 'best_valid_loss': inf, 'best_train_epoch': -1, 'best_valid_epoch': -1, 'batch_idx_train': 0, 'log_interval': 50, 'reset_interval': 200, 'valid_interval': 3000, 'feature_dim': 80, 'subsampling_factor': 4, 'warm_step': 2000, 'env_info': {'k2-version': '1.23.3', 'k2-build-type': 'Debug', 'k2-with-cuda': True, 'k2-git-sha1': '3b81ac9686aee539d447bb2085b2cdfc131c7c91', 'k2-git-date': 'Thu Jan 26 20:40:25 2023', 'lhotse-version': '1.9.0.dev+git.97bf4b0.dirty', 'torch-version': '1.10.0+cu102', 'torch-cuda-available': True, 'torch-cuda-version': '10.2', 'python-version': '3.8', 'icefall-git-branch': 'surt', 'icefall-git-sha1': 'e9931b7-dirty', 'icefall-git-date': 'Fri Mar 3 16:27:17 2023', 'icefall-path': '/exp/draj/mini_scale_2022/icefall', 'k2-path': '/exp/draj/mini_scale_2022/k2/k2/python/k2/__init__.py', 'lhotse-path': '/exp/draj/mini_scale_2022/lhotse/lhotse/__init__.py', 'hostname': 'r8n04', 'IP address': '10.1.8.4'}, 'beam_size': 10, 'reduction': 'sum', 'use_double_scores': True, 'world_size': 4, 'master_port': 12368, 'tensorboard': True, 'num_epochs': 30, 'start_epoch': 1, 'start_batch': 0, 'exp_dir': PosixPath('zipformer_ctc_att/exp/v0'), 'lang_dir': PosixPath('data/lang_bpe_500'), 'base_lr': 0.05, 'lr_batches': 5000, 'lr_epochs': 3.5, 'att_rate': 0.8, 'num_decoder_layers': 6, 'seed': 42, 'print_diagnostics': False, 'inf_check': False, 'save_every_n': 2000, 'keep_last_k': 10, 'average_period': 200, 'use_fp16': True, 'num_encoder_layers': '2,4,3,2,4', 'feedforward_dims': '1024,1024,2048,2048,1024', 'nhead': '8,8,8,8,8', 'encoder_dims': '384,384,384,384,384', 'attention_dims': '192,192,192,192,192', 'encoder_unmasked_dims': '256,256,256,256,256', 'zipformer_downsampling_factors': '1,2,4,8,2', 'cnn_module_kernels': '31,31,31,31,31', 'full_libri': True, 'manifest_dir': PosixPath('data/manifests'), 'max_duration': 1000, 'bucketing_sampler': True, 'num_buckets': 30, 'concatenate_cuts': False, 'duration_factor': 1.0, 'gap': 1.0, 'on_the_fly_feats': False, 'shuffle': True, 'drop_last': True, 'return_cuts': True, 'num_workers': 2, 'enable_spec_aug': True, 'spec_aug_time_warp_factor': 80, 'enable_musan': True, 'input_strategy': 'PrecomputedFeatures'} 2023-03-07 10:14:36,469 INFO [lexicon.py:168] (0/4) Loading pre-compiled data/lang_bpe_500/Linv.pt 2023-03-07 10:14:37,049 INFO [train2.py:902] (0/4) About to create model 2023-03-07 10:14:37,534 INFO [zipformer.py:178] (0/4) At encoder stack 4, which has downsampling_factor=2, we will combine the outputs of layers 1 and 3, with downsampling_factors=2 and 8. 2023-03-07 10:14:37,595 INFO [train2.py:906] (0/4) Number of model parameters: 86083707 2023-03-07 10:14:42,480 INFO [train2.py:921] (0/4) Using DDP 2023-03-07 10:14:42,770 INFO [asr_datamodule.py:420] (0/4) About to get the shuffled train-clean-100, train-clean-360 and train-other-500 cuts 2023-03-07 10:14:42,878 INFO [asr_datamodule.py:224] (0/4) Enable MUSAN 2023-03-07 10:14:42,878 INFO [asr_datamodule.py:225] (0/4) About to get Musan cuts 2023-03-07 10:14:44,364 INFO [asr_datamodule.py:249] (0/4) Enable SpecAugment 2023-03-07 10:14:44,364 INFO [asr_datamodule.py:250] (0/4) Time warp factor: 80 2023-03-07 10:14:44,364 INFO [asr_datamodule.py:260] (0/4) Num frame mask: 10 2023-03-07 10:14:44,364 INFO [asr_datamodule.py:273] (0/4) About to create train dataset 2023-03-07 10:14:44,364 INFO [asr_datamodule.py:300] (0/4) Using DynamicBucketingSampler. 2023-03-07 10:14:46,721 INFO [asr_datamodule.py:316] (0/4) About to create train dataloader 2023-03-07 10:14:46,722 INFO [asr_datamodule.py:440] (0/4) About to get dev-clean cuts 2023-03-07 10:14:46,723 INFO [asr_datamodule.py:447] (0/4) About to get dev-other cuts 2023-03-07 10:14:46,723 INFO [asr_datamodule.py:347] (0/4) About to create dev dataset 2023-03-07 10:14:47,006 INFO [asr_datamodule.py:364] (0/4) About to create dev dataloader 2023-03-07 10:15:00,130 INFO [train2.py:809] (0/4) Epoch 1, batch 0, loss[ctc_loss=5.359, att_loss=1.249, loss=2.071, over 15506.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008071, over 36.00 utterances.], tot_loss[ctc_loss=5.359, att_loss=1.249, loss=2.071, over 15506.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008071, over 36.00 utterances.], batch size: 36, lr: 2.50e-02, grad_scale: 2.0 2023-03-07 10:15:00,132 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 10:15:12,517 INFO [train2.py:843] (0/4) Epoch 1, validation: ctc_loss=5.278, att_loss=1.419, loss=2.191, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 10:15:12,518 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 10964MB 2023-03-07 10:15:18,031 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5.0, num_to_drop=2, layers_to_drop={0, 3} 2023-03-07 10:15:43,070 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:16:22,267 INFO [train2.py:809] (0/4) Epoch 1, batch 50, loss[ctc_loss=1.213, att_loss=1.034, loss=1.07, over 16771.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006407, over 48.00 utterances.], tot_loss[ctc_loss=2.243, att_loss=1.105, loss=1.333, over 731045.98 frames. utt_duration=1416 frames, utt_pad_proportion=0.02243, over 2067.48 utterances.], batch size: 48, lr: 2.75e-02, grad_scale: 2.0 2023-03-07 10:16:35,786 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=53.41 vs. limit=5.0 2023-03-07 10:17:06,712 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:17:21,429 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6844, 5.7059, 5.5049, 5.5902, 4.0259, 5.6582, 4.5094, 5.7286], device='cuda:0'), covar=tensor([0.0093, 0.0059, 0.0098, 0.0182, 0.0141, 0.0069, 0.0045, 0.0048], device='cuda:0'), in_proj_covar=tensor([0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([9.1298e-06, 9.3010e-06, 9.2167e-06, 9.4481e-06, 9.0769e-06, 9.2244e-06, 9.0779e-06, 9.2573e-06], device='cuda:0') 2023-03-07 10:17:31,214 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.328e+01 1.076e+02 2.257e+02 4.349e+02 2.620e+03, threshold=4.514e+02, percent-clipped=0.0 2023-03-07 10:17:31,258 INFO [train2.py:809] (0/4) Epoch 1, batch 100, loss[ctc_loss=1.253, att_loss=1.06, loss=1.098, over 17135.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01326, over 56.00 utterances.], tot_loss[ctc_loss=1.648, att_loss=1.041, loss=1.163, over 1298504.34 frames. utt_duration=1324 frames, utt_pad_proportion=0.0362, over 3926.04 utterances.], batch size: 56, lr: 3.00e-02, grad_scale: 2.0 2023-03-07 10:18:30,870 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1041, 5.1548, 5.1389, 5.1265, 5.1516, 5.1207, 5.1074, 5.1497], device='cuda:0'), covar=tensor([0.0009, 0.0022, 0.0048, 0.0012, 0.0029, 0.0022, 0.0010, 0.0023], device='cuda:0'), in_proj_covar=tensor([0.0008, 0.0008, 0.0008, 0.0008, 0.0008, 0.0009, 0.0008, 0.0008], device='cuda:0'), out_proj_covar=tensor([8.1901e-06, 8.3179e-06, 8.2548e-06, 8.2292e-06, 8.4302e-06, 8.3979e-06, 8.4542e-06, 8.3302e-06], device='cuda:0') 2023-03-07 10:18:33,459 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=144.0, num_to_drop=2, layers_to_drop={0, 3} 2023-03-07 10:18:42,513 INFO [train2.py:809] (0/4) Epoch 1, batch 150, loss[ctc_loss=1.053, att_loss=0.8623, loss=0.9005, over 11607.00 frames. utt_duration=1859 frames, utt_pad_proportion=0.1736, over 25.00 utterances.], tot_loss[ctc_loss=1.45, att_loss=1.013, loss=1.1, over 1735260.73 frames. utt_duration=1310 frames, utt_pad_proportion=0.039, over 5304.54 utterances.], batch size: 25, lr: 3.25e-02, grad_scale: 2.0 2023-03-07 10:19:05,503 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.57 vs. limit=2.0 2023-03-07 10:19:48,619 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.064e+01 7.732e+01 1.287e+02 1.925e+02 9.711e+02, threshold=2.575e+02, percent-clipped=2.0 2023-03-07 10:19:48,663 INFO [train2.py:809] (0/4) Epoch 1, batch 200, loss[ctc_loss=1.128, att_loss=0.92, loss=0.9616, over 16463.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006091, over 46.00 utterances.], tot_loss[ctc_loss=1.353, att_loss=0.9952, loss=1.067, over 2083396.54 frames. utt_duration=1287 frames, utt_pad_proportion=0.04116, over 6480.97 utterances.], batch size: 46, lr: 3.50e-02, grad_scale: 2.0 2023-03-07 10:19:55,758 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=7.88 vs. limit=2.0 2023-03-07 10:20:54,630 INFO [train2.py:809] (0/4) Epoch 1, batch 250, loss[ctc_loss=1.242, att_loss=1.021, loss=1.065, over 17057.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009383, over 53.00 utterances.], tot_loss[ctc_loss=1.3, att_loss=0.9866, loss=1.049, over 2358733.86 frames. utt_duration=1293 frames, utt_pad_proportion=0.03556, over 7304.84 utterances.], batch size: 53, lr: 3.75e-02, grad_scale: 2.0 2023-03-07 10:21:54,654 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=296.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:21:59,471 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=300.0, num_to_drop=2, layers_to_drop={0, 3} 2023-03-07 10:22:00,457 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.252e+01 6.827e+01 1.043e+02 1.996e+02 1.331e+03, threshold=2.085e+02, percent-clipped=17.0 2023-03-07 10:22:00,503 INFO [train2.py:809] (0/4) Epoch 1, batch 300, loss[ctc_loss=1.216, att_loss=0.9801, loss=1.027, over 16758.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.007013, over 48.00 utterances.], tot_loss[ctc_loss=1.268, att_loss=0.9793, loss=1.037, over 2570459.40 frames. utt_duration=1287 frames, utt_pad_proportion=0.03428, over 7998.30 utterances.], batch size: 48, lr: 4.00e-02, grad_scale: 2.0 2023-03-07 10:22:40,704 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=11.73 vs. limit=5.0 2023-03-07 10:23:06,528 INFO [train2.py:809] (0/4) Epoch 1, batch 350, loss[ctc_loss=1.22, att_loss=0.9818, loss=1.03, over 17070.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.00876, over 53.00 utterances.], tot_loss[ctc_loss=1.241, att_loss=0.968, loss=1.023, over 2729725.36 frames. utt_duration=1276 frames, utt_pad_proportion=0.03835, over 8569.88 utterances.], batch size: 53, lr: 4.25e-02, grad_scale: 2.0 2023-03-07 10:23:14,309 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=357.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 10:23:54,006 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=387.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:24:12,778 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 4.983e+01 8.126e+01 1.923e+02 2.981e+02 6.249e+02, threshold=3.846e+02, percent-clipped=46.0 2023-03-07 10:24:12,823 INFO [train2.py:809] (0/4) Epoch 1, batch 400, loss[ctc_loss=1.06, att_loss=0.864, loss=0.9033, over 15632.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009104, over 37.00 utterances.], tot_loss[ctc_loss=1.219, att_loss=0.9579, loss=1.01, over 2847490.94 frames. utt_duration=1238 frames, utt_pad_proportion=0.05136, over 9214.11 utterances.], batch size: 37, lr: 4.50e-02, grad_scale: 4.0 2023-03-07 10:24:28,995 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([0.9215, 3.4542, 3.7296, 4.2843, 3.7752, 4.1045, 2.7550, 3.6675], device='cuda:0'), covar=tensor([0.2656, 0.0276, 0.0345, 0.0083, 0.0070, 0.0043, 0.0268, 0.0041], device='cuda:0'), in_proj_covar=tensor([0.0008, 0.0007, 0.0007, 0.0007, 0.0007, 0.0007, 0.0007, 0.0007], device='cuda:0'), out_proj_covar=tensor([8.1428e-06, 7.7461e-06, 7.5910e-06, 6.9688e-06, 7.7912e-06, 7.3126e-06, 7.6485e-06, 7.2962e-06], device='cuda:0') 2023-03-07 10:25:02,035 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=439.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 10:25:13,928 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=448.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 10:25:17,463 INFO [train2.py:809] (0/4) Epoch 1, batch 450, loss[ctc_loss=1.25, att_loss=0.9906, loss=1.043, over 17097.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01557, over 56.00 utterances.], tot_loss[ctc_loss=1.196, att_loss=0.9445, loss=0.9947, over 2932009.39 frames. utt_duration=1213 frames, utt_pad_proportion=0.06157, over 9681.90 utterances.], batch size: 56, lr: 4.75e-02, grad_scale: 4.0 2023-03-07 10:25:44,906 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.11 vs. limit=5.0 2023-03-07 10:25:45,333 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2102, 5.6788, 5.3160, 5.5907, 5.6160, 5.7558, 5.4341, 5.3246], device='cuda:0'), covar=tensor([0.0047, 0.0093, 0.0317, 0.0172, 0.0089, 0.0152, 0.0532, 0.1048], device='cuda:0'), in_proj_covar=tensor([0.0008, 0.0008, 0.0009, 0.0008, 0.0008, 0.0008, 0.0009, 0.0009], device='cuda:0'), out_proj_covar=tensor([7.8249e-06, 8.4381e-06, 8.5146e-06, 8.4684e-06, 7.9419e-06, 8.5240e-06, 8.9518e-06, 9.4280e-06], device='cuda:0') 2023-03-07 10:26:22,551 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 5.753e+01 1.641e+02 2.817e+02 4.953e+02 1.195e+03, threshold=5.634e+02, percent-clipped=34.0 2023-03-07 10:26:22,596 INFO [train2.py:809] (0/4) Epoch 1, batch 500, loss[ctc_loss=1.05, att_loss=0.8287, loss=0.873, over 16167.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007788, over 41.00 utterances.], tot_loss[ctc_loss=1.173, att_loss=0.9284, loss=0.9774, over 2998645.29 frames. utt_duration=1205 frames, utt_pad_proportion=0.06586, over 9963.96 utterances.], batch size: 41, lr: 4.99e-02, grad_scale: 4.0 2023-03-07 10:27:27,826 INFO [train2.py:809] (0/4) Epoch 1, batch 550, loss[ctc_loss=1.077, att_loss=0.8718, loss=0.9128, over 17454.00 frames. utt_duration=1110 frames, utt_pad_proportion=0.02985, over 63.00 utterances.], tot_loss[ctc_loss=1.152, att_loss=0.9165, loss=0.9635, over 3051608.64 frames. utt_duration=1158 frames, utt_pad_proportion=0.08168, over 10558.30 utterances.], batch size: 63, lr: 4.98e-02, grad_scale: 4.0 2023-03-07 10:27:41,701 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=562.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:27:49,261 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=568.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:28:19,131 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=590.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:28:28,263 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=7.06 vs. limit=5.0 2023-03-07 10:28:32,089 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=600.0, num_to_drop=2, layers_to_drop={1, 2} 2023-03-07 10:28:33,064 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.073e+01 2.172e+02 3.350e+02 5.891e+02 2.324e+03, threshold=6.699e+02, percent-clipped=26.0 2023-03-07 10:28:33,107 INFO [train2.py:809] (0/4) Epoch 1, batch 600, loss[ctc_loss=1.041, att_loss=0.8909, loss=0.9209, over 16776.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005978, over 48.00 utterances.], tot_loss[ctc_loss=1.122, att_loss=0.8989, loss=0.9434, over 3093571.65 frames. utt_duration=1194 frames, utt_pad_proportion=0.0729, over 10380.64 utterances.], batch size: 48, lr: 4.98e-02, grad_scale: 4.0 2023-03-07 10:28:33,707 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.82 vs. limit=5.0 2023-03-07 10:28:45,125 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.39 vs. limit=2.0 2023-03-07 10:29:01,197 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=623.0, num_to_drop=2, layers_to_drop={2, 3} 2023-03-07 10:29:09,923 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=629.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 10:29:33,549 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=648.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:29:37,165 INFO [train2.py:809] (0/4) Epoch 1, batch 650, loss[ctc_loss=0.9721, att_loss=0.8853, loss=0.9027, over 17102.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.016, over 56.00 utterances.], tot_loss[ctc_loss=1.086, att_loss=0.8852, loss=0.9254, over 3134394.51 frames. utt_duration=1211 frames, utt_pad_proportion=0.06687, over 10363.28 utterances.], batch size: 56, lr: 4.98e-02, grad_scale: 4.0 2023-03-07 10:29:37,378 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=651.0, num_to_drop=2, layers_to_drop={2, 3} 2023-03-07 10:29:38,578 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=652.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 10:30:39,888 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1913, 5.6188, 5.7609, 5.9950, 6.0777, 5.9124, 5.9621, 5.8970], device='cuda:0'), covar=tensor([0.1305, 0.0590, 0.1530, 0.1395, 0.1043, 0.1330, 0.2042, 0.1176], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0023, 0.0025, 0.0024, 0.0023, 0.0025, 0.0025, 0.0026], device='cuda:0'), out_proj_covar=tensor([2.2219e-05, 2.2402e-05, 2.4209e-05, 2.3733e-05, 2.3110e-05, 2.4990e-05, 2.5330e-05, 2.6115e-05], device='cuda:0') 2023-03-07 10:30:42,278 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 2.720e+02 3.813e+02 5.930e+02 1.074e+03, threshold=7.626e+02, percent-clipped=15.0 2023-03-07 10:30:42,321 INFO [train2.py:809] (0/4) Epoch 1, batch 700, loss[ctc_loss=0.7937, att_loss=0.7617, loss=0.7681, over 15744.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.009568, over 38.00 utterances.], tot_loss[ctc_loss=1.04, att_loss=0.8709, loss=0.9048, over 3158982.65 frames. utt_duration=1218 frames, utt_pad_proportion=0.06576, over 10390.89 utterances.], batch size: 38, lr: 4.98e-02, grad_scale: 4.0 2023-03-07 10:30:50,658 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.20 vs. limit=5.0 2023-03-07 10:31:32,128 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=739.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 10:31:37,624 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=743.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 10:31:47,678 INFO [train2.py:809] (0/4) Epoch 1, batch 750, loss[ctc_loss=0.8823, att_loss=0.88, loss=0.8805, over 17018.00 frames. utt_duration=1286 frames, utt_pad_proportion=0.0114, over 53.00 utterances.], tot_loss[ctc_loss=0.992, att_loss=0.8602, loss=0.8865, over 3179260.76 frames. utt_duration=1239 frames, utt_pad_proportion=0.06079, over 10276.19 utterances.], batch size: 53, lr: 4.97e-02, grad_scale: 4.0 2023-03-07 10:32:12,020 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.04 vs. limit=2.0 2023-03-07 10:32:34,574 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=787.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:32:37,483 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2023-03-07 10:32:52,870 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.221e+02 2.689e+02 3.178e+02 4.558e+02 8.298e+02, threshold=6.355e+02, percent-clipped=3.0 2023-03-07 10:32:52,913 INFO [train2.py:809] (0/4) Epoch 1, batch 800, loss[ctc_loss=0.7686, att_loss=0.8112, loss=0.8027, over 16517.00 frames. utt_duration=1470 frames, utt_pad_proportion=0.007547, over 45.00 utterances.], tot_loss[ctc_loss=0.9466, att_loss=0.8519, loss=0.8709, over 3203520.82 frames. utt_duration=1224 frames, utt_pad_proportion=0.06285, over 10482.37 utterances.], batch size: 45, lr: 4.97e-02, grad_scale: 8.0 2023-03-07 10:33:06,822 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-03-07 10:33:52,567 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=847.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:33:57,208 INFO [train2.py:809] (0/4) Epoch 1, batch 850, loss[ctc_loss=0.7707, att_loss=0.7583, loss=0.7608, over 16399.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007001, over 44.00 utterances.], tot_loss[ctc_loss=0.9051, att_loss=0.8384, loss=0.8517, over 3210938.39 frames. utt_duration=1235 frames, utt_pad_proportion=0.06207, over 10410.74 utterances.], batch size: 44, lr: 4.96e-02, grad_scale: 8.0 2023-03-07 10:34:01,873 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-07 10:35:01,440 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 3.625e+02 4.631e+02 5.738e+02 1.582e+03, threshold=9.262e+02, percent-clipped=18.0 2023-03-07 10:35:01,482 INFO [train2.py:809] (0/4) Epoch 1, batch 900, loss[ctc_loss=0.7545, att_loss=0.7483, loss=0.7495, over 17269.00 frames. utt_duration=876.1 frames, utt_pad_proportion=0.0807, over 79.00 utterances.], tot_loss[ctc_loss=0.866, att_loss=0.8175, loss=0.8272, over 3224452.87 frames. utt_duration=1231 frames, utt_pad_proportion=0.06174, over 10486.14 utterances.], batch size: 79, lr: 4.96e-02, grad_scale: 8.0 2023-03-07 10:35:10,201 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=908.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 10:35:23,167 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=918.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 10:35:25,278 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-03-07 10:35:31,209 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=924.0, num_to_drop=2, layers_to_drop={0, 3} 2023-03-07 10:35:50,160 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-07 10:35:59,510 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=946.0, num_to_drop=2, layers_to_drop={0, 3} 2023-03-07 10:36:05,673 INFO [train2.py:809] (0/4) Epoch 1, batch 950, loss[ctc_loss=0.8004, att_loss=0.699, loss=0.7193, over 16479.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006594, over 46.00 utterances.], tot_loss[ctc_loss=0.8336, att_loss=0.7907, loss=0.7993, over 3236438.24 frames. utt_duration=1270 frames, utt_pad_proportion=0.05195, over 10207.93 utterances.], batch size: 46, lr: 4.96e-02, grad_scale: 8.0 2023-03-07 10:36:07,088 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=952.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 10:36:31,592 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2023-03-07 10:37:08,811 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1000.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:37:09,959 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.067e+02 5.280e+02 6.177e+02 7.418e+02 1.446e+03, threshold=1.235e+03, percent-clipped=9.0 2023-03-07 10:37:10,000 INFO [train2.py:809] (0/4) Epoch 1, batch 1000, loss[ctc_loss=0.7127, att_loss=0.6364, loss=0.6517, over 16947.00 frames. utt_duration=686 frames, utt_pad_proportion=0.1403, over 99.00 utterances.], tot_loss[ctc_loss=0.8017, att_loss=0.757, loss=0.7659, over 3240477.37 frames. utt_duration=1255 frames, utt_pad_proportion=0.05539, over 10343.53 utterances.], batch size: 99, lr: 4.95e-02, grad_scale: 8.0 2023-03-07 10:37:32,905 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6993, 4.1095, 1.4484, 4.8554, 4.5972, 3.5057, 4.7262, 4.6409], device='cuda:0'), covar=tensor([0.4574, 0.1283, 0.0934, 0.0374, 0.0616, 0.1505, 0.0508, 0.0592], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0019, 0.0015, 0.0020, 0.0021, 0.0022, 0.0020, 0.0021], device='cuda:0'), out_proj_covar=tensor([2.0031e-05, 1.5173e-05, 1.0174e-05, 1.3487e-05, 1.5037e-05, 1.6615e-05, 1.3713e-05, 1.4622e-05], device='cuda:0') 2023-03-07 10:38:04,772 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1043.0, num_to_drop=2, layers_to_drop={0, 2} 2023-03-07 10:38:15,413 INFO [train2.py:809] (0/4) Epoch 1, batch 1050, loss[ctc_loss=0.6955, att_loss=0.6427, loss=0.6532, over 16876.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007899, over 49.00 utterances.], tot_loss[ctc_loss=0.7683, att_loss=0.7206, loss=0.7301, over 3244696.45 frames. utt_duration=1293 frames, utt_pad_proportion=0.04692, over 10048.97 utterances.], batch size: 49, lr: 4.95e-02, grad_scale: 8.0 2023-03-07 10:39:07,941 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1091.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:39:21,061 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.042e+02 5.471e+02 6.537e+02 8.714e+02 1.508e+03, threshold=1.307e+03, percent-clipped=3.0 2023-03-07 10:39:21,104 INFO [train2.py:809] (0/4) Epoch 1, batch 1100, loss[ctc_loss=0.6025, att_loss=0.5333, loss=0.5472, over 16387.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.008251, over 44.00 utterances.], tot_loss[ctc_loss=0.7415, att_loss=0.6862, loss=0.6973, over 3250936.26 frames. utt_duration=1274 frames, utt_pad_proportion=0.05133, over 10222.41 utterances.], batch size: 44, lr: 4.94e-02, grad_scale: 8.0 2023-03-07 10:40:27,178 INFO [train2.py:809] (0/4) Epoch 1, batch 1150, loss[ctc_loss=0.5804, att_loss=0.5024, loss=0.518, over 15972.00 frames. utt_duration=1560 frames, utt_pad_proportion=0.005907, over 41.00 utterances.], tot_loss[ctc_loss=0.7197, att_loss=0.6577, loss=0.6701, over 3255258.59 frames. utt_duration=1224 frames, utt_pad_proportion=0.06453, over 10651.91 utterances.], batch size: 41, lr: 4.94e-02, grad_scale: 8.0 2023-03-07 10:40:30,066 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1153.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:41:32,899 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.510e+02 5.540e+02 7.016e+02 8.798e+02 2.623e+03, threshold=1.403e+03, percent-clipped=4.0 2023-03-07 10:41:32,941 INFO [train2.py:809] (0/4) Epoch 1, batch 1200, loss[ctc_loss=0.5503, att_loss=0.4855, loss=0.4984, over 16415.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006018, over 44.00 utterances.], tot_loss[ctc_loss=0.6968, att_loss=0.6295, loss=0.643, over 3263822.61 frames. utt_duration=1214 frames, utt_pad_proportion=0.06427, over 10766.63 utterances.], batch size: 44, lr: 4.93e-02, grad_scale: 8.0 2023-03-07 10:41:35,557 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1203.0, num_to_drop=2, layers_to_drop={2, 3} 2023-03-07 10:41:49,213 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1214.0, num_to_drop=2, layers_to_drop={0, 3} 2023-03-07 10:41:54,873 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1218.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 10:42:02,756 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1224.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 10:42:08,542 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9926, 4.8317, 4.0912, 5.3443, 4.6340, 4.7717, 4.3640, 5.2220], device='cuda:0'), covar=tensor([0.0222, 0.0257, 0.4688, 0.0157, 0.0285, 0.0330, 0.0788, 0.0157], device='cuda:0'), in_proj_covar=tensor([0.0024, 0.0019, 0.0029, 0.0024, 0.0017, 0.0024, 0.0022, 0.0024], device='cuda:0'), out_proj_covar=tensor([1.4394e-05, 1.2134e-05, 1.9954e-05, 1.4750e-05, 1.1112e-05, 1.5555e-05, 1.4110e-05, 1.4643e-05], device='cuda:0') 2023-03-07 10:42:31,457 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1246.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:42:37,640 INFO [train2.py:809] (0/4) Epoch 1, batch 1250, loss[ctc_loss=0.5592, att_loss=0.4785, loss=0.4946, over 15943.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007186, over 41.00 utterances.], tot_loss[ctc_loss=0.6744, att_loss=0.6026, loss=0.617, over 3262635.64 frames. utt_duration=1208 frames, utt_pad_proportion=0.06514, over 10820.75 utterances.], batch size: 41, lr: 4.92e-02, grad_scale: 8.0 2023-03-07 10:42:51,157 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9845, 4.9802, 5.2648, 5.1215, 5.2439, 5.1373, 5.2199, 5.2557], device='cuda:0'), covar=tensor([0.0437, 0.0648, 0.0447, 0.0365, 0.1021, 0.0430, 0.0390, 0.0438], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0040, 0.0040, 0.0039, 0.0037, 0.0043, 0.0030, 0.0036], device='cuda:0'), out_proj_covar=tensor([3.8202e-05, 3.6335e-05, 3.4333e-05, 3.3224e-05, 3.3424e-05, 3.6847e-05, 2.4988e-05, 3.1459e-05], device='cuda:0') 2023-03-07 10:42:56,385 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1266.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:43:05,032 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1272.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 10:43:23,040 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.18 vs. limit=2.0 2023-03-07 10:43:34,645 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1294.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 10:43:43,368 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.330e+02 5.046e+02 6.291e+02 8.119e+02 1.875e+03, threshold=1.258e+03, percent-clipped=2.0 2023-03-07 10:43:43,410 INFO [train2.py:809] (0/4) Epoch 1, batch 1300, loss[ctc_loss=0.5922, att_loss=0.5094, loss=0.526, over 16687.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005773, over 46.00 utterances.], tot_loss[ctc_loss=0.6564, att_loss=0.581, loss=0.5961, over 3263188.07 frames. utt_duration=1214 frames, utt_pad_proportion=0.06483, over 10764.69 utterances.], batch size: 46, lr: 4.92e-02, grad_scale: 8.0 2023-03-07 10:44:46,952 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3730, 1.6934, 2.6712, 3.3645, 3.2608, 3.0850, 3.3996, 3.3139], device='cuda:0'), covar=tensor([0.1738, 1.0229, 0.3243, 0.1828, 0.1651, 0.2484, 0.1924, 0.1929], device='cuda:0'), in_proj_covar=tensor([0.0060, 0.0040, 0.0040, 0.0049, 0.0036, 0.0061, 0.0068, 0.0059], device='cuda:0'), out_proj_covar=tensor([4.7440e-05, 3.9791e-05, 3.7976e-05, 4.3368e-05, 2.9589e-05, 4.8748e-05, 5.4671e-05, 4.6614e-05], device='cuda:0') 2023-03-07 10:44:50,634 INFO [train2.py:809] (0/4) Epoch 1, batch 1350, loss[ctc_loss=0.5705, att_loss=0.4972, loss=0.5119, over 17367.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03485, over 63.00 utterances.], tot_loss[ctc_loss=0.6387, att_loss=0.5614, loss=0.5769, over 3269726.00 frames. utt_duration=1217 frames, utt_pad_proportion=0.06094, over 10756.83 utterances.], batch size: 63, lr: 4.91e-02, grad_scale: 8.0 2023-03-07 10:45:59,935 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.741e+02 4.980e+02 5.949e+02 7.400e+02 1.342e+03, threshold=1.190e+03, percent-clipped=3.0 2023-03-07 10:45:59,977 INFO [train2.py:809] (0/4) Epoch 1, batch 1400, loss[ctc_loss=0.5939, att_loss=0.4927, loss=0.513, over 17003.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.009469, over 51.00 utterances.], tot_loss[ctc_loss=0.6213, att_loss=0.5423, loss=0.5581, over 3271459.12 frames. utt_duration=1212 frames, utt_pad_proportion=0.06322, over 10811.97 utterances.], batch size: 51, lr: 4.91e-02, grad_scale: 8.0 2023-03-07 10:46:45,447 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1434.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:47:08,136 INFO [train2.py:809] (0/4) Epoch 1, batch 1450, loss[ctc_loss=0.6188, att_loss=0.4924, loss=0.5177, over 16981.00 frames. utt_duration=687.5 frames, utt_pad_proportion=0.1374, over 99.00 utterances.], tot_loss[ctc_loss=0.6054, att_loss=0.5259, loss=0.5418, over 3270788.73 frames. utt_duration=1227 frames, utt_pad_proportion=0.05957, over 10678.71 utterances.], batch size: 99, lr: 4.90e-02, grad_scale: 8.0 2023-03-07 10:47:19,486 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-03-07 10:48:08,678 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1495.0, num_to_drop=2, layers_to_drop={2, 3} 2023-03-07 10:48:16,251 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.693e+02 5.276e+02 6.895e+02 9.021e+02 2.982e+03, threshold=1.379e+03, percent-clipped=6.0 2023-03-07 10:48:16,297 INFO [train2.py:809] (0/4) Epoch 1, batch 1500, loss[ctc_loss=0.5726, att_loss=0.4938, loss=0.5095, over 17335.00 frames. utt_duration=1102 frames, utt_pad_proportion=0.03667, over 63.00 utterances.], tot_loss[ctc_loss=0.5905, att_loss=0.5121, loss=0.5278, over 3260930.71 frames. utt_duration=1201 frames, utt_pad_proportion=0.07046, over 10877.02 utterances.], batch size: 63, lr: 4.89e-02, grad_scale: 8.0 2023-03-07 10:48:19,320 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1503.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:48:27,268 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1509.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 10:48:34,635 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-03-07 10:49:26,092 INFO [train2.py:809] (0/4) Epoch 1, batch 1550, loss[ctc_loss=0.5187, att_loss=0.4468, loss=0.4612, over 17068.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.008171, over 53.00 utterances.], tot_loss[ctc_loss=0.5739, att_loss=0.4977, loss=0.5129, over 3266520.61 frames. utt_duration=1214 frames, utt_pad_proportion=0.06532, over 10779.25 utterances.], batch size: 53, lr: 4.89e-02, grad_scale: 8.0 2023-03-07 10:49:26,174 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1551.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 10:49:41,347 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-03-07 10:49:46,959 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1566.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:50:21,719 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3321, 4.6159, 4.8216, 4.5086, 4.6505, 4.5021, 4.4276, 4.4590], device='cuda:0'), covar=tensor([0.0417, 0.0427, 0.0344, 0.0441, 0.0492, 0.0432, 0.0590, 0.0432], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0045, 0.0052, 0.0039, 0.0052, 0.0059, 0.0047, 0.0053], device='cuda:0'), out_proj_covar=tensor([4.7916e-05, 4.0184e-05, 4.4191e-05, 3.2225e-05, 4.6377e-05, 5.2658e-05, 4.3243e-05, 4.9053e-05], device='cuda:0') 2023-03-07 10:50:21,787 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5026, 2.0765, 3.9722, 3.7731, 3.6217, 3.2834, 3.1011, 3.5108], device='cuda:0'), covar=tensor([0.0444, 0.3474, 0.0330, 0.0573, 0.0626, 0.1164, 0.1233, 0.0961], device='cuda:0'), in_proj_covar=tensor([0.0045, 0.0047, 0.0060, 0.0081, 0.0075, 0.0089, 0.0048, 0.0079], device='cuda:0'), out_proj_covar=tensor([3.6737e-05, 3.9280e-05, 4.4970e-05, 5.9087e-05, 5.2861e-05, 6.7751e-05, 4.0617e-05, 5.7787e-05], device='cuda:0') 2023-03-07 10:50:36,311 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.585e+02 5.591e+02 7.426e+02 9.323e+02 2.170e+03, threshold=1.485e+03, percent-clipped=5.0 2023-03-07 10:50:36,354 INFO [train2.py:809] (0/4) Epoch 1, batch 1600, loss[ctc_loss=0.4693, att_loss=0.4162, loss=0.4269, over 16116.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006834, over 42.00 utterances.], tot_loss[ctc_loss=0.5596, att_loss=0.4857, loss=0.5004, over 3258659.54 frames. utt_duration=1198 frames, utt_pad_proportion=0.07118, over 10894.38 utterances.], batch size: 42, lr: 4.88e-02, grad_scale: 8.0 2023-03-07 10:50:36,483 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.3372, 6.1501, 5.7877, 6.1041, 6.1282, 5.7827, 6.0745, 5.9521], device='cuda:0'), covar=tensor([0.0321, 0.0360, 0.0377, 0.0327, 0.0322, 0.0570, 0.0368, 0.0490], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0091, 0.0101, 0.0084, 0.0088, 0.0109, 0.0086, 0.0106], device='cuda:0'), out_proj_covar=tensor([8.1256e-05, 9.4088e-05, 1.0475e-04, 8.6808e-05, 8.5110e-05, 1.2683e-04, 8.7462e-05, 1.1662e-04], device='cuda:0') 2023-03-07 10:50:50,094 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1611.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:51:13,075 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1627.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 10:51:15,694 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1875, 5.6766, 4.4585, 5.6864, 4.6848, 4.9225, 4.4882, 5.3768], device='cuda:0'), covar=tensor([0.0221, 0.0069, 0.3251, 0.0072, 0.0583, 0.0400, 0.1923, 0.0115], device='cuda:0'), in_proj_covar=tensor([0.0041, 0.0030, 0.0034, 0.0038, 0.0029, 0.0039, 0.0034, 0.0037], device='cuda:0'), out_proj_covar=tensor([2.2672e-05, 1.5286e-05, 2.5321e-05, 2.0330e-05, 1.5663e-05, 2.2086e-05, 2.0572e-05, 1.9864e-05], device='cuda:0') 2023-03-07 10:51:16,852 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9036, 3.8069, 3.8861, 3.9063, 3.9539, 3.5096, 3.6529, 3.6866], device='cuda:0'), covar=tensor([0.0226, 0.0255, 0.0238, 0.0208, 0.0214, 0.0566, 0.0364, 0.0348], device='cuda:0'), in_proj_covar=tensor([0.0034, 0.0036, 0.0038, 0.0037, 0.0033, 0.0036, 0.0039, 0.0038], device='cuda:0'), out_proj_covar=tensor([2.6364e-05, 2.8386e-05, 3.0887e-05, 2.9088e-05, 2.6385e-05, 2.9517e-05, 3.1273e-05, 2.9619e-05], device='cuda:0') 2023-03-07 10:51:44,866 INFO [train2.py:809] (0/4) Epoch 1, batch 1650, loss[ctc_loss=0.5408, att_loss=0.4732, loss=0.4867, over 16864.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.008545, over 49.00 utterances.], tot_loss[ctc_loss=0.5477, att_loss=0.4757, loss=0.4901, over 3261258.60 frames. utt_duration=1198 frames, utt_pad_proportion=0.07, over 10899.53 utterances.], batch size: 49, lr: 4.87e-02, grad_scale: 8.0 2023-03-07 10:52:15,862 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1672.0, num_to_drop=2, layers_to_drop={0, 3} 2023-03-07 10:52:55,824 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.825e+02 4.712e+02 6.880e+02 8.240e+02 2.356e+03, threshold=1.376e+03, percent-clipped=2.0 2023-03-07 10:52:55,867 INFO [train2.py:809] (0/4) Epoch 1, batch 1700, loss[ctc_loss=0.4929, att_loss=0.435, loss=0.4466, over 17292.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01231, over 55.00 utterances.], tot_loss[ctc_loss=0.5361, att_loss=0.4666, loss=0.4805, over 3267235.03 frames. utt_duration=1209 frames, utt_pad_proportion=0.06547, over 10824.53 utterances.], batch size: 55, lr: 4.86e-02, grad_scale: 8.0 2023-03-07 10:54:07,062 INFO [train2.py:809] (0/4) Epoch 1, batch 1750, loss[ctc_loss=0.4823, att_loss=0.4466, loss=0.4538, over 16866.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.007567, over 49.00 utterances.], tot_loss[ctc_loss=0.5216, att_loss=0.4572, loss=0.4701, over 3263214.94 frames. utt_duration=1213 frames, utt_pad_proportion=0.06308, over 10777.03 utterances.], batch size: 49, lr: 4.86e-02, grad_scale: 8.0 2023-03-07 10:55:02,387 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1790.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:55:05,898 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.34 vs. limit=5.0 2023-03-07 10:55:11,868 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9788, 5.2753, 4.2513, 5.3036, 4.6575, 4.5872, 4.3323, 4.6924], device='cuda:0'), covar=tensor([0.0220, 0.0063, 0.1614, 0.0068, 0.0370, 0.0538, 0.1270, 0.0193], device='cuda:0'), in_proj_covar=tensor([0.0045, 0.0033, 0.0033, 0.0039, 0.0032, 0.0040, 0.0038, 0.0040], device='cuda:0'), out_proj_covar=tensor([2.5740e-05, 1.6498e-05, 2.4948e-05, 2.0231e-05, 1.6661e-05, 2.3581e-05, 2.2370e-05, 2.0597e-05], device='cuda:0') 2023-03-07 10:55:17,469 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.104e+02 4.727e+02 5.545e+02 7.358e+02 1.291e+03, threshold=1.109e+03, percent-clipped=0.0 2023-03-07 10:55:17,511 INFO [train2.py:809] (0/4) Epoch 1, batch 1800, loss[ctc_loss=0.4086, att_loss=0.3728, loss=0.3799, over 15371.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01053, over 35.00 utterances.], tot_loss[ctc_loss=0.5097, att_loss=0.4489, loss=0.461, over 3265816.69 frames. utt_duration=1220 frames, utt_pad_proportion=0.06154, over 10721.49 utterances.], batch size: 35, lr: 4.85e-02, grad_scale: 8.0 2023-03-07 10:55:28,533 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=1809.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 10:56:09,980 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1838.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:56:27,794 INFO [train2.py:809] (0/4) Epoch 1, batch 1850, loss[ctc_loss=0.5914, att_loss=0.5016, loss=0.5196, over 14017.00 frames. utt_duration=382.9 frames, utt_pad_proportion=0.3282, over 147.00 utterances.], tot_loss[ctc_loss=0.5002, att_loss=0.4437, loss=0.455, over 3264394.59 frames. utt_duration=1233 frames, utt_pad_proportion=0.05995, over 10601.28 utterances.], batch size: 147, lr: 4.84e-02, grad_scale: 8.0 2023-03-07 10:56:36,554 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=1857.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 10:56:36,786 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1857.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 10:56:56,752 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=1870.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 10:57:37,072 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1899.0, num_to_drop=2, layers_to_drop={1, 2} 2023-03-07 10:57:39,636 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.994e+02 5.025e+02 6.383e+02 8.643e+02 1.619e+03, threshold=1.277e+03, percent-clipped=10.0 2023-03-07 10:57:39,680 INFO [train2.py:809] (0/4) Epoch 1, batch 1900, loss[ctc_loss=0.4724, att_loss=0.4381, loss=0.445, over 17025.00 frames. utt_duration=1286 frames, utt_pad_proportion=0.01118, over 53.00 utterances.], tot_loss[ctc_loss=0.4902, att_loss=0.4381, loss=0.4485, over 3267890.89 frames. utt_duration=1244 frames, utt_pad_proportion=0.0576, over 10521.83 utterances.], batch size: 53, lr: 4.83e-02, grad_scale: 8.0 2023-03-07 10:57:55,880 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8975, 1.9275, 2.9476, 4.0442, 3.6030, 2.9076, 3.5646, 3.4738], device='cuda:0'), covar=tensor([0.0463, 0.2638, 0.1359, 0.0403, 0.0751, 0.2422, 0.0831, 0.1416], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0048, 0.0048, 0.0065, 0.0054, 0.0089, 0.0084, 0.0077], device='cuda:0'), out_proj_covar=tensor([5.9068e-05, 4.6381e-05, 4.5498e-05, 5.1616e-05, 4.2671e-05, 8.6679e-05, 6.6779e-05, 5.8304e-05], device='cuda:0') 2023-03-07 10:58:05,227 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1918.0, num_to_drop=2, layers_to_drop={0, 2} 2023-03-07 10:58:10,232 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1922.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 10:58:22,720 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=1931.0, num_to_drop=2, layers_to_drop={0, 2} 2023-03-07 10:58:50,732 INFO [train2.py:809] (0/4) Epoch 1, batch 1950, loss[ctc_loss=0.4414, att_loss=0.4226, loss=0.4263, over 16124.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006264, over 42.00 utterances.], tot_loss[ctc_loss=0.4792, att_loss=0.4319, loss=0.4414, over 3272099.09 frames. utt_duration=1261 frames, utt_pad_proportion=0.05336, over 10392.76 utterances.], batch size: 42, lr: 4.83e-02, grad_scale: 8.0 2023-03-07 10:58:52,364 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0549, 4.0690, 3.1192, 4.4910, 3.5390, 4.4639, 4.9224, 4.7653], device='cuda:0'), covar=tensor([0.1714, 0.0963, 0.0522, 0.0278, 0.1296, 0.0255, 0.0170, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0037, 0.0033, 0.0047, 0.0053, 0.0030, 0.0043, 0.0041], device='cuda:0'), out_proj_covar=tensor([2.8255e-05, 2.5336e-05, 1.8433e-05, 2.7474e-05, 3.7170e-05, 1.9624e-05, 2.4937e-05, 2.3199e-05], device='cuda:0') 2023-03-07 10:59:14,909 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=1967.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:00:01,373 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-2000.pt 2023-03-07 11:00:06,965 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.041e+02 5.008e+02 6.125e+02 7.778e+02 1.503e+03, threshold=1.225e+03, percent-clipped=3.0 2023-03-07 11:00:07,008 INFO [train2.py:809] (0/4) Epoch 1, batch 2000, loss[ctc_loss=0.4062, att_loss=0.39, loss=0.3933, over 15997.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007822, over 40.00 utterances.], tot_loss[ctc_loss=0.4678, att_loss=0.4254, loss=0.4339, over 3266824.37 frames. utt_duration=1280 frames, utt_pad_proportion=0.05019, over 10222.35 utterances.], batch size: 40, lr: 4.82e-02, grad_scale: 16.0 2023-03-07 11:00:15,623 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-03-07 11:00:57,029 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1820, 5.9328, 5.5672, 6.1219, 5.9314, 5.6021, 5.9464, 5.5769], device='cuda:0'), covar=tensor([0.0429, 0.0463, 0.0514, 0.0298, 0.0457, 0.0723, 0.0482, 0.0892], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0123, 0.0124, 0.0102, 0.0102, 0.0136, 0.0111, 0.0134], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-07 11:01:23,773 INFO [train2.py:809] (0/4) Epoch 1, batch 2050, loss[ctc_loss=0.4856, att_loss=0.444, loss=0.4523, over 14473.00 frames. utt_duration=398.1 frames, utt_pad_proportion=0.3053, over 146.00 utterances.], tot_loss[ctc_loss=0.4521, att_loss=0.4172, loss=0.4242, over 3266933.84 frames. utt_duration=1275 frames, utt_pad_proportion=0.04991, over 10265.28 utterances.], batch size: 146, lr: 4.81e-02, grad_scale: 16.0 2023-03-07 11:02:23,645 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2090.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 11:02:39,886 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.629e+02 3.921e+02 4.697e+02 6.442e+02 1.106e+03, threshold=9.394e+02, percent-clipped=0.0 2023-03-07 11:02:39,928 INFO [train2.py:809] (0/4) Epoch 1, batch 2100, loss[ctc_loss=0.4447, att_loss=0.4148, loss=0.4208, over 17171.00 frames. utt_duration=695.2 frames, utt_pad_proportion=0.1255, over 99.00 utterances.], tot_loss[ctc_loss=0.4378, att_loss=0.4105, loss=0.416, over 3264155.70 frames. utt_duration=1250 frames, utt_pad_proportion=0.05626, over 10458.16 utterances.], batch size: 99, lr: 4.80e-02, grad_scale: 16.0 2023-03-07 11:03:35,192 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2138.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:03:55,161 INFO [train2.py:809] (0/4) Epoch 1, batch 2150, loss[ctc_loss=0.3867, att_loss=0.3906, loss=0.3898, over 17351.00 frames. utt_duration=1178 frames, utt_pad_proportion=0.02169, over 59.00 utterances.], tot_loss[ctc_loss=0.4269, att_loss=0.4056, loss=0.4099, over 3263284.75 frames. utt_duration=1265 frames, utt_pad_proportion=0.05296, over 10330.97 utterances.], batch size: 59, lr: 4.79e-02, grad_scale: 16.0 2023-03-07 11:05:00,212 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2194.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:05:10,566 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.636e+02 4.045e+02 5.653e+02 7.128e+02 1.127e+03, threshold=1.131e+03, percent-clipped=5.0 2023-03-07 11:05:10,609 INFO [train2.py:809] (0/4) Epoch 1, batch 2200, loss[ctc_loss=0.3382, att_loss=0.3566, loss=0.3529, over 15896.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008599, over 39.00 utterances.], tot_loss[ctc_loss=0.414, att_loss=0.3998, loss=0.4026, over 3272544.03 frames. utt_duration=1270 frames, utt_pad_proportion=0.04984, over 10319.52 utterances.], batch size: 39, lr: 4.78e-02, grad_scale: 16.0 2023-03-07 11:05:30,174 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2213.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:05:43,398 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2222.0, num_to_drop=2, layers_to_drop={0, 2} 2023-03-07 11:05:49,117 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2226.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:06:17,857 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8066, 4.9484, 4.4318, 5.1036, 5.1179, 4.8183, 4.5029, 3.7070], device='cuda:0'), covar=tensor([0.0221, 0.0107, 0.0703, 0.0055, 0.0104, 0.0174, 0.0656, 0.0710], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0039, 0.0034, 0.0045, 0.0038, 0.0048, 0.0043, 0.0053], device='cuda:0'), out_proj_covar=tensor([3.3436e-05, 1.9926e-05, 2.5668e-05, 2.0964e-05, 1.9350e-05, 2.7717e-05, 2.6327e-05, 2.9669e-05], device='cuda:0') 2023-03-07 11:06:26,395 INFO [train2.py:809] (0/4) Epoch 1, batch 2250, loss[ctc_loss=0.3905, att_loss=0.3933, loss=0.3927, over 16621.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005484, over 47.00 utterances.], tot_loss[ctc_loss=0.4056, att_loss=0.3966, loss=0.3984, over 3278618.23 frames. utt_duration=1263 frames, utt_pad_proportion=0.05115, over 10399.81 utterances.], batch size: 47, lr: 4.77e-02, grad_scale: 16.0 2023-03-07 11:06:52,780 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2267.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 11:06:55,779 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3898, 4.1332, 4.6270, 4.5862, 4.7849, 4.0272, 4.5291, 4.8908], device='cuda:0'), covar=tensor([0.0145, 0.0156, 0.0128, 0.0122, 0.0060, 0.0197, 0.0124, 0.0078], device='cuda:0'), in_proj_covar=tensor([0.0015, 0.0014, 0.0013, 0.0013, 0.0014, 0.0017, 0.0014, 0.0014], device='cuda:0'), out_proj_covar=tensor([1.1568e-05, 1.0889e-05, 1.0491e-05, 9.7861e-06, 1.0266e-05, 1.3851e-05, 1.0331e-05, 1.1415e-05], device='cuda:0') 2023-03-07 11:06:57,086 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2270.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:07:43,372 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.224e+02 3.835e+02 4.889e+02 6.269e+02 9.585e+02, threshold=9.779e+02, percent-clipped=0.0 2023-03-07 11:07:43,417 INFO [train2.py:809] (0/4) Epoch 1, batch 2300, loss[ctc_loss=0.3652, att_loss=0.3876, loss=0.3831, over 16896.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.006798, over 49.00 utterances.], tot_loss[ctc_loss=0.3955, att_loss=0.3924, loss=0.393, over 3268196.61 frames. utt_duration=1264 frames, utt_pad_proportion=0.05338, over 10355.69 utterances.], batch size: 49, lr: 4.77e-02, grad_scale: 16.0 2023-03-07 11:08:06,459 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2315.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:08:43,539 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2340.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 11:08:59,800 INFO [train2.py:809] (0/4) Epoch 1, batch 2350, loss[ctc_loss=0.4631, att_loss=0.4392, loss=0.444, over 14611.00 frames. utt_duration=404.6 frames, utt_pad_proportion=0.2964, over 145.00 utterances.], tot_loss[ctc_loss=0.3883, att_loss=0.39, loss=0.3897, over 3276292.23 frames. utt_duration=1259 frames, utt_pad_proportion=0.05298, over 10424.85 utterances.], batch size: 145, lr: 4.76e-02, grad_scale: 16.0 2023-03-07 11:10:15,553 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.741e+02 4.246e+02 5.457e+02 7.092e+02 1.687e+03, threshold=1.091e+03, percent-clipped=6.0 2023-03-07 11:10:15,596 INFO [train2.py:809] (0/4) Epoch 1, batch 2400, loss[ctc_loss=0.3248, att_loss=0.3671, loss=0.3586, over 15955.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.007026, over 41.00 utterances.], tot_loss[ctc_loss=0.3802, att_loss=0.3863, loss=0.3851, over 3275713.95 frames. utt_duration=1270 frames, utt_pad_proportion=0.04983, over 10327.67 utterances.], batch size: 41, lr: 4.75e-02, grad_scale: 16.0 2023-03-07 11:10:15,984 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2401.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 11:11:19,812 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6406, 3.3195, 3.4732, 3.8344, 3.6993, 3.5698, 2.9089, 3.3372], device='cuda:0'), covar=tensor([0.0263, 0.0425, 0.0212, 0.0132, 0.0434, 0.0384, 0.0745, 0.0378], device='cuda:0'), in_proj_covar=tensor([0.0028, 0.0028, 0.0027, 0.0027, 0.0028, 0.0027, 0.0037, 0.0030], device='cuda:0'), out_proj_covar=tensor([2.1082e-05, 2.0093e-05, 1.7984e-05, 1.8398e-05, 2.0459e-05, 1.8355e-05, 2.8535e-05, 2.1146e-05], device='cuda:0') 2023-03-07 11:11:30,933 INFO [train2.py:809] (0/4) Epoch 1, batch 2450, loss[ctc_loss=0.3326, att_loss=0.3454, loss=0.3428, over 16001.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007714, over 40.00 utterances.], tot_loss[ctc_loss=0.3743, att_loss=0.3836, loss=0.3817, over 3266892.11 frames. utt_duration=1231 frames, utt_pad_proportion=0.06186, over 10625.37 utterances.], batch size: 40, lr: 4.74e-02, grad_scale: 16.0 2023-03-07 11:12:36,896 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2494.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 11:12:47,384 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.845e+02 4.360e+02 5.388e+02 6.966e+02 1.787e+03, threshold=1.078e+03, percent-clipped=1.0 2023-03-07 11:12:47,427 INFO [train2.py:809] (0/4) Epoch 1, batch 2500, loss[ctc_loss=0.3907, att_loss=0.4027, loss=0.4003, over 17055.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008926, over 52.00 utterances.], tot_loss[ctc_loss=0.368, att_loss=0.3816, loss=0.3789, over 3268074.96 frames. utt_duration=1242 frames, utt_pad_proportion=0.05823, over 10534.17 utterances.], batch size: 52, lr: 4.73e-02, grad_scale: 16.0 2023-03-07 11:13:06,090 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2513.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:13:25,311 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2526.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:13:45,428 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7783, 5.1089, 4.9710, 4.1743, 4.5121, 4.2616, 4.8444, 4.4994], device='cuda:0'), covar=tensor([0.0726, 0.0255, 0.0563, 0.2319, 0.1051, 0.3529, 0.1720, 0.3355], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0049, 0.0046, 0.0070, 0.0060, 0.0058, 0.0043, 0.0037], device='cuda:0'), out_proj_covar=tensor([1.8117e-05, 2.0575e-05, 2.0847e-05, 4.3480e-05, 3.0017e-05, 3.3510e-05, 2.7077e-05, 2.3890e-05], device='cuda:0') 2023-03-07 11:13:49,469 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2542.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:14:02,690 INFO [train2.py:809] (0/4) Epoch 1, batch 2550, loss[ctc_loss=0.3258, att_loss=0.382, loss=0.3707, over 16484.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005604, over 46.00 utterances.], tot_loss[ctc_loss=0.362, att_loss=0.3791, loss=0.3757, over 3260909.30 frames. utt_duration=1226 frames, utt_pad_proportion=0.06374, over 10652.47 utterances.], batch size: 46, lr: 4.72e-02, grad_scale: 16.0 2023-03-07 11:14:19,961 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2561.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:14:30,574 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5166, 3.1947, 2.5027, 3.8122, 2.6656, 3.8779, 3.2259, 3.0568], device='cuda:0'), covar=tensor([0.0306, 0.0284, 0.0843, 0.0187, 0.0556, 0.0169, 0.0354, 0.0353], device='cuda:0'), in_proj_covar=tensor([0.0025, 0.0020, 0.0030, 0.0024, 0.0021, 0.0021, 0.0025, 0.0022], device='cuda:0'), out_proj_covar=tensor([1.9869e-05, 2.0585e-05, 2.6819e-05, 1.8751e-05, 1.8433e-05, 1.7202e-05, 1.9914e-05, 1.8170e-05], device='cuda:0') 2023-03-07 11:14:36,621 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7171, 5.3194, 4.2876, 5.4422, 5.4175, 4.7847, 4.6927, 3.5654], device='cuda:0'), covar=tensor([0.0437, 0.0091, 0.1003, 0.0052, 0.0168, 0.0312, 0.0610, 0.1644], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0049, 0.0041, 0.0058, 0.0050, 0.0063, 0.0050, 0.0087], device='cuda:0'), out_proj_covar=tensor([4.7741e-05, 2.2399e-05, 2.9957e-05, 2.4081e-05, 2.6951e-05, 3.4022e-05, 3.2164e-05, 5.1816e-05], device='cuda:0') 2023-03-07 11:14:39,251 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=2574.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:15:19,989 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.773e+02 4.230e+02 5.272e+02 6.534e+02 1.107e+03, threshold=1.054e+03, percent-clipped=1.0 2023-03-07 11:15:20,033 INFO [train2.py:809] (0/4) Epoch 1, batch 2600, loss[ctc_loss=0.3074, att_loss=0.3529, loss=0.3438, over 16973.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007012, over 50.00 utterances.], tot_loss[ctc_loss=0.357, att_loss=0.3767, loss=0.3728, over 3261276.69 frames. utt_duration=1209 frames, utt_pad_proportion=0.06689, over 10803.40 utterances.], batch size: 50, lr: 4.71e-02, grad_scale: 16.0 2023-03-07 11:16:37,076 INFO [train2.py:809] (0/4) Epoch 1, batch 2650, loss[ctc_loss=0.3248, att_loss=0.3413, loss=0.338, over 10710.00 frames. utt_duration=1864 frames, utt_pad_proportion=0.222, over 23.00 utterances.], tot_loss[ctc_loss=0.352, att_loss=0.3748, loss=0.3702, over 3264778.11 frames. utt_duration=1212 frames, utt_pad_proportion=0.06503, over 10787.64 utterances.], batch size: 23, lr: 4.70e-02, grad_scale: 16.0 2023-03-07 11:17:15,820 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9662, 4.6683, 4.9157, 4.6565, 4.3246, 4.2114, 4.9249, 4.9044], device='cuda:0'), covar=tensor([0.0360, 0.0191, 0.0139, 0.0207, 0.0287, 0.0156, 0.0304, 0.0135], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0050, 0.0055, 0.0046, 0.0062, 0.0051, 0.0059, 0.0046], device='cuda:0'), out_proj_covar=tensor([7.1478e-05, 4.8611e-05, 5.0506e-05, 4.2810e-05, 6.1530e-05, 5.1595e-05, 5.8746e-05, 3.8568e-05], device='cuda:0') 2023-03-07 11:17:29,881 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2685.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 11:17:45,992 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=2696.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 11:17:53,968 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.491e+02 4.149e+02 5.114e+02 6.615e+02 1.308e+03, threshold=1.023e+03, percent-clipped=1.0 2023-03-07 11:17:54,010 INFO [train2.py:809] (0/4) Epoch 1, batch 2700, loss[ctc_loss=0.4156, att_loss=0.4142, loss=0.4145, over 16627.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005229, over 47.00 utterances.], tot_loss[ctc_loss=0.3465, att_loss=0.3725, loss=0.3673, over 3270585.95 frames. utt_duration=1228 frames, utt_pad_proportion=0.05964, over 10670.57 utterances.], batch size: 47, lr: 4.69e-02, grad_scale: 16.0 2023-03-07 11:18:21,645 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-03-07 11:19:03,801 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2746.0, num_to_drop=2, layers_to_drop={2, 3} 2023-03-07 11:19:10,751 INFO [train2.py:809] (0/4) Epoch 1, batch 2750, loss[ctc_loss=0.3483, att_loss=0.3772, loss=0.3714, over 17015.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008079, over 51.00 utterances.], tot_loss[ctc_loss=0.3425, att_loss=0.3709, loss=0.3652, over 3276082.07 frames. utt_duration=1243 frames, utt_pad_proportion=0.0547, over 10553.46 utterances.], batch size: 51, lr: 4.68e-02, grad_scale: 16.0 2023-03-07 11:19:30,587 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.05 vs. limit=2.0 2023-03-07 11:20:27,835 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.038e+02 4.978e+02 5.756e+02 7.064e+02 1.353e+03, threshold=1.151e+03, percent-clipped=3.0 2023-03-07 11:20:27,878 INFO [train2.py:809] (0/4) Epoch 1, batch 2800, loss[ctc_loss=0.3214, att_loss=0.359, loss=0.3514, over 15959.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006762, over 41.00 utterances.], tot_loss[ctc_loss=0.3391, att_loss=0.3693, loss=0.3633, over 3277059.68 frames. utt_duration=1248 frames, utt_pad_proportion=0.05301, over 10517.05 utterances.], batch size: 41, lr: 4.67e-02, grad_scale: 16.0 2023-03-07 11:21:04,316 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2824.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 11:21:44,972 INFO [train2.py:809] (0/4) Epoch 1, batch 2850, loss[ctc_loss=0.2976, att_loss=0.3399, loss=0.3314, over 15367.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01151, over 35.00 utterances.], tot_loss[ctc_loss=0.3361, att_loss=0.3681, loss=0.3617, over 3265704.78 frames. utt_duration=1235 frames, utt_pad_proportion=0.06053, over 10594.21 utterances.], batch size: 35, lr: 4.66e-02, grad_scale: 16.0 2023-03-07 11:22:36,857 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2885.0, num_to_drop=2, layers_to_drop={0, 1} 2023-03-07 11:23:01,455 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.575e+02 4.414e+02 5.747e+02 7.042e+02 1.597e+03, threshold=1.149e+03, percent-clipped=5.0 2023-03-07 11:23:01,502 INFO [train2.py:809] (0/4) Epoch 1, batch 2900, loss[ctc_loss=0.3137, att_loss=0.3735, loss=0.3615, over 16883.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006744, over 49.00 utterances.], tot_loss[ctc_loss=0.3314, att_loss=0.3655, loss=0.3587, over 3266461.37 frames. utt_duration=1262 frames, utt_pad_proportion=0.05336, over 10367.32 utterances.], batch size: 49, lr: 4.65e-02, grad_scale: 16.0 2023-03-07 11:23:57,989 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=2937.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 11:24:18,944 INFO [train2.py:809] (0/4) Epoch 1, batch 2950, loss[ctc_loss=0.3173, att_loss=0.3679, loss=0.3578, over 17054.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.00897, over 52.00 utterances.], tot_loss[ctc_loss=0.3263, att_loss=0.363, loss=0.3556, over 3266295.43 frames. utt_duration=1255 frames, utt_pad_proportion=0.05651, over 10420.26 utterances.], batch size: 52, lr: 4.64e-02, grad_scale: 16.0 2023-03-07 11:24:40,038 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-03-07 11:25:29,020 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=2996.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 11:25:32,112 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=2998.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 11:25:36,469 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.432e+02 4.332e+02 5.294e+02 6.689e+02 1.698e+03, threshold=1.059e+03, percent-clipped=3.0 2023-03-07 11:25:36,515 INFO [train2.py:809] (0/4) Epoch 1, batch 3000, loss[ctc_loss=0.2839, att_loss=0.3397, loss=0.3285, over 16403.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006759, over 44.00 utterances.], tot_loss[ctc_loss=0.3237, att_loss=0.362, loss=0.3544, over 3269399.98 frames. utt_duration=1250 frames, utt_pad_proportion=0.05627, over 10477.27 utterances.], batch size: 44, lr: 4.63e-02, grad_scale: 16.0 2023-03-07 11:25:36,518 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 11:25:51,183 INFO [train2.py:843] (0/4) Epoch 1, validation: ctc_loss=0.2388, att_loss=0.3154, loss=0.3001, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 11:25:51,183 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 14491MB 2023-03-07 11:26:18,663 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3019.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 11:26:51,036 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3041.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:26:55,333 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3044.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 11:27:07,499 INFO [train2.py:809] (0/4) Epoch 1, batch 3050, loss[ctc_loss=0.325, att_loss=0.3622, loss=0.3547, over 16408.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006502, over 44.00 utterances.], tot_loss[ctc_loss=0.3203, att_loss=0.3605, loss=0.3525, over 3266429.27 frames. utt_duration=1248 frames, utt_pad_proportion=0.05876, over 10480.98 utterances.], batch size: 44, lr: 4.62e-02, grad_scale: 16.0 2023-03-07 11:27:38,520 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.68 vs. limit=5.0 2023-03-07 11:27:52,673 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3080.0, num_to_drop=2, layers_to_drop={1, 2} 2023-03-07 11:28:24,852 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.428e+02 4.335e+02 5.357e+02 6.400e+02 1.084e+03, threshold=1.071e+03, percent-clipped=0.0 2023-03-07 11:28:24,897 INFO [train2.py:809] (0/4) Epoch 1, batch 3100, loss[ctc_loss=0.3407, att_loss=0.3451, loss=0.3442, over 15648.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008766, over 37.00 utterances.], tot_loss[ctc_loss=0.3171, att_loss=0.3588, loss=0.3505, over 3272656.74 frames. utt_duration=1254 frames, utt_pad_proportion=0.05518, over 10454.52 utterances.], batch size: 37, lr: 4.61e-02, grad_scale: 16.0 2023-03-07 11:29:25,627 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5819, 5.1549, 4.9025, 4.8785, 5.2394, 5.1087, 5.0472, 4.8597], device='cuda:0'), covar=tensor([0.0355, 0.0139, 0.0198, 0.0232, 0.0193, 0.0156, 0.0144, 0.0192], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0088, 0.0076, 0.0070, 0.0097, 0.0113, 0.0081, 0.0096], device='cuda:0'), out_proj_covar=tensor([1.2674e-04, 9.5662e-05, 8.4908e-05, 8.4500e-05, 1.1425e-04, 1.4209e-04, 8.6582e-05, 1.1783e-04], device='cuda:0') 2023-03-07 11:29:41,931 INFO [train2.py:809] (0/4) Epoch 1, batch 3150, loss[ctc_loss=0.2902, att_loss=0.357, loss=0.3436, over 17019.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007773, over 51.00 utterances.], tot_loss[ctc_loss=0.3155, att_loss=0.3588, loss=0.3501, over 3268029.19 frames. utt_duration=1220 frames, utt_pad_proportion=0.06387, over 10727.83 utterances.], batch size: 51, lr: 4.60e-02, grad_scale: 16.0 2023-03-07 11:30:09,754 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1035, 3.2580, 3.3146, 2.0315, 3.2961, 3.2381, 2.9013, 2.5080], device='cuda:0'), covar=tensor([0.0331, 0.0235, 0.0238, 0.1178, 0.0213, 0.0204, 0.0520, 0.0797], device='cuda:0'), in_proj_covar=tensor([0.0033, 0.0031, 0.0030, 0.0042, 0.0029, 0.0024, 0.0029, 0.0041], device='cuda:0'), out_proj_covar=tensor([2.4461e-05, 2.2138e-05, 2.3259e-05, 3.5914e-05, 2.1135e-05, 1.8832e-05, 2.2446e-05, 3.4831e-05], device='cuda:0') 2023-03-07 11:30:25,776 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3180.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:30:58,387 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.970e+02 4.582e+02 5.665e+02 7.020e+02 1.937e+03, threshold=1.133e+03, percent-clipped=6.0 2023-03-07 11:30:58,429 INFO [train2.py:809] (0/4) Epoch 1, batch 3200, loss[ctc_loss=0.4166, att_loss=0.4109, loss=0.412, over 14269.00 frames. utt_duration=392.4 frames, utt_pad_proportion=0.3152, over 146.00 utterances.], tot_loss[ctc_loss=0.315, att_loss=0.3588, loss=0.35, over 3267919.04 frames. utt_duration=1221 frames, utt_pad_proportion=0.06339, over 10721.98 utterances.], batch size: 146, lr: 4.59e-02, grad_scale: 16.0 2023-03-07 11:32:13,829 INFO [train2.py:809] (0/4) Epoch 1, batch 3250, loss[ctc_loss=0.2693, att_loss=0.3466, loss=0.3312, over 16613.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006066, over 47.00 utterances.], tot_loss[ctc_loss=0.3166, att_loss=0.3606, loss=0.3518, over 3270850.75 frames. utt_duration=1192 frames, utt_pad_proportion=0.06952, over 10991.69 utterances.], batch size: 47, lr: 4.58e-02, grad_scale: 16.0 2023-03-07 11:32:41,063 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9975, 2.3147, 2.5346, 3.5268, 3.1311, 3.1178, 2.4487, 3.3029], device='cuda:0'), covar=tensor([0.0340, 0.0690, 0.0867, 0.0268, 0.0314, 0.0509, 0.0698, 0.0244], device='cuda:0'), in_proj_covar=tensor([0.0026, 0.0023, 0.0035, 0.0028, 0.0025, 0.0025, 0.0032, 0.0024], device='cuda:0'), out_proj_covar=tensor([2.0899e-05, 2.3347e-05, 3.5483e-05, 2.2593e-05, 2.1712e-05, 2.1333e-05, 2.7438e-05, 2.1432e-05], device='cuda:0') 2023-03-07 11:33:09,808 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3288.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 11:33:17,830 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3293.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:33:29,892 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.939e+02 4.361e+02 5.805e+02 6.999e+02 1.268e+03, threshold=1.161e+03, percent-clipped=3.0 2023-03-07 11:33:29,935 INFO [train2.py:809] (0/4) Epoch 1, batch 3300, loss[ctc_loss=0.285, att_loss=0.3519, loss=0.3385, over 17094.00 frames. utt_duration=1222 frames, utt_pad_proportion=0.01661, over 56.00 utterances.], tot_loss[ctc_loss=0.3141, att_loss=0.3596, loss=0.3505, over 3264837.92 frames. utt_duration=1169 frames, utt_pad_proportion=0.07709, over 11182.41 utterances.], batch size: 56, lr: 4.57e-02, grad_scale: 16.0 2023-03-07 11:34:30,955 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3341.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:34:43,732 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3349.0, num_to_drop=2, layers_to_drop={0, 2} 2023-03-07 11:34:46,401 INFO [train2.py:809] (0/4) Epoch 1, batch 3350, loss[ctc_loss=0.2892, att_loss=0.3363, loss=0.3269, over 15943.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007015, over 41.00 utterances.], tot_loss[ctc_loss=0.3086, att_loss=0.3565, loss=0.3469, over 3259477.02 frames. utt_duration=1193 frames, utt_pad_proportion=0.07233, over 10942.18 utterances.], batch size: 41, lr: 4.56e-02, grad_scale: 16.0 2023-03-07 11:35:23,012 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3375.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:35:29,483 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9265, 3.5144, 3.0059, 1.7948, 3.7973, 3.9950, 3.2033, 2.4916], device='cuda:0'), covar=tensor([0.0207, 0.0346, 0.0474, 0.1764, 0.0237, 0.0171, 0.1057, 0.1297], device='cuda:0'), in_proj_covar=tensor([0.0038, 0.0038, 0.0038, 0.0054, 0.0037, 0.0028, 0.0034, 0.0053], device='cuda:0'), out_proj_covar=tensor([2.8123e-05, 2.7107e-05, 3.0748e-05, 4.5992e-05, 2.7304e-05, 2.1731e-05, 2.8213e-05, 4.5572e-05], device='cuda:0') 2023-03-07 11:35:31,089 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1189, 4.0244, 4.4943, 4.3913, 4.7310, 3.4299, 4.0905, 4.4711], device='cuda:0'), covar=tensor([0.0133, 0.0166, 0.0179, 0.0134, 0.0072, 0.0237, 0.0187, 0.0156], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0016, 0.0018, 0.0020, 0.0017, 0.0019, 0.0020, 0.0023], device='cuda:0'), out_proj_covar=tensor([1.8943e-05, 1.8322e-05, 2.1480e-05, 1.9606e-05, 1.6983e-05, 2.3709e-05, 2.0405e-05, 2.4964e-05], device='cuda:0') 2023-03-07 11:35:44,634 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3389.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:36:04,116 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.598e+02 4.455e+02 5.465e+02 6.622e+02 1.273e+03, threshold=1.093e+03, percent-clipped=2.0 2023-03-07 11:36:04,158 INFO [train2.py:809] (0/4) Epoch 1, batch 3400, loss[ctc_loss=0.2633, att_loss=0.3299, loss=0.3166, over 15957.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006284, over 41.00 utterances.], tot_loss[ctc_loss=0.3023, att_loss=0.3531, loss=0.343, over 3257425.98 frames. utt_duration=1240 frames, utt_pad_proportion=0.06204, over 10524.71 utterances.], batch size: 41, lr: 4.55e-02, grad_scale: 16.0 2023-03-07 11:37:19,748 INFO [train2.py:809] (0/4) Epoch 1, batch 3450, loss[ctc_loss=0.2543, att_loss=0.3058, loss=0.2955, over 15385.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.01022, over 35.00 utterances.], tot_loss[ctc_loss=0.3022, att_loss=0.3534, loss=0.3431, over 3266815.26 frames. utt_duration=1248 frames, utt_pad_proportion=0.05696, over 10483.01 utterances.], batch size: 35, lr: 4.54e-02, grad_scale: 16.0 2023-03-07 11:37:33,676 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-07 11:38:03,905 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3480.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:38:05,695 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2023-03-07 11:38:35,895 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.427e+02 4.628e+02 6.013e+02 7.728e+02 1.949e+03, threshold=1.203e+03, percent-clipped=6.0 2023-03-07 11:38:35,938 INFO [train2.py:809] (0/4) Epoch 1, batch 3500, loss[ctc_loss=0.3054, att_loss=0.3451, loss=0.3371, over 15642.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008376, over 37.00 utterances.], tot_loss[ctc_loss=0.3035, att_loss=0.3538, loss=0.3438, over 3267747.58 frames. utt_duration=1223 frames, utt_pad_proportion=0.06312, over 10701.34 utterances.], batch size: 37, lr: 4.53e-02, grad_scale: 16.0 2023-03-07 11:38:51,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-03-07 11:39:16,983 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3528.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:39:45,657 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3546.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 11:39:52,690 INFO [train2.py:809] (0/4) Epoch 1, batch 3550, loss[ctc_loss=0.2619, att_loss=0.3293, loss=0.3158, over 16682.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006795, over 46.00 utterances.], tot_loss[ctc_loss=0.3049, att_loss=0.3543, loss=0.3444, over 3265652.27 frames. utt_duration=1238 frames, utt_pad_proportion=0.06113, over 10561.47 utterances.], batch size: 46, lr: 4.51e-02, grad_scale: 16.0 2023-03-07 11:40:02,243 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4264, 4.1806, 4.3449, 4.2096, 4.7586, 4.6499, 3.0874, 3.5946], device='cuda:0'), covar=tensor([0.0108, 0.0207, 0.0144, 0.0217, 0.0053, 0.0084, 0.0815, 0.0441], device='cuda:0'), in_proj_covar=tensor([0.0042, 0.0037, 0.0046, 0.0036, 0.0034, 0.0039, 0.0066, 0.0055], device='cuda:0'), out_proj_covar=tensor([3.0878e-05, 3.3488e-05, 3.4639e-05, 3.0807e-05, 2.8248e-05, 2.8448e-05, 5.8531e-05, 4.3224e-05], device='cuda:0') 2023-03-07 11:40:29,982 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0612, 3.0503, 2.9217, 3.1992, 3.2453, 3.2454, 2.4812, 2.4385], device='cuda:0'), covar=tensor([0.0174, 0.0224, 0.0265, 0.0221, 0.0122, 0.0130, 0.0839, 0.0609], device='cuda:0'), in_proj_covar=tensor([0.0042, 0.0037, 0.0047, 0.0037, 0.0034, 0.0040, 0.0068, 0.0057], device='cuda:0'), out_proj_covar=tensor([3.1421e-05, 3.4228e-05, 3.5984e-05, 3.1733e-05, 2.8701e-05, 2.9267e-05, 6.0174e-05, 4.4997e-05], device='cuda:0') 2023-03-07 11:40:58,957 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3593.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:41:10,752 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.449e+02 6.503e+02 7.868e+02 1.011e+03 2.300e+03, threshold=1.574e+03, percent-clipped=15.0 2023-03-07 11:41:10,795 INFO [train2.py:809] (0/4) Epoch 1, batch 3600, loss[ctc_loss=0.2793, att_loss=0.305, loss=0.2998, over 15899.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008443, over 39.00 utterances.], tot_loss[ctc_loss=0.3028, att_loss=0.3536, loss=0.3434, over 3265911.83 frames. utt_duration=1241 frames, utt_pad_proportion=0.06003, over 10536.81 utterances.], batch size: 39, lr: 4.50e-02, grad_scale: 16.0 2023-03-07 11:41:20,362 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3607.0, num_to_drop=2, layers_to_drop={1, 3} 2023-03-07 11:41:36,658 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-07 11:42:12,226 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3641.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:42:17,579 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3644.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 11:42:27,537 INFO [train2.py:809] (0/4) Epoch 1, batch 3650, loss[ctc_loss=0.2756, att_loss=0.3442, loss=0.3305, over 17285.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01213, over 55.00 utterances.], tot_loss[ctc_loss=0.3004, att_loss=0.3524, loss=0.342, over 3270477.62 frames. utt_duration=1242 frames, utt_pad_proportion=0.0586, over 10542.44 utterances.], batch size: 55, lr: 4.49e-02, grad_scale: 16.0 2023-03-07 11:43:05,254 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3675.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:43:06,912 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3676.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 11:43:47,093 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.430e+02 5.749e+02 7.107e+02 9.009e+02 1.799e+03, threshold=1.421e+03, percent-clipped=2.0 2023-03-07 11:43:47,136 INFO [train2.py:809] (0/4) Epoch 1, batch 3700, loss[ctc_loss=0.2866, att_loss=0.3383, loss=0.328, over 16410.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007178, over 44.00 utterances.], tot_loss[ctc_loss=0.2987, att_loss=0.3525, loss=0.3417, over 3279021.42 frames. utt_duration=1228 frames, utt_pad_proportion=0.05947, over 10692.56 utterances.], batch size: 44, lr: 4.48e-02, grad_scale: 16.0 2023-03-07 11:44:12,507 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7133, 4.4543, 4.7203, 4.7652, 4.6763, 3.8948, 4.4919, 4.3658], device='cuda:0'), covar=tensor([0.0106, 0.0087, 0.0111, 0.0090, 0.0053, 0.0128, 0.0120, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0018, 0.0016, 0.0018, 0.0021, 0.0018, 0.0019, 0.0020, 0.0025], device='cuda:0'), out_proj_covar=tensor([2.0181e-05, 2.0906e-05, 2.4795e-05, 2.2190e-05, 1.9329e-05, 2.5376e-05, 2.2492e-05, 2.8601e-05], device='cuda:0') 2023-03-07 11:44:21,455 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3723.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:44:45,185 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3737.0, num_to_drop=2, layers_to_drop={2, 3} 2023-03-07 11:45:06,155 INFO [train2.py:809] (0/4) Epoch 1, batch 3750, loss[ctc_loss=0.3014, att_loss=0.3664, loss=0.3534, over 17332.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.01004, over 55.00 utterances.], tot_loss[ctc_loss=0.2957, att_loss=0.3506, loss=0.3396, over 3283077.58 frames. utt_duration=1252 frames, utt_pad_proportion=0.05228, over 10504.84 utterances.], batch size: 55, lr: 4.47e-02, grad_scale: 16.0 2023-03-07 11:46:23,999 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.021e+02 4.717e+02 6.819e+02 8.989e+02 1.986e+03, threshold=1.364e+03, percent-clipped=6.0 2023-03-07 11:46:24,042 INFO [train2.py:809] (0/4) Epoch 1, batch 3800, loss[ctc_loss=0.2736, att_loss=0.3103, loss=0.303, over 12355.00 frames. utt_duration=1832 frames, utt_pad_proportion=0.1436, over 27.00 utterances.], tot_loss[ctc_loss=0.2962, att_loss=0.3513, loss=0.3403, over 3280637.56 frames. utt_duration=1238 frames, utt_pad_proportion=0.05476, over 10613.29 utterances.], batch size: 27, lr: 4.46e-02, grad_scale: 16.0 2023-03-07 11:46:58,991 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3265, 4.7129, 5.0384, 4.9624, 4.5369, 4.7505, 5.0280, 4.9298], device='cuda:0'), covar=tensor([0.0245, 0.0219, 0.0123, 0.0141, 0.0252, 0.0099, 0.0221, 0.0108], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0065, 0.0073, 0.0059, 0.0078, 0.0066, 0.0075, 0.0059], device='cuda:0'), out_proj_covar=tensor([1.1258e-04, 8.2440e-05, 8.3590e-05, 7.0031e-05, 9.9821e-05, 8.9647e-05, 9.7200e-05, 6.4231e-05], device='cuda:0') 2023-03-07 11:47:43,586 INFO [train2.py:809] (0/4) Epoch 1, batch 3850, loss[ctc_loss=0.2709, att_loss=0.3386, loss=0.325, over 16178.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006443, over 41.00 utterances.], tot_loss[ctc_loss=0.2936, att_loss=0.3501, loss=0.3388, over 3281383.78 frames. utt_duration=1238 frames, utt_pad_proportion=0.05553, over 10614.06 utterances.], batch size: 41, lr: 4.45e-02, grad_scale: 16.0 2023-03-07 11:49:01,029 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.357e+02 5.503e+02 6.521e+02 8.081e+02 1.434e+03, threshold=1.304e+03, percent-clipped=2.0 2023-03-07 11:49:01,073 INFO [train2.py:809] (0/4) Epoch 1, batch 3900, loss[ctc_loss=0.2814, att_loss=0.3204, loss=0.3126, over 15504.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008658, over 36.00 utterances.], tot_loss[ctc_loss=0.293, att_loss=0.3498, loss=0.3384, over 3275875.00 frames. utt_duration=1227 frames, utt_pad_proportion=0.06038, over 10694.16 utterances.], batch size: 36, lr: 4.44e-02, grad_scale: 16.0 2023-03-07 11:49:02,613 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=3902.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 11:49:39,813 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=3926.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:50:08,100 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=3944.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 11:50:11,347 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6873, 3.5567, 3.5924, 3.8500, 2.9372, 3.8021, 3.3672, 3.6292], device='cuda:0'), covar=tensor([0.0378, 0.0218, 0.0326, 0.0388, 0.2814, 0.0388, 0.0479, 0.0214], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0077, 0.0101, 0.0150, 0.0218, 0.0074, 0.0123, 0.0111], device='cuda:0'), out_proj_covar=tensor([4.9706e-05, 5.2771e-05, 6.0915e-05, 9.3366e-05, 1.4633e-04, 4.6561e-05, 7.1162e-05, 5.5944e-05], device='cuda:0') 2023-03-07 11:50:18,519 INFO [train2.py:809] (0/4) Epoch 1, batch 3950, loss[ctc_loss=0.2882, att_loss=0.352, loss=0.3393, over 17011.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008239, over 51.00 utterances.], tot_loss[ctc_loss=0.2917, att_loss=0.3491, loss=0.3377, over 3276750.54 frames. utt_duration=1221 frames, utt_pad_proportion=0.06161, over 10745.21 utterances.], batch size: 51, lr: 4.43e-02, grad_scale: 16.0 2023-03-07 11:50:43,521 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1925, 1.8681, 2.7784, 2.6212, 3.4506, 2.5737, 2.7438, 1.7427], device='cuda:0'), covar=tensor([0.0485, 0.0984, 0.0426, 0.0681, 0.0273, 0.0572, 0.0758, 0.2337], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0071, 0.0068, 0.0094, 0.0090, 0.0078, 0.0068, 0.0112], device='cuda:0'), out_proj_covar=tensor([5.2927e-05, 5.2005e-05, 5.0180e-05, 6.3243e-05, 5.3334e-05, 7.0222e-05, 5.1086e-05, 9.4293e-05], device='cuda:0') 2023-03-07 11:51:09,989 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-1.pt 2023-03-07 11:51:37,723 INFO [train2.py:809] (0/4) Epoch 2, batch 0, loss[ctc_loss=0.293, att_loss=0.3635, loss=0.3494, over 17281.00 frames. utt_duration=1099 frames, utt_pad_proportion=0.03946, over 63.00 utterances.], tot_loss[ctc_loss=0.293, att_loss=0.3635, loss=0.3494, over 17281.00 frames. utt_duration=1099 frames, utt_pad_proportion=0.03946, over 63.00 utterances.], batch size: 63, lr: 4.34e-02, grad_scale: 8.0 2023-03-07 11:51:37,725 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 11:51:49,517 INFO [train2.py:843] (0/4) Epoch 2, validation: ctc_loss=0.1604, att_loss=0.2954, loss=0.2684, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 11:51:49,518 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 15678MB 2023-03-07 11:51:52,965 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=3987.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:52:01,257 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=3992.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 11:52:13,728 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-4000.pt 2023-03-07 11:52:20,458 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.599e+02 5.286e+02 6.545e+02 8.035e+02 1.268e+03, threshold=1.309e+03, percent-clipped=0.0 2023-03-07 11:53:10,009 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4032.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 11:53:14,640 INFO [train2.py:809] (0/4) Epoch 2, batch 50, loss[ctc_loss=0.2812, att_loss=0.3383, loss=0.3269, over 16114.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.007034, over 42.00 utterances.], tot_loss[ctc_loss=0.285, att_loss=0.3479, loss=0.3353, over 734109.97 frames. utt_duration=1309 frames, utt_pad_proportion=0.05063, over 2245.86 utterances.], batch size: 42, lr: 4.33e-02, grad_scale: 8.0 2023-03-07 11:53:35,030 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7061, 5.8356, 5.1209, 5.8382, 5.1886, 5.3305, 5.1103, 5.3099], device='cuda:0'), covar=tensor([0.0847, 0.0717, 0.0746, 0.0499, 0.0730, 0.0966, 0.1828, 0.1320], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0225, 0.0186, 0.0167, 0.0148, 0.0225, 0.0243, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-03-07 11:54:37,347 INFO [train2.py:809] (0/4) Epoch 2, batch 100, loss[ctc_loss=0.3267, att_loss=0.3664, loss=0.3585, over 17345.00 frames. utt_duration=879.8 frames, utt_pad_proportion=0.07974, over 79.00 utterances.], tot_loss[ctc_loss=0.2884, att_loss=0.3481, loss=0.3362, over 1296314.22 frames. utt_duration=1227 frames, utt_pad_proportion=0.06791, over 4229.79 utterances.], batch size: 79, lr: 4.31e-02, grad_scale: 8.0 2023-03-07 11:55:05,385 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.894e+02 5.347e+02 6.194e+02 8.279e+02 2.356e+03, threshold=1.239e+03, percent-clipped=4.0 2023-03-07 11:55:18,593 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0075, 1.5547, 2.7859, 2.4045, 3.4303, 2.3173, 2.6181, 1.9231], device='cuda:0'), covar=tensor([0.0335, 0.1143, 0.0374, 0.0815, 0.0303, 0.0641, 0.0827, 0.1745], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0066, 0.0064, 0.0089, 0.0080, 0.0069, 0.0064, 0.0108], device='cuda:0'), out_proj_covar=tensor([4.7068e-05, 4.9572e-05, 4.9170e-05, 6.1849e-05, 4.9795e-05, 6.0747e-05, 4.8373e-05, 8.9587e-05], device='cuda:0') 2023-03-07 11:56:00,898 INFO [train2.py:809] (0/4) Epoch 2, batch 150, loss[ctc_loss=0.2417, att_loss=0.2979, loss=0.2867, over 15645.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008297, over 37.00 utterances.], tot_loss[ctc_loss=0.2848, att_loss=0.3468, loss=0.3344, over 1746521.44 frames. utt_duration=1236 frames, utt_pad_proportion=0.05552, over 5657.55 utterances.], batch size: 37, lr: 4.30e-02, grad_scale: 8.0 2023-03-07 11:56:11,603 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4141.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:56:41,930 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-03-07 11:57:24,047 INFO [train2.py:809] (0/4) Epoch 2, batch 200, loss[ctc_loss=0.2099, att_loss=0.2846, loss=0.2697, over 12311.00 frames. utt_duration=1825 frames, utt_pad_proportion=0.1488, over 27.00 utterances.], tot_loss[ctc_loss=0.2834, att_loss=0.346, loss=0.3335, over 2083107.97 frames. utt_duration=1265 frames, utt_pad_proportion=0.04982, over 6596.81 utterances.], batch size: 27, lr: 4.29e-02, grad_scale: 8.0 2023-03-07 11:57:52,227 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.168e+02 5.150e+02 6.618e+02 8.460e+02 1.655e+03, threshold=1.324e+03, percent-clipped=5.0 2023-03-07 11:57:52,580 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4202.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 11:57:52,653 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4202.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 11:58:28,334 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-07 11:58:46,861 INFO [train2.py:809] (0/4) Epoch 2, batch 250, loss[ctc_loss=0.3705, att_loss=0.3895, loss=0.3857, over 17065.00 frames. utt_duration=691 frames, utt_pad_proportion=0.1341, over 99.00 utterances.], tot_loss[ctc_loss=0.2794, att_loss=0.3427, loss=0.3301, over 2344366.54 frames. utt_duration=1292 frames, utt_pad_proportion=0.0453, over 7268.64 utterances.], batch size: 99, lr: 4.28e-02, grad_scale: 8.0 2023-03-07 11:59:11,768 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4250.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 11:59:26,931 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-03-07 12:00:03,997 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4282.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:00:08,476 INFO [train2.py:809] (0/4) Epoch 2, batch 300, loss[ctc_loss=0.2846, att_loss=0.3522, loss=0.3387, over 16869.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007288, over 49.00 utterances.], tot_loss[ctc_loss=0.2818, att_loss=0.3455, loss=0.3328, over 2557247.46 frames. utt_duration=1230 frames, utt_pad_proportion=0.05788, over 8327.66 utterances.], batch size: 49, lr: 4.27e-02, grad_scale: 8.0 2023-03-07 12:00:14,392 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=4288.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:00:22,945 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-03-07 12:00:36,610 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.675e+02 5.336e+02 6.631e+02 8.677e+02 1.956e+03, threshold=1.326e+03, percent-clipped=4.0 2023-03-07 12:01:25,933 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4332.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 12:01:30,247 INFO [train2.py:809] (0/4) Epoch 2, batch 350, loss[ctc_loss=0.2556, att_loss=0.3254, loss=0.3114, over 16380.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.008181, over 44.00 utterances.], tot_loss[ctc_loss=0.2793, att_loss=0.3444, loss=0.3314, over 2716224.36 frames. utt_duration=1229 frames, utt_pad_proportion=0.05587, over 8847.99 utterances.], batch size: 44, lr: 4.26e-02, grad_scale: 8.0 2023-03-07 12:01:52,885 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=4349.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:02:19,122 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-07 12:02:42,478 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4380.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 12:02:49,846 INFO [train2.py:809] (0/4) Epoch 2, batch 400, loss[ctc_loss=0.2896, att_loss=0.3635, loss=0.3488, over 16884.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006524, over 49.00 utterances.], tot_loss[ctc_loss=0.2787, att_loss=0.3445, loss=0.3314, over 2837516.50 frames. utt_duration=1216 frames, utt_pad_proportion=0.06131, over 9345.54 utterances.], batch size: 49, lr: 4.25e-02, grad_scale: 8.0 2023-03-07 12:03:17,595 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.419e+02 5.648e+02 6.669e+02 8.746e+02 3.020e+03, threshold=1.334e+03, percent-clipped=4.0 2023-03-07 12:04:10,866 INFO [train2.py:809] (0/4) Epoch 2, batch 450, loss[ctc_loss=0.2983, att_loss=0.3641, loss=0.351, over 17472.00 frames. utt_duration=1014 frames, utt_pad_proportion=0.04229, over 69.00 utterances.], tot_loss[ctc_loss=0.281, att_loss=0.347, loss=0.3338, over 2944117.86 frames. utt_duration=1227 frames, utt_pad_proportion=0.05617, over 9612.04 utterances.], batch size: 69, lr: 4.24e-02, grad_scale: 8.0 2023-03-07 12:05:33,256 INFO [train2.py:809] (0/4) Epoch 2, batch 500, loss[ctc_loss=0.3137, att_loss=0.364, loss=0.3539, over 17114.00 frames. utt_duration=693.1 frames, utt_pad_proportion=0.1304, over 99.00 utterances.], tot_loss[ctc_loss=0.2775, att_loss=0.3439, loss=0.3306, over 3012045.29 frames. utt_duration=1243 frames, utt_pad_proportion=0.0535, over 9700.83 utterances.], batch size: 99, lr: 4.23e-02, grad_scale: 8.0 2023-03-07 12:05:51,003 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7127, 5.4258, 5.0564, 4.8288, 5.4653, 5.1418, 5.1250, 5.1136], device='cuda:0'), covar=tensor([0.0619, 0.0176, 0.0209, 0.0446, 0.0253, 0.0269, 0.0185, 0.0227], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0123, 0.0090, 0.0094, 0.0130, 0.0148, 0.0107, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-07 12:05:52,554 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4497.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:06:00,026 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.132e+02 4.968e+02 5.977e+02 8.490e+02 1.955e+03, threshold=1.195e+03, percent-clipped=9.0 2023-03-07 12:06:53,452 INFO [train2.py:809] (0/4) Epoch 2, batch 550, loss[ctc_loss=0.2554, att_loss=0.3389, loss=0.3222, over 16622.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.00459, over 47.00 utterances.], tot_loss[ctc_loss=0.2754, att_loss=0.3423, loss=0.3289, over 3070339.54 frames. utt_duration=1241 frames, utt_pad_proportion=0.05408, over 9906.09 utterances.], batch size: 47, lr: 4.22e-02, grad_scale: 8.0 2023-03-07 12:07:01,776 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2850, 4.6249, 5.0614, 4.9678, 1.8662, 3.7836, 4.9536, 3.8777], device='cuda:0'), covar=tensor([0.2376, 0.1213, 0.0687, 0.1206, 3.3794, 0.3810, 0.1093, 0.8252], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0067, 0.0092, 0.0102, 0.0272, 0.0147, 0.0102, 0.0095], device='cuda:0'), out_proj_covar=tensor([5.4662e-05, 3.1077e-05, 3.4584e-05, 3.9198e-05, 1.4926e-04, 6.8240e-05, 3.7874e-05, 5.4625e-05], device='cuda:0') 2023-03-07 12:08:08,817 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4582.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:08:13,846 INFO [train2.py:809] (0/4) Epoch 2, batch 600, loss[ctc_loss=0.2747, att_loss=0.3485, loss=0.3337, over 17430.00 frames. utt_duration=1108 frames, utt_pad_proportion=0.03037, over 63.00 utterances.], tot_loss[ctc_loss=0.2769, att_loss=0.3428, loss=0.3296, over 3120089.48 frames. utt_duration=1243 frames, utt_pad_proportion=0.05339, over 10052.81 utterances.], batch size: 63, lr: 4.21e-02, grad_scale: 8.0 2023-03-07 12:08:40,958 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.676e+02 4.778e+02 6.448e+02 8.266e+02 2.224e+03, threshold=1.290e+03, percent-clipped=6.0 2023-03-07 12:09:12,460 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-07 12:09:24,677 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-07 12:09:26,909 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4630.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:09:35,421 INFO [train2.py:809] (0/4) Epoch 2, batch 650, loss[ctc_loss=0.3685, att_loss=0.3983, loss=0.3923, over 14401.00 frames. utt_duration=396 frames, utt_pad_proportion=0.3077, over 146.00 utterances.], tot_loss[ctc_loss=0.2742, att_loss=0.3415, loss=0.328, over 3156587.26 frames. utt_duration=1250 frames, utt_pad_proportion=0.05123, over 10109.99 utterances.], batch size: 146, lr: 4.20e-02, grad_scale: 8.0 2023-03-07 12:09:50,146 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=4644.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:10:56,697 INFO [train2.py:809] (0/4) Epoch 2, batch 700, loss[ctc_loss=0.3085, att_loss=0.3678, loss=0.3559, over 16778.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005889, over 48.00 utterances.], tot_loss[ctc_loss=0.2738, att_loss=0.3411, loss=0.3276, over 3182533.27 frames. utt_duration=1236 frames, utt_pad_proportion=0.05499, over 10311.60 utterances.], batch size: 48, lr: 4.19e-02, grad_scale: 8.0 2023-03-07 12:11:23,312 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.556e+02 4.876e+02 6.372e+02 7.940e+02 1.497e+03, threshold=1.274e+03, percent-clipped=2.0 2023-03-07 12:12:18,128 INFO [train2.py:809] (0/4) Epoch 2, batch 750, loss[ctc_loss=0.2647, att_loss=0.3341, loss=0.3202, over 15960.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006607, over 41.00 utterances.], tot_loss[ctc_loss=0.2731, att_loss=0.3409, loss=0.3273, over 3204091.92 frames. utt_duration=1220 frames, utt_pad_proportion=0.05918, over 10521.45 utterances.], batch size: 41, lr: 4.18e-02, grad_scale: 8.0 2023-03-07 12:12:43,401 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8159, 5.9671, 5.4526, 5.9723, 5.4924, 5.5950, 5.4065, 5.3501], device='cuda:0'), covar=tensor([0.0928, 0.0660, 0.0596, 0.0550, 0.0540, 0.0910, 0.1873, 0.1450], device='cuda:0'), in_proj_covar=tensor([0.0192, 0.0227, 0.0199, 0.0177, 0.0162, 0.0240, 0.0254, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 12:13:38,639 INFO [train2.py:809] (0/4) Epoch 2, batch 800, loss[ctc_loss=0.2381, att_loss=0.3128, loss=0.2979, over 15951.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.005899, over 41.00 utterances.], tot_loss[ctc_loss=0.2717, att_loss=0.3401, loss=0.3264, over 3216880.10 frames. utt_duration=1236 frames, utt_pad_proportion=0.05604, over 10419.99 utterances.], batch size: 41, lr: 4.17e-02, grad_scale: 8.0 2023-03-07 12:13:57,529 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4797.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:14:01,356 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.19 vs. limit=2.0 2023-03-07 12:14:05,888 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.406e+02 5.666e+02 6.834e+02 8.456e+02 1.435e+03, threshold=1.367e+03, percent-clipped=1.0 2023-03-07 12:14:52,301 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-07 12:14:58,934 INFO [train2.py:809] (0/4) Epoch 2, batch 850, loss[ctc_loss=0.2303, att_loss=0.321, loss=0.3028, over 17015.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008728, over 51.00 utterances.], tot_loss[ctc_loss=0.272, att_loss=0.3405, loss=0.3268, over 3232237.62 frames. utt_duration=1230 frames, utt_pad_proportion=0.05757, over 10524.87 utterances.], batch size: 51, lr: 4.16e-02, grad_scale: 8.0 2023-03-07 12:15:14,882 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4845.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:16:05,520 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.21 vs. limit=5.0 2023-03-07 12:16:16,492 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5994, 4.0561, 3.8404, 3.7868, 4.1254, 3.9793, 3.8875, 3.9067], device='cuda:0'), covar=tensor([0.0584, 0.0263, 0.0268, 0.0321, 0.0229, 0.0252, 0.0252, 0.0249], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0123, 0.0090, 0.0094, 0.0129, 0.0143, 0.0112, 0.0123], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-07 12:16:19,353 INFO [train2.py:809] (0/4) Epoch 2, batch 900, loss[ctc_loss=0.2821, att_loss=0.3503, loss=0.3366, over 16959.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007918, over 50.00 utterances.], tot_loss[ctc_loss=0.2712, att_loss=0.3405, loss=0.3266, over 3244436.46 frames. utt_duration=1220 frames, utt_pad_proportion=0.06023, over 10649.43 utterances.], batch size: 50, lr: 4.15e-02, grad_scale: 8.0 2023-03-07 12:16:46,358 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.888e+02 4.698e+02 5.859e+02 7.531e+02 1.383e+03, threshold=1.172e+03, percent-clipped=2.0 2023-03-07 12:17:13,386 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-07 12:17:15,804 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2764, 2.7324, 3.2875, 3.3715, 2.6216, 2.8841, 1.7332, 3.3224], device='cuda:0'), covar=tensor([0.0323, 0.0279, 0.0584, 0.0360, 0.0473, 0.0841, 0.1179, 0.0143], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0029, 0.0053, 0.0047, 0.0036, 0.0059, 0.0055, 0.0030], device='cuda:0'), out_proj_covar=tensor([3.9588e-05, 3.9456e-05, 6.9170e-05, 4.8904e-05, 4.2524e-05, 6.9374e-05, 5.7792e-05, 3.3900e-05], device='cuda:0') 2023-03-07 12:17:40,695 INFO [train2.py:809] (0/4) Epoch 2, batch 950, loss[ctc_loss=0.2431, att_loss=0.3204, loss=0.3049, over 16314.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.007053, over 45.00 utterances.], tot_loss[ctc_loss=0.2691, att_loss=0.339, loss=0.325, over 3251610.13 frames. utt_duration=1246 frames, utt_pad_proportion=0.05311, over 10449.78 utterances.], batch size: 45, lr: 4.14e-02, grad_scale: 8.0 2023-03-07 12:17:55,059 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=4944.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:19:01,004 INFO [train2.py:809] (0/4) Epoch 2, batch 1000, loss[ctc_loss=0.2126, att_loss=0.3051, loss=0.2866, over 16406.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007344, over 44.00 utterances.], tot_loss[ctc_loss=0.266, att_loss=0.337, loss=0.3228, over 3249417.46 frames. utt_duration=1258 frames, utt_pad_proportion=0.05287, over 10347.38 utterances.], batch size: 44, lr: 4.13e-02, grad_scale: 8.0 2023-03-07 12:19:12,012 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=4992.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:19:27,689 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.851e+02 4.989e+02 6.201e+02 7.841e+02 1.460e+03, threshold=1.240e+03, percent-clipped=6.0 2023-03-07 12:20:21,707 INFO [train2.py:809] (0/4) Epoch 2, batch 1050, loss[ctc_loss=0.3747, att_loss=0.3964, loss=0.3921, over 14196.00 frames. utt_duration=393 frames, utt_pad_proportion=0.3177, over 145.00 utterances.], tot_loss[ctc_loss=0.2652, att_loss=0.3368, loss=0.3225, over 3260642.89 frames. utt_duration=1257 frames, utt_pad_proportion=0.05151, over 10390.14 utterances.], batch size: 145, lr: 4.12e-02, grad_scale: 8.0 2023-03-07 12:20:29,914 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3412, 4.5461, 5.3196, 4.3285, 3.9831, 4.2758, 4.6229, 4.6380], device='cuda:0'), covar=tensor([0.0229, 0.0705, 0.0096, 0.0875, 0.1570, 0.0681, 0.0324, 0.0352], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0145, 0.0089, 0.0165, 0.0182, 0.0106, 0.0070, 0.0071], device='cuda:0'), out_proj_covar=tensor([3.9068e-05, 7.7401e-05, 4.5084e-05, 1.0303e-04, 1.1192e-04, 7.0947e-05, 4.2713e-05, 4.1868e-05], device='cuda:0') 2023-03-07 12:21:38,067 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5408, 4.1388, 4.4831, 4.6309, 4.4566, 3.0643, 4.3349, 2.7651], device='cuda:0'), covar=tensor([0.0113, 0.0112, 0.0202, 0.0099, 0.0101, 0.0251, 0.0121, 0.0517], device='cuda:0'), in_proj_covar=tensor([0.0021, 0.0020, 0.0020, 0.0024, 0.0023, 0.0023, 0.0024, 0.0038], device='cuda:0'), out_proj_covar=tensor([3.1276e-05, 3.3963e-05, 3.8454e-05, 3.4153e-05, 3.4642e-05, 4.1014e-05, 3.5099e-05, 5.7498e-05], device='cuda:0') 2023-03-07 12:21:42,517 INFO [train2.py:809] (0/4) Epoch 2, batch 1100, loss[ctc_loss=0.2792, att_loss=0.3517, loss=0.3372, over 17039.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.01056, over 53.00 utterances.], tot_loss[ctc_loss=0.2666, att_loss=0.3379, loss=0.3236, over 3265939.03 frames. utt_duration=1205 frames, utt_pad_proportion=0.06393, over 10858.94 utterances.], batch size: 53, lr: 4.11e-02, grad_scale: 8.0 2023-03-07 12:22:09,662 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.609e+02 5.009e+02 6.335e+02 7.458e+02 1.646e+03, threshold=1.267e+03, percent-clipped=3.0 2023-03-07 12:23:03,365 INFO [train2.py:809] (0/4) Epoch 2, batch 1150, loss[ctc_loss=0.2149, att_loss=0.3153, loss=0.2952, over 16410.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007042, over 44.00 utterances.], tot_loss[ctc_loss=0.2655, att_loss=0.3368, loss=0.3225, over 3264739.87 frames. utt_duration=1236 frames, utt_pad_proportion=0.05763, over 10576.25 utterances.], batch size: 44, lr: 4.10e-02, grad_scale: 8.0 2023-03-07 12:24:22,230 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4146, 4.7028, 4.2393, 4.8129, 4.9242, 4.5730, 4.4079, 4.6825], device='cuda:0'), covar=tensor([0.0114, 0.0207, 0.0171, 0.0123, 0.0104, 0.0109, 0.0200, 0.0155], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0039, 0.0039, 0.0031, 0.0028, 0.0031, 0.0041, 0.0037], device='cuda:0'), out_proj_covar=tensor([5.2810e-05, 5.9958e-05, 6.3950e-05, 4.6374e-05, 4.0168e-05, 5.0086e-05, 6.0165e-05, 5.3114e-05], device='cuda:0') 2023-03-07 12:24:24,938 INFO [train2.py:809] (0/4) Epoch 2, batch 1200, loss[ctc_loss=0.3072, att_loss=0.3775, loss=0.3635, over 17036.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009797, over 52.00 utterances.], tot_loss[ctc_loss=0.2636, att_loss=0.3361, loss=0.3216, over 3261901.06 frames. utt_duration=1257 frames, utt_pad_proportion=0.05332, over 10390.64 utterances.], batch size: 52, lr: 4.08e-02, grad_scale: 8.0 2023-03-07 12:24:52,563 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.06 vs. limit=2.0 2023-03-07 12:24:53,173 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.294e+02 5.285e+02 6.236e+02 7.599e+02 1.317e+03, threshold=1.247e+03, percent-clipped=1.0 2023-03-07 12:25:01,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9301, 2.8065, 3.6946, 4.1275, 4.5705, 4.5468, 2.9973, 2.0561], device='cuda:0'), covar=tensor([0.0343, 0.1050, 0.0581, 0.0339, 0.0100, 0.0137, 0.1231, 0.2008], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0086, 0.0086, 0.0059, 0.0049, 0.0060, 0.0104, 0.0102], device='cuda:0'), out_proj_covar=tensor([5.9945e-05, 7.8549e-05, 7.5621e-05, 6.2890e-05, 4.9865e-05, 4.8504e-05, 9.7154e-05, 8.9171e-05], device='cuda:0') 2023-03-07 12:25:18,087 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-03-07 12:25:45,334 INFO [train2.py:809] (0/4) Epoch 2, batch 1250, loss[ctc_loss=0.2253, att_loss=0.2892, loss=0.2764, over 15634.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009136, over 37.00 utterances.], tot_loss[ctc_loss=0.262, att_loss=0.3342, loss=0.3198, over 3250743.76 frames. utt_duration=1268 frames, utt_pad_proportion=0.05377, over 10266.43 utterances.], batch size: 37, lr: 4.07e-02, grad_scale: 8.0 2023-03-07 12:25:47,603 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2023-03-07 12:26:12,513 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.15 vs. limit=2.0 2023-03-07 12:27:05,707 INFO [train2.py:809] (0/4) Epoch 2, batch 1300, loss[ctc_loss=0.25, att_loss=0.3103, loss=0.2983, over 14508.00 frames. utt_duration=1815 frames, utt_pad_proportion=0.04076, over 32.00 utterances.], tot_loss[ctc_loss=0.2611, att_loss=0.3336, loss=0.3191, over 3258650.48 frames. utt_duration=1261 frames, utt_pad_proportion=0.05412, over 10351.44 utterances.], batch size: 32, lr: 4.06e-02, grad_scale: 8.0 2023-03-07 12:27:28,065 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-03-07 12:27:33,502 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.402e+02 4.574e+02 5.700e+02 7.129e+02 3.450e+03, threshold=1.140e+03, percent-clipped=2.0 2023-03-07 12:28:26,302 INFO [train2.py:809] (0/4) Epoch 2, batch 1350, loss[ctc_loss=0.2358, att_loss=0.3094, loss=0.2946, over 16302.00 frames. utt_duration=1518 frames, utt_pad_proportion=0.006092, over 43.00 utterances.], tot_loss[ctc_loss=0.2603, att_loss=0.3338, loss=0.3191, over 3261713.68 frames. utt_duration=1260 frames, utt_pad_proportion=0.05421, over 10370.63 utterances.], batch size: 43, lr: 4.05e-02, grad_scale: 8.0 2023-03-07 12:28:34,176 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7631, 5.7346, 5.4106, 5.8358, 5.4726, 5.3914, 5.3066, 5.2618], device='cuda:0'), covar=tensor([0.0755, 0.0693, 0.0659, 0.0521, 0.0445, 0.0977, 0.1886, 0.1639], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0240, 0.0209, 0.0180, 0.0164, 0.0255, 0.0264, 0.0237], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 12:29:47,506 INFO [train2.py:809] (0/4) Epoch 2, batch 1400, loss[ctc_loss=0.2348, att_loss=0.3216, loss=0.3042, over 16540.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006362, over 45.00 utterances.], tot_loss[ctc_loss=0.2585, att_loss=0.3328, loss=0.318, over 3259313.75 frames. utt_duration=1257 frames, utt_pad_proportion=0.05556, over 10385.09 utterances.], batch size: 45, lr: 4.04e-02, grad_scale: 8.0 2023-03-07 12:30:15,419 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.881e+02 4.556e+02 5.757e+02 7.116e+02 2.241e+03, threshold=1.151e+03, percent-clipped=2.0 2023-03-07 12:30:27,092 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-03-07 12:31:08,625 INFO [train2.py:809] (0/4) Epoch 2, batch 1450, loss[ctc_loss=0.2597, att_loss=0.3079, loss=0.2983, over 15511.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.008178, over 36.00 utterances.], tot_loss[ctc_loss=0.2568, att_loss=0.3317, loss=0.3167, over 3261915.66 frames. utt_duration=1274 frames, utt_pad_proportion=0.04998, over 10255.44 utterances.], batch size: 36, lr: 4.03e-02, grad_scale: 8.0 2023-03-07 12:31:34,230 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-07 12:32:21,525 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2496, 4.8873, 5.0714, 5.0625, 4.5368, 4.6684, 5.1590, 5.0539], device='cuda:0'), covar=tensor([0.0331, 0.0278, 0.0168, 0.0167, 0.0293, 0.0133, 0.0253, 0.0124], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0086, 0.0093, 0.0070, 0.0100, 0.0082, 0.0094, 0.0076], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-03-07 12:32:29,328 INFO [train2.py:809] (0/4) Epoch 2, batch 1500, loss[ctc_loss=0.2499, att_loss=0.3245, loss=0.3095, over 16012.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007048, over 40.00 utterances.], tot_loss[ctc_loss=0.2545, att_loss=0.3305, loss=0.3153, over 3263846.36 frames. utt_duration=1268 frames, utt_pad_proportion=0.05126, over 10305.43 utterances.], batch size: 40, lr: 4.02e-02, grad_scale: 8.0 2023-03-07 12:32:57,036 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.303e+02 4.697e+02 5.525e+02 6.949e+02 3.924e+03, threshold=1.105e+03, percent-clipped=4.0 2023-03-07 12:33:21,791 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.4899, 1.3526, 1.5654, 1.7035, 2.0391, 1.6704, 1.4005, 2.6403], device='cuda:0'), covar=tensor([0.0921, 0.1243, 0.1120, 0.1101, 0.0614, 0.0995, 0.1098, 0.0468], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0062, 0.0059, 0.0070, 0.0062, 0.0069, 0.0061, 0.0081], device='cuda:0'), out_proj_covar=tensor([5.3435e-05, 5.4144e-05, 5.4990e-05, 4.9770e-05, 4.7706e-05, 7.1060e-05, 5.9049e-05, 4.9154e-05], device='cuda:0') 2023-03-07 12:33:49,178 INFO [train2.py:809] (0/4) Epoch 2, batch 1550, loss[ctc_loss=0.2205, att_loss=0.3128, loss=0.2943, over 16398.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007016, over 44.00 utterances.], tot_loss[ctc_loss=0.256, att_loss=0.3314, loss=0.3163, over 3265976.17 frames. utt_duration=1244 frames, utt_pad_proportion=0.05718, over 10516.39 utterances.], batch size: 44, lr: 4.01e-02, grad_scale: 8.0 2023-03-07 12:34:06,306 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3292, 4.9443, 4.8821, 4.7851, 5.0048, 4.9699, 4.9026, 4.7393], device='cuda:0'), covar=tensor([0.0807, 0.0358, 0.0234, 0.0404, 0.0530, 0.0299, 0.0234, 0.0337], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0133, 0.0091, 0.0101, 0.0138, 0.0149, 0.0116, 0.0125], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 12:34:18,985 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4567, 3.6851, 3.4343, 2.1575, 3.2837, 4.4475, 3.7981, 2.9985], device='cuda:0'), covar=tensor([0.0158, 0.0382, 0.0638, 0.1524, 0.0704, 0.0098, 0.0336, 0.1053], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0078, 0.0080, 0.0111, 0.0102, 0.0053, 0.0058, 0.0109], device='cuda:0'), out_proj_covar=tensor([7.2151e-05, 6.9111e-05, 8.8550e-05, 1.0030e-04, 9.5852e-05, 5.1057e-05, 6.5782e-05, 9.9134e-05], device='cuda:0') 2023-03-07 12:35:10,350 INFO [train2.py:809] (0/4) Epoch 2, batch 1600, loss[ctc_loss=0.2401, att_loss=0.3339, loss=0.3151, over 16463.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006855, over 46.00 utterances.], tot_loss[ctc_loss=0.2549, att_loss=0.331, loss=0.3158, over 3260736.72 frames. utt_duration=1252 frames, utt_pad_proportion=0.05685, over 10428.51 utterances.], batch size: 46, lr: 4.00e-02, grad_scale: 8.0 2023-03-07 12:35:38,630 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.176e+02 4.466e+02 5.617e+02 6.515e+02 1.895e+03, threshold=1.123e+03, percent-clipped=5.0 2023-03-07 12:36:31,275 INFO [train2.py:809] (0/4) Epoch 2, batch 1650, loss[ctc_loss=0.2268, att_loss=0.2985, loss=0.2842, over 14039.00 frames. utt_duration=1813 frames, utt_pad_proportion=0.03928, over 31.00 utterances.], tot_loss[ctc_loss=0.2545, att_loss=0.3309, loss=0.3156, over 3260974.13 frames. utt_duration=1253 frames, utt_pad_proportion=0.05647, over 10420.02 utterances.], batch size: 31, lr: 3.99e-02, grad_scale: 8.0 2023-03-07 12:36:33,786 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-03-07 12:36:48,008 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2463, 0.8118, 1.9840, 0.9138, 2.1399, 1.9185, 1.7606, 2.6050], device='cuda:0'), covar=tensor([0.0558, 0.1403, 0.0780, 0.1084, 0.0464, 0.0819, 0.0693, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0060, 0.0058, 0.0070, 0.0058, 0.0068, 0.0059, 0.0079], device='cuda:0'), out_proj_covar=tensor([4.8133e-05, 5.2117e-05, 5.2304e-05, 5.0364e-05, 4.3356e-05, 6.7640e-05, 5.7373e-05, 4.7673e-05], device='cuda:0') 2023-03-07 12:36:55,769 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5650.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 12:37:51,240 INFO [train2.py:809] (0/4) Epoch 2, batch 1700, loss[ctc_loss=0.2448, att_loss=0.3273, loss=0.3108, over 16332.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005928, over 45.00 utterances.], tot_loss[ctc_loss=0.2528, att_loss=0.3299, loss=0.3145, over 3263778.95 frames. utt_duration=1262 frames, utt_pad_proportion=0.05525, over 10355.37 utterances.], batch size: 45, lr: 3.98e-02, grad_scale: 8.0 2023-03-07 12:38:18,539 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.296e+02 4.646e+02 5.757e+02 7.379e+02 2.321e+03, threshold=1.151e+03, percent-clipped=8.0 2023-03-07 12:38:33,501 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5711.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 12:39:10,811 INFO [train2.py:809] (0/4) Epoch 2, batch 1750, loss[ctc_loss=0.2638, att_loss=0.3397, loss=0.3245, over 17395.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03348, over 63.00 utterances.], tot_loss[ctc_loss=0.2544, att_loss=0.3313, loss=0.3159, over 3265398.09 frames. utt_duration=1227 frames, utt_pad_proportion=0.06324, over 10654.13 utterances.], batch size: 63, lr: 3.97e-02, grad_scale: 8.0 2023-03-07 12:39:44,208 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5755.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:40:07,349 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-03-07 12:40:16,459 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0932, 3.5229, 5.0026, 4.0208, 3.7473, 4.1974, 4.7832, 4.3999], device='cuda:0'), covar=tensor([0.0418, 0.1712, 0.0166, 0.1545, 0.2498, 0.0998, 0.0270, 0.0545], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0163, 0.0096, 0.0187, 0.0210, 0.0117, 0.0073, 0.0078], device='cuda:0'), out_proj_covar=tensor([5.0222e-05, 9.6598e-05, 5.6329e-05, 1.2641e-04, 1.3395e-04, 8.4789e-05, 5.0393e-05, 5.4145e-05], device='cuda:0') 2023-03-07 12:40:31,797 INFO [train2.py:809] (0/4) Epoch 2, batch 1800, loss[ctc_loss=0.304, att_loss=0.3612, loss=0.3497, over 16341.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005472, over 45.00 utterances.], tot_loss[ctc_loss=0.2544, att_loss=0.3323, loss=0.3167, over 3274121.23 frames. utt_duration=1230 frames, utt_pad_proportion=0.0605, over 10658.82 utterances.], batch size: 45, lr: 3.96e-02, grad_scale: 8.0 2023-03-07 12:40:33,777 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4108, 4.5029, 4.7902, 4.2005, 2.1884, 4.8038, 3.0181, 5.1845], device='cuda:0'), covar=tensor([0.0230, 0.0195, 0.0289, 0.0412, 0.4489, 0.0151, 0.1020, 0.0134], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0068, 0.0121, 0.0121, 0.0208, 0.0070, 0.0126, 0.0110], device='cuda:0'), out_proj_covar=tensor([5.8208e-05, 5.5683e-05, 8.8636e-05, 8.5316e-05, 1.4034e-04, 5.5403e-05, 8.8726e-05, 7.4191e-05], device='cuda:0') 2023-03-07 12:40:55,475 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.06 vs. limit=2.0 2023-03-07 12:40:59,739 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.935e+02 4.735e+02 5.722e+02 7.362e+02 1.615e+03, threshold=1.144e+03, percent-clipped=5.0 2023-03-07 12:41:23,382 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5816.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:41:53,264 INFO [train2.py:809] (0/4) Epoch 2, batch 1850, loss[ctc_loss=0.1927, att_loss=0.2873, loss=0.2684, over 16133.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005236, over 42.00 utterances.], tot_loss[ctc_loss=0.2521, att_loss=0.3305, loss=0.3148, over 3272801.01 frames. utt_duration=1247 frames, utt_pad_proportion=0.0561, over 10513.47 utterances.], batch size: 42, lr: 3.95e-02, grad_scale: 8.0 2023-03-07 12:42:30,755 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5858.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:43:13,667 INFO [train2.py:809] (0/4) Epoch 2, batch 1900, loss[ctc_loss=0.2059, att_loss=0.3067, loss=0.2865, over 16895.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.005951, over 49.00 utterances.], tot_loss[ctc_loss=0.2515, att_loss=0.3298, loss=0.3142, over 3263653.55 frames. utt_duration=1229 frames, utt_pad_proportion=0.06339, over 10636.46 utterances.], batch size: 49, lr: 3.95e-02, grad_scale: 8.0 2023-03-07 12:43:41,191 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.885e+02 5.042e+02 5.993e+02 7.245e+02 1.405e+03, threshold=1.199e+03, percent-clipped=1.0 2023-03-07 12:44:09,819 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=5919.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:44:35,118 INFO [train2.py:809] (0/4) Epoch 2, batch 1950, loss[ctc_loss=0.2715, att_loss=0.354, loss=0.3375, over 17322.00 frames. utt_duration=1101 frames, utt_pad_proportion=0.03727, over 63.00 utterances.], tot_loss[ctc_loss=0.2501, att_loss=0.3287, loss=0.313, over 3261415.75 frames. utt_duration=1236 frames, utt_pad_proportion=0.06016, over 10567.77 utterances.], batch size: 63, lr: 3.94e-02, grad_scale: 8.0 2023-03-07 12:44:38,563 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8247, 3.0062, 3.3563, 1.8566, 3.0884, 3.9136, 3.5948, 2.8958], device='cuda:0'), covar=tensor([0.0284, 0.0470, 0.0748, 0.1635, 0.0740, 0.0117, 0.0382, 0.1064], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0088, 0.0088, 0.0117, 0.0112, 0.0058, 0.0064, 0.0119], device='cuda:0'), out_proj_covar=tensor([8.3644e-05, 8.0137e-05, 9.9747e-05, 1.0767e-04, 1.0692e-04, 5.7792e-05, 7.5124e-05, 1.0936e-04], device='cuda:0') 2023-03-07 12:44:56,554 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4932, 3.3252, 3.4094, 2.9573, 3.3953, 3.1755, 1.8610, 3.8280], device='cuda:0'), covar=tensor([0.0566, 0.0186, 0.0552, 0.0542, 0.0213, 0.0671, 0.1089, 0.0086], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0035, 0.0070, 0.0059, 0.0043, 0.0077, 0.0067, 0.0035], device='cuda:0'), out_proj_covar=tensor([6.4445e-05, 5.2139e-05, 9.6915e-05, 7.0739e-05, 5.8837e-05, 1.0206e-04, 7.8489e-05, 4.7348e-05], device='cuda:0') 2023-03-07 12:45:55,276 INFO [train2.py:809] (0/4) Epoch 2, batch 2000, loss[ctc_loss=0.2876, att_loss=0.3572, loss=0.3433, over 17039.00 frames. utt_duration=689.8 frames, utt_pad_proportion=0.1334, over 99.00 utterances.], tot_loss[ctc_loss=0.2495, att_loss=0.3286, loss=0.3128, over 3268927.88 frames. utt_duration=1251 frames, utt_pad_proportion=0.05371, over 10467.83 utterances.], batch size: 99, lr: 3.93e-02, grad_scale: 16.0 2023-03-07 12:46:04,069 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=5990.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:46:13,105 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9987, 4.2659, 4.1229, 4.2983, 4.4999, 4.2529, 4.1524, 4.1331], device='cuda:0'), covar=tensor([0.0147, 0.0258, 0.0143, 0.0166, 0.0114, 0.0123, 0.0267, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0037, 0.0042, 0.0040, 0.0032, 0.0028, 0.0032, 0.0046, 0.0039], device='cuda:0'), out_proj_covar=tensor([6.3099e-05, 7.1557e-05, 7.5209e-05, 5.5740e-05, 4.4670e-05, 5.8438e-05, 7.6472e-05, 6.8369e-05], device='cuda:0') 2023-03-07 12:46:19,409 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-6000.pt 2023-03-07 12:46:26,496 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.443e+02 4.806e+02 6.010e+02 7.878e+02 2.247e+03, threshold=1.202e+03, percent-clipped=5.0 2023-03-07 12:46:33,512 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6006.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 12:46:51,194 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4805, 4.0348, 5.1497, 3.9631, 3.6866, 4.1240, 4.5757, 4.5029], device='cuda:0'), covar=tensor([0.0426, 0.1369, 0.0362, 0.2168, 0.3447, 0.1811, 0.0447, 0.0585], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0169, 0.0101, 0.0203, 0.0236, 0.0127, 0.0078, 0.0085], device='cuda:0'), out_proj_covar=tensor([5.3921e-05, 1.0240e-04, 5.9805e-05, 1.3869e-04, 1.5132e-04, 9.4064e-05, 5.4973e-05, 5.9519e-05], device='cuda:0') 2023-03-07 12:47:19,310 INFO [train2.py:809] (0/4) Epoch 2, batch 2050, loss[ctc_loss=0.2801, att_loss=0.3539, loss=0.3391, over 17081.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008078, over 53.00 utterances.], tot_loss[ctc_loss=0.2523, att_loss=0.3298, loss=0.3143, over 3266224.33 frames. utt_duration=1237 frames, utt_pad_proportion=0.05799, over 10572.16 utterances.], batch size: 53, lr: 3.92e-02, grad_scale: 8.0 2023-03-07 12:47:33,232 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6043.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:47:45,319 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6051.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:47:56,920 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8608, 4.5394, 5.1539, 4.3815, 2.0820, 3.9954, 3.4381, 5.2124], device='cuda:0'), covar=tensor([0.0125, 0.0201, 0.0262, 0.0226, 0.4144, 0.0320, 0.0781, 0.0103], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0070, 0.0129, 0.0122, 0.0214, 0.0076, 0.0130, 0.0112], device='cuda:0'), out_proj_covar=tensor([5.8699e-05, 5.8405e-05, 9.6496e-05, 8.6002e-05, 1.4516e-04, 6.1064e-05, 9.2695e-05, 7.7846e-05], device='cuda:0') 2023-03-07 12:48:40,075 INFO [train2.py:809] (0/4) Epoch 2, batch 2100, loss[ctc_loss=0.3458, att_loss=0.3867, loss=0.3786, over 13971.00 frames. utt_duration=386.8 frames, utt_pad_proportion=0.3284, over 145.00 utterances.], tot_loss[ctc_loss=0.2524, att_loss=0.3301, loss=0.3145, over 3269475.47 frames. utt_duration=1223 frames, utt_pad_proportion=0.06036, over 10706.47 utterances.], batch size: 145, lr: 3.91e-02, grad_scale: 8.0 2023-03-07 12:49:08,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.112e+02 4.379e+02 5.307e+02 7.391e+02 2.758e+03, threshold=1.061e+03, percent-clipped=4.0 2023-03-07 12:49:10,906 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6104.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:49:23,167 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6111.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:49:32,394 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-07 12:50:01,354 INFO [train2.py:809] (0/4) Epoch 2, batch 2150, loss[ctc_loss=0.2407, att_loss=0.338, loss=0.3186, over 17378.00 frames. utt_duration=1105 frames, utt_pad_proportion=0.03344, over 63.00 utterances.], tot_loss[ctc_loss=0.2494, att_loss=0.3281, loss=0.3124, over 3266410.66 frames. utt_duration=1248 frames, utt_pad_proportion=0.05602, over 10483.80 utterances.], batch size: 63, lr: 3.90e-02, grad_scale: 8.0 2023-03-07 12:50:01,690 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6135.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:50:57,754 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6170.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:51:21,402 INFO [train2.py:809] (0/4) Epoch 2, batch 2200, loss[ctc_loss=0.2271, att_loss=0.2875, loss=0.2754, over 15626.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009579, over 37.00 utterances.], tot_loss[ctc_loss=0.2505, att_loss=0.3289, loss=0.3132, over 3262181.53 frames. utt_duration=1230 frames, utt_pad_proportion=0.06153, over 10620.86 utterances.], batch size: 37, lr: 3.89e-02, grad_scale: 8.0 2023-03-07 12:51:39,623 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6196.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:51:50,203 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.181e+02 4.728e+02 5.597e+02 7.112e+02 1.691e+03, threshold=1.119e+03, percent-clipped=4.0 2023-03-07 12:51:50,630 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6203.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:52:04,335 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6211.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:52:08,250 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2023-03-07 12:52:08,899 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6214.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:52:35,403 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5453, 3.5059, 3.6211, 2.1035, 3.5107, 4.4091, 3.9617, 2.9995], device='cuda:0'), covar=tensor([0.0225, 0.0603, 0.0856, 0.1839, 0.0798, 0.0110, 0.0717, 0.1565], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0096, 0.0091, 0.0120, 0.0119, 0.0060, 0.0072, 0.0125], device='cuda:0'), out_proj_covar=tensor([9.0645e-05, 8.9948e-05, 1.0400e-04, 1.1168e-04, 1.1486e-04, 6.1502e-05, 8.4105e-05, 1.1701e-04], device='cuda:0') 2023-03-07 12:52:35,462 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6231.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:52:42,165 INFO [train2.py:809] (0/4) Epoch 2, batch 2250, loss[ctc_loss=0.2773, att_loss=0.3483, loss=0.3341, over 17308.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01172, over 55.00 utterances.], tot_loss[ctc_loss=0.2496, att_loss=0.3282, loss=0.3124, over 3264202.84 frames. utt_duration=1224 frames, utt_pad_proportion=0.06187, over 10683.58 utterances.], batch size: 55, lr: 3.88e-02, grad_scale: 8.0 2023-03-07 12:53:29,754 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6264.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:53:32,797 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6266.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:53:42,112 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6272.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:53:51,286 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.1894, 1.6519, 1.8179, 0.9954, 1.6496, 2.5736, 1.7569, 2.0149], device='cuda:0'), covar=tensor([0.0731, 0.0713, 0.0681, 0.1091, 0.0454, 0.0465, 0.0719, 0.0524], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0064, 0.0063, 0.0070, 0.0063, 0.0073, 0.0066, 0.0090], device='cuda:0'), out_proj_covar=tensor([5.0327e-05, 4.9892e-05, 5.0559e-05, 5.8057e-05, 4.0570e-05, 6.4440e-05, 5.4747e-05, 4.8140e-05], device='cuda:0') 2023-03-07 12:54:02,554 INFO [train2.py:809] (0/4) Epoch 2, batch 2300, loss[ctc_loss=0.2569, att_loss=0.3385, loss=0.3222, over 16117.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.00616, over 42.00 utterances.], tot_loss[ctc_loss=0.2483, att_loss=0.3274, loss=0.3115, over 3267139.82 frames. utt_duration=1253 frames, utt_pad_proportion=0.05381, over 10443.95 utterances.], batch size: 42, lr: 3.87e-02, grad_scale: 8.0 2023-03-07 12:54:30,722 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.781e+02 4.301e+02 5.622e+02 7.152e+02 1.270e+03, threshold=1.124e+03, percent-clipped=3.0 2023-03-07 12:54:36,264 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6306.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 12:55:09,167 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6327.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:55:21,933 INFO [train2.py:809] (0/4) Epoch 2, batch 2350, loss[ctc_loss=0.2627, att_loss=0.3203, loss=0.3088, over 15760.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009028, over 38.00 utterances.], tot_loss[ctc_loss=0.2485, att_loss=0.3276, loss=0.3118, over 3264824.48 frames. utt_duration=1264 frames, utt_pad_proportion=0.05154, over 10342.97 utterances.], batch size: 38, lr: 3.86e-02, grad_scale: 8.0 2023-03-07 12:55:39,718 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6346.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:55:39,889 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2984, 2.8784, 3.2665, 2.9547, 3.0302, 2.8696, 2.1430, 3.3051], device='cuda:0'), covar=tensor([0.0830, 0.0290, 0.0618, 0.0509, 0.0332, 0.0868, 0.0979, 0.0149], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0041, 0.0081, 0.0067, 0.0048, 0.0087, 0.0076, 0.0037], device='cuda:0'), out_proj_covar=tensor([8.0530e-05, 6.2340e-05, 1.1447e-04, 8.2203e-05, 6.8950e-05, 1.1787e-04, 9.1674e-05, 5.2802e-05], device='cuda:0') 2023-03-07 12:55:52,504 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6354.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 12:56:38,672 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7183, 2.9938, 3.2298, 1.5888, 3.3727, 4.2069, 3.7535, 2.3069], device='cuda:0'), covar=tensor([0.0522, 0.0744, 0.0924, 0.2048, 0.0822, 0.0127, 0.0438, 0.1843], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0100, 0.0098, 0.0123, 0.0126, 0.0061, 0.0074, 0.0128], device='cuda:0'), out_proj_covar=tensor([9.7250e-05, 9.4397e-05, 1.1179e-04, 1.1505e-04, 1.2315e-04, 6.3268e-05, 8.7529e-05, 1.2153e-04], device='cuda:0') 2023-03-07 12:56:41,288 INFO [train2.py:809] (0/4) Epoch 2, batch 2400, loss[ctc_loss=0.2514, att_loss=0.338, loss=0.3207, over 16543.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006107, over 45.00 utterances.], tot_loss[ctc_loss=0.2494, att_loss=0.3285, loss=0.3127, over 3263638.28 frames. utt_duration=1241 frames, utt_pad_proportion=0.05896, over 10532.47 utterances.], batch size: 45, lr: 3.85e-02, grad_scale: 8.0 2023-03-07 12:57:03,925 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6399.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:57:10,896 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.496e+02 4.794e+02 5.680e+02 6.939e+02 1.517e+03, threshold=1.136e+03, percent-clipped=5.0 2023-03-07 12:57:24,330 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6411.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:58:03,299 INFO [train2.py:809] (0/4) Epoch 2, batch 2450, loss[ctc_loss=0.2103, att_loss=0.2973, loss=0.2799, over 15354.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.0119, over 35.00 utterances.], tot_loss[ctc_loss=0.2488, att_loss=0.3279, loss=0.3121, over 3265420.08 frames. utt_duration=1225 frames, utt_pad_proportion=0.06226, over 10672.14 utterances.], batch size: 35, lr: 3.84e-02, grad_scale: 8.0 2023-03-07 12:58:42,229 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6459.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:58:47,114 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6462.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:58:54,648 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6467.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:59:23,398 INFO [train2.py:809] (0/4) Epoch 2, batch 2500, loss[ctc_loss=0.2887, att_loss=0.3569, loss=0.3432, over 17045.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009789, over 53.00 utterances.], tot_loss[ctc_loss=0.2484, att_loss=0.3285, loss=0.3125, over 3272884.55 frames. utt_duration=1231 frames, utt_pad_proportion=0.05894, over 10644.48 utterances.], batch size: 53, lr: 3.83e-02, grad_scale: 8.0 2023-03-07 12:59:33,177 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6491.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 12:59:51,987 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.651e+02 4.550e+02 5.536e+02 6.688e+02 2.770e+03, threshold=1.107e+03, percent-clipped=3.0 2023-03-07 13:00:09,859 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6514.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:00:24,411 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6523.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 13:00:29,522 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6526.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:00:32,863 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6528.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:00:36,690 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-07 13:00:44,096 INFO [train2.py:809] (0/4) Epoch 2, batch 2550, loss[ctc_loss=0.2202, att_loss=0.309, loss=0.2913, over 16180.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.007021, over 41.00 utterances.], tot_loss[ctc_loss=0.2482, att_loss=0.328, loss=0.312, over 3275316.59 frames. utt_duration=1229 frames, utt_pad_proportion=0.05872, over 10671.66 utterances.], batch size: 41, lr: 3.82e-02, grad_scale: 8.0 2023-03-07 13:01:15,296 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.13 vs. limit=2.0 2023-03-07 13:01:23,814 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6559.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:01:28,553 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6562.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:01:28,873 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.1286, 1.3981, 1.6017, 1.3663, 1.1626, 2.2058, 1.5089, 2.1594], device='cuda:0'), covar=tensor([0.0986, 0.0953, 0.0762, 0.0807, 0.0784, 0.0684, 0.0987, 0.0480], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0071, 0.0067, 0.0068, 0.0069, 0.0085, 0.0072, 0.0095], device='cuda:0'), out_proj_covar=tensor([5.1579e-05, 5.3265e-05, 5.3258e-05, 5.5384e-05, 4.3438e-05, 6.9513e-05, 5.7250e-05, 5.0345e-05], device='cuda:0') 2023-03-07 13:01:36,559 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6567.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:01:43,016 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4300, 3.7790, 3.2180, 3.2594, 3.3890, 3.4772, 2.0449, 3.8579], device='cuda:0'), covar=tensor([0.0723, 0.0204, 0.0623, 0.0467, 0.0262, 0.0592, 0.0924, 0.0094], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0044, 0.0083, 0.0071, 0.0050, 0.0087, 0.0078, 0.0037], device='cuda:0'), out_proj_covar=tensor([8.6096e-05, 6.7710e-05, 1.1767e-04, 8.9134e-05, 7.3462e-05, 1.1974e-04, 9.5777e-05, 5.6296e-05], device='cuda:0') 2023-03-07 13:02:05,598 INFO [train2.py:809] (0/4) Epoch 2, batch 2600, loss[ctc_loss=0.2277, att_loss=0.3386, loss=0.3164, over 17285.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01281, over 55.00 utterances.], tot_loss[ctc_loss=0.2462, att_loss=0.3269, loss=0.3107, over 3269253.36 frames. utt_duration=1211 frames, utt_pad_proportion=0.0638, over 10811.34 utterances.], batch size: 55, lr: 3.81e-02, grad_scale: 8.0 2023-03-07 13:02:35,426 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.190e+02 4.654e+02 5.537e+02 6.841e+02 1.492e+03, threshold=1.107e+03, percent-clipped=3.0 2023-03-07 13:03:05,288 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6622.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:03:16,798 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6629.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:03:26,460 INFO [train2.py:809] (0/4) Epoch 2, batch 2650, loss[ctc_loss=0.3027, att_loss=0.3663, loss=0.3536, over 14100.00 frames. utt_duration=387.7 frames, utt_pad_proportion=0.3245, over 146.00 utterances.], tot_loss[ctc_loss=0.245, att_loss=0.3265, loss=0.3102, over 3268517.65 frames. utt_duration=1182 frames, utt_pad_proportion=0.07208, over 11079.39 utterances.], batch size: 146, lr: 3.80e-02, grad_scale: 8.0 2023-03-07 13:03:44,194 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6646.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:04:45,710 INFO [train2.py:809] (0/4) Epoch 2, batch 2700, loss[ctc_loss=0.2028, att_loss=0.2925, loss=0.2746, over 15744.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.01024, over 38.00 utterances.], tot_loss[ctc_loss=0.2448, att_loss=0.3255, loss=0.3093, over 3259826.27 frames. utt_duration=1193 frames, utt_pad_proportion=0.07201, over 10941.60 utterances.], batch size: 38, lr: 3.79e-02, grad_scale: 8.0 2023-03-07 13:04:54,021 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6690.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:05:00,042 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6694.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:05:08,571 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6699.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:05:15,303 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.955e+02 4.204e+02 5.612e+02 7.162e+02 1.326e+03, threshold=1.122e+03, percent-clipped=2.0 2023-03-07 13:06:06,238 INFO [train2.py:809] (0/4) Epoch 2, batch 2750, loss[ctc_loss=0.2646, att_loss=0.304, loss=0.2961, over 15492.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.009313, over 36.00 utterances.], tot_loss[ctc_loss=0.2443, att_loss=0.325, loss=0.3089, over 3254271.26 frames. utt_duration=1186 frames, utt_pad_proportion=0.07576, over 10986.35 utterances.], batch size: 36, lr: 3.79e-02, grad_scale: 8.0 2023-03-07 13:06:26,249 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6747.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:06:30,197 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-03-07 13:07:25,950 INFO [train2.py:809] (0/4) Epoch 2, batch 2800, loss[ctc_loss=0.2272, att_loss=0.3059, loss=0.2902, over 16175.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006688, over 41.00 utterances.], tot_loss[ctc_loss=0.2438, att_loss=0.3248, loss=0.3086, over 3260798.76 frames. utt_duration=1199 frames, utt_pad_proportion=0.07116, over 10889.79 utterances.], batch size: 41, lr: 3.78e-02, grad_scale: 8.0 2023-03-07 13:07:36,261 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6791.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:07:55,847 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.567e+02 4.477e+02 5.364e+02 6.587e+02 1.505e+03, threshold=1.073e+03, percent-clipped=3.0 2023-03-07 13:08:01,996 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.75 vs. limit=2.0 2023-03-07 13:08:20,295 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6818.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 13:08:21,915 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9771, 4.2106, 4.7390, 4.4722, 4.7106, 3.7763, 4.1161, 1.9271], device='cuda:0'), covar=tensor([0.0087, 0.0152, 0.0223, 0.0166, 0.0088, 0.0199, 0.0190, 0.1273], device='cuda:0'), in_proj_covar=tensor([0.0031, 0.0028, 0.0029, 0.0040, 0.0031, 0.0034, 0.0039, 0.0070], device='cuda:0'), out_proj_covar=tensor([6.0257e-05, 6.4051e-05, 7.4789e-05, 7.6823e-05, 6.1905e-05, 8.3456e-05, 7.5855e-05, 1.3110e-04], device='cuda:0') 2023-03-07 13:08:28,459 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6823.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:08:33,303 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6826.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:08:46,712 INFO [train2.py:809] (0/4) Epoch 2, batch 2850, loss[ctc_loss=0.1814, att_loss=0.2813, loss=0.2613, over 15862.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.00942, over 39.00 utterances.], tot_loss[ctc_loss=0.242, att_loss=0.3244, loss=0.3079, over 3274232.71 frames. utt_duration=1215 frames, utt_pad_proportion=0.06316, over 10790.41 utterances.], batch size: 39, lr: 3.77e-02, grad_scale: 8.0 2023-03-07 13:08:53,145 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6839.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:09:22,104 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1221, 5.3423, 4.6932, 5.3451, 4.9568, 4.7476, 4.7473, 4.6649], device='cuda:0'), covar=tensor([0.0919, 0.0756, 0.0660, 0.0506, 0.0488, 0.1065, 0.1843, 0.1499], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0270, 0.0226, 0.0202, 0.0171, 0.0275, 0.0285, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 13:09:25,225 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6859.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:09:37,082 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7860, 3.7623, 4.0563, 3.8686, 2.3787, 2.7957, 3.8948, 3.4094], device='cuda:0'), covar=tensor([0.0785, 0.0276, 0.0180, 0.0465, 0.8639, 0.1952, 0.0297, 0.2499], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0110, 0.0135, 0.0153, 0.0375, 0.0244, 0.0136, 0.0179], device='cuda:0'), out_proj_covar=tensor([1.0267e-04, 5.7019e-05, 6.3157e-05, 7.3561e-05, 1.8188e-04, 1.1962e-04, 6.5229e-05, 1.0012e-04], device='cuda:0') 2023-03-07 13:09:38,375 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6867.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:09:49,559 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6874.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:09:56,254 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6878.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 13:10:06,817 INFO [train2.py:809] (0/4) Epoch 2, batch 2900, loss[ctc_loss=0.2528, att_loss=0.3331, loss=0.317, over 16620.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005379, over 47.00 utterances.], tot_loss[ctc_loss=0.2418, att_loss=0.3245, loss=0.308, over 3276655.66 frames. utt_duration=1218 frames, utt_pad_proportion=0.06038, over 10770.10 utterances.], batch size: 47, lr: 3.76e-02, grad_scale: 8.0 2023-03-07 13:10:36,473 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.536e+02 4.377e+02 5.489e+02 7.218e+02 1.392e+03, threshold=1.098e+03, percent-clipped=5.0 2023-03-07 13:10:42,875 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6907.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:10:46,191 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=6909.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:10:55,825 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6915.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:11:07,564 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=6922.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:11:20,727 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-07 13:11:27,221 INFO [train2.py:809] (0/4) Epoch 2, batch 2950, loss[ctc_loss=0.1723, att_loss=0.2742, loss=0.2538, over 15509.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.0078, over 36.00 utterances.], tot_loss[ctc_loss=0.2394, att_loss=0.3235, loss=0.3066, over 3276227.57 frames. utt_duration=1216 frames, utt_pad_proportion=0.06206, over 10794.80 utterances.], batch size: 36, lr: 3.75e-02, grad_scale: 8.0 2023-03-07 13:11:35,186 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6939.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 13:12:00,341 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-07 13:12:12,954 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7857, 5.9005, 5.4195, 5.9087, 5.4191, 5.3807, 5.2196, 5.2951], device='cuda:0'), covar=tensor([0.0790, 0.0696, 0.0552, 0.0456, 0.0539, 0.0997, 0.2043, 0.1723], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0270, 0.0224, 0.0198, 0.0173, 0.0267, 0.0283, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 13:12:24,490 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:12:24,787 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=6970.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:12:47,618 INFO [train2.py:809] (0/4) Epoch 2, batch 3000, loss[ctc_loss=0.2119, att_loss=0.3135, loss=0.2932, over 17154.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01335, over 56.00 utterances.], tot_loss[ctc_loss=0.2384, att_loss=0.3229, loss=0.306, over 3277244.25 frames. utt_duration=1202 frames, utt_pad_proportion=0.06553, over 10917.32 utterances.], batch size: 56, lr: 3.74e-02, grad_scale: 8.0 2023-03-07 13:12:47,620 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 13:13:01,225 INFO [train2.py:843] (0/4) Epoch 2, validation: ctc_loss=0.1245, att_loss=0.2759, loss=0.2456, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 13:13:01,225 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 13:13:01,423 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=6985.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:13:30,571 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 3.232e+02 4.393e+02 5.286e+02 7.029e+02 1.971e+03, threshold=1.057e+03, percent-clipped=9.0 2023-03-07 13:13:50,317 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.79 vs. limit=5.0 2023-03-07 13:14:20,773 INFO [train2.py:809] (0/4) Epoch 2, batch 3050, loss[ctc_loss=0.2654, att_loss=0.3243, loss=0.3125, over 16011.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006908, over 40.00 utterances.], tot_loss[ctc_loss=0.2388, att_loss=0.323, loss=0.3062, over 3270686.00 frames. utt_duration=1216 frames, utt_pad_proportion=0.06519, over 10772.41 utterances.], batch size: 40, lr: 3.73e-02, grad_scale: 8.0 2023-03-07 13:14:27,773 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2023-03-07 13:14:42,941 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7048.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 13:15:41,471 INFO [train2.py:809] (0/4) Epoch 2, batch 3100, loss[ctc_loss=0.3075, att_loss=0.3452, loss=0.3377, over 17136.00 frames. utt_duration=693.9 frames, utt_pad_proportion=0.1293, over 99.00 utterances.], tot_loss[ctc_loss=0.2377, att_loss=0.3226, loss=0.3056, over 3273799.99 frames. utt_duration=1212 frames, utt_pad_proportion=0.06473, over 10819.13 utterances.], batch size: 99, lr: 3.72e-02, grad_scale: 8.0 2023-03-07 13:16:04,541 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3899, 5.6827, 5.1774, 5.7033, 5.2554, 5.1531, 5.0585, 5.1560], device='cuda:0'), covar=tensor([0.1122, 0.0864, 0.0752, 0.0534, 0.0549, 0.1238, 0.2274, 0.1864], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0272, 0.0230, 0.0203, 0.0176, 0.0275, 0.0291, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 13:16:10,552 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.443e+02 4.796e+02 5.430e+02 6.688e+02 1.495e+03, threshold=1.086e+03, percent-clipped=6.0 2023-03-07 13:16:20,401 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7109.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 13:16:34,983 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7118.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 13:16:42,314 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7123.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:17:01,197 INFO [train2.py:809] (0/4) Epoch 2, batch 3150, loss[ctc_loss=0.1902, att_loss=0.3073, loss=0.2839, over 16132.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005772, over 42.00 utterances.], tot_loss[ctc_loss=0.2367, att_loss=0.322, loss=0.3049, over 3271639.30 frames. utt_duration=1237 frames, utt_pad_proportion=0.0592, over 10594.28 utterances.], batch size: 42, lr: 3.71e-02, grad_scale: 8.0 2023-03-07 13:17:46,715 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7163.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:17:51,612 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7166.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:17:51,871 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.0704, 0.7614, 1.9606, 1.2734, 2.1709, 1.5697, 2.0896, 1.2476], device='cuda:0'), covar=tensor([0.1198, 0.1169, 0.0699, 0.1280, 0.0710, 0.3404, 0.1075, 0.1839], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0063, 0.0056, 0.0084, 0.0070, 0.0062, 0.0073, 0.0092], device='cuda:0'), out_proj_covar=tensor([4.8175e-05, 4.1567e-05, 4.0646e-05, 4.6693e-05, 4.1045e-05, 4.1502e-05, 4.3059e-05, 6.2036e-05], device='cuda:0') 2023-03-07 13:17:59,076 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7171.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:18:22,230 INFO [train2.py:809] (0/4) Epoch 2, batch 3200, loss[ctc_loss=0.1823, att_loss=0.2765, loss=0.2577, over 15780.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008756, over 38.00 utterances.], tot_loss[ctc_loss=0.2355, att_loss=0.3213, loss=0.3042, over 3274476.80 frames. utt_duration=1249 frames, utt_pad_proportion=0.05522, over 10503.40 utterances.], batch size: 38, lr: 3.71e-02, grad_scale: 8.0 2023-03-07 13:18:49,775 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3863, 4.9222, 4.9182, 5.0164, 1.9882, 2.8857, 5.1986, 3.8633], device='cuda:0'), covar=tensor([0.0706, 0.0214, 0.0196, 0.0353, 1.2148, 0.2797, 0.0131, 0.3001], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0114, 0.0133, 0.0152, 0.0370, 0.0248, 0.0130, 0.0188], device='cuda:0'), out_proj_covar=tensor([1.0685e-04, 6.0000e-05, 6.4646e-05, 7.4236e-05, 1.7925e-04, 1.2257e-04, 6.4050e-05, 1.0558e-04], device='cuda:0') 2023-03-07 13:18:50,783 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 4.512e+02 5.517e+02 7.088e+02 1.562e+03, threshold=1.103e+03, percent-clipped=2.0 2023-03-07 13:19:25,266 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7224.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 13:19:32,688 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2847, 5.4311, 4.9713, 5.5004, 5.1571, 5.0405, 5.0181, 4.8465], device='cuda:0'), covar=tensor([0.0959, 0.0854, 0.0759, 0.0615, 0.0603, 0.1016, 0.1794, 0.1743], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0274, 0.0230, 0.0201, 0.0183, 0.0284, 0.0289, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 13:19:41,623 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7234.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 13:19:42,924 INFO [train2.py:809] (0/4) Epoch 2, batch 3250, loss[ctc_loss=0.256, att_loss=0.3344, loss=0.3187, over 16958.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.00802, over 50.00 utterances.], tot_loss[ctc_loss=0.2367, att_loss=0.3215, loss=0.3045, over 3276738.19 frames. utt_duration=1235 frames, utt_pad_proportion=0.05683, over 10625.74 utterances.], batch size: 50, lr: 3.70e-02, grad_scale: 8.0 2023-03-07 13:19:43,720 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-07 13:20:29,912 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7265.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:20:45,957 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7275.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:21:02,604 INFO [train2.py:809] (0/4) Epoch 2, batch 3300, loss[ctc_loss=0.1783, att_loss=0.2734, loss=0.2543, over 15776.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008385, over 38.00 utterances.], tot_loss[ctc_loss=0.2362, att_loss=0.3217, loss=0.3046, over 3277754.15 frames. utt_duration=1240 frames, utt_pad_proportion=0.05645, over 10588.68 utterances.], batch size: 38, lr: 3.69e-02, grad_scale: 8.0 2023-03-07 13:21:02,939 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7285.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:21:30,734 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.392e+02 4.714e+02 6.021e+02 7.352e+02 2.285e+03, threshold=1.204e+03, percent-clipped=4.0 2023-03-07 13:22:02,636 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-07 13:22:20,315 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7333.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:22:23,592 INFO [train2.py:809] (0/4) Epoch 2, batch 3350, loss[ctc_loss=0.2328, att_loss=0.3258, loss=0.3072, over 16973.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007319, over 50.00 utterances.], tot_loss[ctc_loss=0.2359, att_loss=0.3215, loss=0.3044, over 3283744.71 frames. utt_duration=1234 frames, utt_pad_proportion=0.05632, over 10658.35 utterances.], batch size: 50, lr: 3.68e-02, grad_scale: 8.0 2023-03-07 13:22:25,544 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7336.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:23:43,566 INFO [train2.py:809] (0/4) Epoch 2, batch 3400, loss[ctc_loss=0.2291, att_loss=0.3274, loss=0.3078, over 16757.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006441, over 48.00 utterances.], tot_loss[ctc_loss=0.2348, att_loss=0.321, loss=0.3037, over 3285741.77 frames. utt_duration=1238 frames, utt_pad_proportion=0.05535, over 10632.67 utterances.], batch size: 48, lr: 3.67e-02, grad_scale: 8.0 2023-03-07 13:24:12,389 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.987e+02 4.124e+02 5.654e+02 7.446e+02 2.114e+03, threshold=1.131e+03, percent-clipped=4.0 2023-03-07 13:24:14,040 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7404.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 13:24:18,714 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7407.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:25:03,542 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-03-07 13:25:03,869 INFO [train2.py:809] (0/4) Epoch 2, batch 3450, loss[ctc_loss=0.231, att_loss=0.3158, loss=0.2989, over 16023.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006614, over 40.00 utterances.], tot_loss[ctc_loss=0.2348, att_loss=0.3209, loss=0.3037, over 3281034.17 frames. utt_duration=1255 frames, utt_pad_proportion=0.05245, over 10471.02 utterances.], batch size: 40, lr: 3.66e-02, grad_scale: 8.0 2023-03-07 13:25:56,126 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7468.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:26:22,844 INFO [train2.py:809] (0/4) Epoch 2, batch 3500, loss[ctc_loss=0.1984, att_loss=0.2837, loss=0.2666, over 15495.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008865, over 36.00 utterances.], tot_loss[ctc_loss=0.2341, att_loss=0.3203, loss=0.3031, over 3281154.65 frames. utt_duration=1251 frames, utt_pad_proportion=0.05369, over 10504.19 utterances.], batch size: 36, lr: 3.65e-02, grad_scale: 8.0 2023-03-07 13:26:50,319 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.396e+02 4.119e+02 4.861e+02 5.923e+02 1.212e+03, threshold=9.723e+02, percent-clipped=1.0 2023-03-07 13:26:55,704 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4951, 5.0857, 4.8474, 4.8225, 5.1589, 5.0459, 4.8720, 4.6587], device='cuda:0'), covar=tensor([0.1044, 0.0357, 0.0279, 0.0497, 0.0263, 0.0301, 0.0276, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0157, 0.0104, 0.0122, 0.0161, 0.0179, 0.0140, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 13:27:16,174 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7519.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 13:27:16,364 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4635, 4.6159, 4.4823, 4.4634, 5.1304, 4.5237, 4.4338, 2.3228], device='cuda:0'), covar=tensor([0.0258, 0.0332, 0.0215, 0.0329, 0.1124, 0.0184, 0.0347, 0.3767], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0101, 0.0096, 0.0099, 0.0173, 0.0114, 0.0084, 0.0227], device='cuda:0'), out_proj_covar=tensor([9.4157e-05, 7.1805e-05, 7.3247e-05, 7.3143e-05, 1.4709e-04, 8.1369e-05, 6.7305e-05, 1.6916e-04], device='cuda:0') 2023-03-07 13:27:39,851 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7534.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 13:27:41,015 INFO [train2.py:809] (0/4) Epoch 2, batch 3550, loss[ctc_loss=0.234, att_loss=0.3028, loss=0.289, over 15493.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.009456, over 36.00 utterances.], tot_loss[ctc_loss=0.2353, att_loss=0.3209, loss=0.3038, over 3279024.67 frames. utt_duration=1271 frames, utt_pad_proportion=0.04841, over 10334.17 utterances.], batch size: 36, lr: 3.65e-02, grad_scale: 8.0 2023-03-07 13:28:27,984 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7565.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:28:36,596 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-03-07 13:28:55,954 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7582.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 13:29:00,313 INFO [train2.py:809] (0/4) Epoch 2, batch 3600, loss[ctc_loss=0.2938, att_loss=0.3558, loss=0.3434, over 16891.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006333, over 49.00 utterances.], tot_loss[ctc_loss=0.2366, att_loss=0.3215, loss=0.3046, over 3278735.79 frames. utt_duration=1261 frames, utt_pad_proportion=0.05026, over 10409.83 utterances.], batch size: 49, lr: 3.64e-02, grad_scale: 8.0 2023-03-07 13:29:28,532 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.614e+02 5.143e+02 6.522e+02 9.491e+02 2.921e+03, threshold=1.304e+03, percent-clipped=20.0 2023-03-07 13:29:38,559 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6013, 4.8018, 5.0078, 5.3927, 4.7849, 5.4798, 5.0722, 5.5458], device='cuda:0'), covar=tensor([0.0497, 0.0616, 0.0609, 0.0422, 0.2079, 0.0713, 0.0455, 0.0553], device='cuda:0'), in_proj_covar=tensor([0.0299, 0.0228, 0.0212, 0.0238, 0.0386, 0.0207, 0.0182, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 13:29:44,794 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7613.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:29:51,840 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.1622, 1.1652, 2.3086, 2.5866, 3.1029, 1.8830, 1.8529, 1.5995], device='cuda:0'), covar=tensor([0.1069, 0.1095, 0.0306, 0.0319, 0.0220, 0.0612, 0.1006, 0.1438], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0058, 0.0049, 0.0069, 0.0055, 0.0051, 0.0067, 0.0082], device='cuda:0'), out_proj_covar=tensor([4.0697e-05, 3.6764e-05, 3.3003e-05, 3.9410e-05, 3.3522e-05, 3.5697e-05, 3.9179e-05, 5.4393e-05], device='cuda:0') 2023-03-07 13:29:59,670 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8141, 4.9982, 5.3725, 5.7157, 4.9806, 5.7494, 5.1622, 5.7971], device='cuda:0'), covar=tensor([0.0413, 0.0548, 0.0415, 0.0303, 0.1821, 0.0562, 0.0443, 0.0454], device='cuda:0'), in_proj_covar=tensor([0.0299, 0.0228, 0.0211, 0.0235, 0.0387, 0.0207, 0.0183, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 13:30:03,033 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7624.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:30:14,837 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7631.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:30:20,987 INFO [train2.py:809] (0/4) Epoch 2, batch 3650, loss[ctc_loss=0.1964, att_loss=0.2801, loss=0.2633, over 15653.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.007647, over 37.00 utterances.], tot_loss[ctc_loss=0.2367, att_loss=0.3219, loss=0.3049, over 3277571.42 frames. utt_duration=1239 frames, utt_pad_proportion=0.05646, over 10592.60 utterances.], batch size: 37, lr: 3.63e-02, grad_scale: 8.0 2023-03-07 13:31:41,707 INFO [train2.py:809] (0/4) Epoch 2, batch 3700, loss[ctc_loss=0.2845, att_loss=0.3529, loss=0.3392, over 17274.00 frames. utt_duration=699.4 frames, utt_pad_proportion=0.1236, over 99.00 utterances.], tot_loss[ctc_loss=0.236, att_loss=0.3209, loss=0.3039, over 3255387.51 frames. utt_duration=1217 frames, utt_pad_proportion=0.06753, over 10714.43 utterances.], batch size: 99, lr: 3.62e-02, grad_scale: 8.0 2023-03-07 13:31:42,157 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7685.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:32:10,450 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.948e+02 4.093e+02 5.003e+02 6.020e+02 1.181e+03, threshold=1.001e+03, percent-clipped=0.0 2023-03-07 13:32:12,236 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7704.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 13:33:01,946 INFO [train2.py:809] (0/4) Epoch 2, batch 3750, loss[ctc_loss=0.2123, att_loss=0.2925, loss=0.2765, over 15375.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.009926, over 35.00 utterances.], tot_loss[ctc_loss=0.2338, att_loss=0.3202, loss=0.3029, over 3269757.66 frames. utt_duration=1220 frames, utt_pad_proportion=0.06245, over 10736.83 utterances.], batch size: 35, lr: 3.61e-02, grad_scale: 8.0 2023-03-07 13:33:14,924 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7743.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:33:29,473 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7752.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 13:33:43,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8829, 5.0469, 5.3436, 5.7102, 5.1436, 5.8415, 5.2300, 5.8906], device='cuda:0'), covar=tensor([0.0455, 0.0519, 0.0424, 0.0411, 0.1877, 0.0500, 0.0372, 0.0422], device='cuda:0'), in_proj_covar=tensor([0.0308, 0.0233, 0.0214, 0.0238, 0.0398, 0.0216, 0.0182, 0.0257], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 13:33:46,948 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7763.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:34:09,681 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8894, 5.1574, 5.4848, 5.7012, 4.9310, 5.7920, 5.1333, 5.8959], device='cuda:0'), covar=tensor([0.0488, 0.0484, 0.0377, 0.0439, 0.2003, 0.0627, 0.0392, 0.0476], device='cuda:0'), in_proj_covar=tensor([0.0309, 0.0233, 0.0214, 0.0238, 0.0399, 0.0216, 0.0182, 0.0259], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 13:34:21,941 INFO [train2.py:809] (0/4) Epoch 2, batch 3800, loss[ctc_loss=0.1967, att_loss=0.2888, loss=0.2704, over 16409.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007102, over 44.00 utterances.], tot_loss[ctc_loss=0.2348, att_loss=0.3212, loss=0.3039, over 3275925.31 frames. utt_duration=1224 frames, utt_pad_proportion=0.05854, over 10716.25 utterances.], batch size: 44, lr: 3.60e-02, grad_scale: 8.0 2023-03-07 13:34:49,465 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2724, 4.6785, 4.8047, 4.8331, 4.2268, 4.6105, 5.0372, 4.7554], device='cuda:0'), covar=tensor([0.0327, 0.0205, 0.0244, 0.0138, 0.0334, 0.0180, 0.0216, 0.0135], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0101, 0.0114, 0.0079, 0.0115, 0.0089, 0.0105, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 13:34:49,594 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=7802.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:34:50,771 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.304e+02 4.201e+02 5.009e+02 6.863e+02 2.448e+03, threshold=1.002e+03, percent-clipped=6.0 2023-03-07 13:34:52,731 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7804.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:35:16,895 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7819.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 13:35:41,998 INFO [train2.py:809] (0/4) Epoch 2, batch 3850, loss[ctc_loss=0.2461, att_loss=0.3121, loss=0.2989, over 16127.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005925, over 42.00 utterances.], tot_loss[ctc_loss=0.2343, att_loss=0.3201, loss=0.3029, over 3265073.98 frames. utt_duration=1205 frames, utt_pad_proportion=0.06762, over 10851.93 utterances.], batch size: 42, lr: 3.60e-02, grad_scale: 8.0 2023-03-07 13:36:25,203 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=7863.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:36:31,177 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7867.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:36:43,379 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-07 13:36:53,863 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6859, 4.6030, 5.0735, 4.4284, 2.1680, 3.6255, 2.4716, 4.5851], device='cuda:0'), covar=tensor([0.0371, 0.0222, 0.0295, 0.0280, 0.3568, 0.0432, 0.1340, 0.0298], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0074, 0.0163, 0.0111, 0.0213, 0.0090, 0.0158, 0.0134], device='cuda:0'), out_proj_covar=tensor([7.5275e-05, 6.4902e-05, 1.2970e-04, 8.3026e-05, 1.5548e-04, 7.5521e-05, 1.2123e-04, 1.0352e-04], device='cuda:0') 2023-03-07 13:36:59,755 INFO [train2.py:809] (0/4) Epoch 2, batch 3900, loss[ctc_loss=0.1984, att_loss=0.2944, loss=0.2752, over 15780.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.008008, over 38.00 utterances.], tot_loss[ctc_loss=0.234, att_loss=0.32, loss=0.3028, over 3256561.48 frames. utt_duration=1203 frames, utt_pad_proportion=0.07118, over 10841.98 utterances.], batch size: 38, lr: 3.59e-02, grad_scale: 8.0 2023-03-07 13:37:27,828 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.554e+02 4.818e+02 5.738e+02 6.928e+02 1.942e+03, threshold=1.148e+03, percent-clipped=11.0 2023-03-07 13:38:11,523 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=7931.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:38:17,769 INFO [train2.py:809] (0/4) Epoch 2, batch 3950, loss[ctc_loss=0.252, att_loss=0.3291, loss=0.3137, over 16968.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007407, over 50.00 utterances.], tot_loss[ctc_loss=0.234, att_loss=0.3203, loss=0.303, over 3261200.76 frames. utt_duration=1213 frames, utt_pad_proportion=0.06736, over 10769.23 utterances.], batch size: 50, lr: 3.58e-02, grad_scale: 8.0 2023-03-07 13:39:09,537 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-2.pt 2023-03-07 13:39:31,703 INFO [train2.py:809] (0/4) Epoch 3, batch 0, loss[ctc_loss=0.2023, att_loss=0.2923, loss=0.2743, over 15526.00 frames. utt_duration=1727 frames, utt_pad_proportion=0.00714, over 36.00 utterances.], tot_loss[ctc_loss=0.2023, att_loss=0.2923, loss=0.2743, over 15526.00 frames. utt_duration=1727 frames, utt_pad_proportion=0.00714, over 36.00 utterances.], batch size: 36, lr: 3.40e-02, grad_scale: 8.0 2023-03-07 13:39:31,705 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 13:39:44,017 INFO [train2.py:843] (0/4) Epoch 3, validation: ctc_loss=0.122, att_loss=0.2748, loss=0.2442, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 13:39:44,018 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 13:40:00,311 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=7979.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:40:01,862 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=7980.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:40:34,066 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-8000.pt 2023-03-07 13:40:42,655 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.716e+02 4.111e+02 5.021e+02 6.373e+02 1.388e+03, threshold=1.004e+03, percent-clipped=3.0 2023-03-07 13:40:47,368 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6643, 5.1455, 5.4990, 5.6766, 4.2662, 5.5755, 4.9699, 5.3621], device='cuda:0'), covar=tensor([0.0818, 0.0731, 0.0528, 0.0672, 0.3644, 0.0995, 0.0710, 0.1089], device='cuda:0'), in_proj_covar=tensor([0.0314, 0.0235, 0.0208, 0.0237, 0.0393, 0.0210, 0.0185, 0.0257], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 13:40:55,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0644, 5.2022, 5.5547, 5.8523, 5.2259, 5.9864, 5.2147, 6.0044], device='cuda:0'), covar=tensor([0.0496, 0.0523, 0.0397, 0.0417, 0.2423, 0.0619, 0.0452, 0.0498], device='cuda:0'), in_proj_covar=tensor([0.0314, 0.0235, 0.0208, 0.0237, 0.0393, 0.0210, 0.0186, 0.0258], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 13:41:08,149 INFO [train2.py:809] (0/4) Epoch 3, batch 50, loss[ctc_loss=0.2427, att_loss=0.3192, loss=0.3039, over 16906.00 frames. utt_duration=684.5 frames, utt_pad_proportion=0.1401, over 99.00 utterances.], tot_loss[ctc_loss=0.221, att_loss=0.3117, loss=0.2936, over 733024.46 frames. utt_duration=1258 frames, utt_pad_proportion=0.05972, over 2333.53 utterances.], batch size: 99, lr: 3.39e-02, grad_scale: 16.0 2023-03-07 13:42:17,909 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8063.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:42:27,408 INFO [train2.py:809] (0/4) Epoch 3, batch 100, loss[ctc_loss=0.2192, att_loss=0.3103, loss=0.2921, over 16418.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006049, over 44.00 utterances.], tot_loss[ctc_loss=0.2228, att_loss=0.3138, loss=0.2956, over 1303866.13 frames. utt_duration=1293 frames, utt_pad_proportion=0.04165, over 4039.12 utterances.], batch size: 44, lr: 3.38e-02, grad_scale: 16.0 2023-03-07 13:42:43,268 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6430, 4.5297, 4.9357, 4.6142, 2.1325, 4.1919, 2.8062, 3.6851], device='cuda:0'), covar=tensor([0.0361, 0.0152, 0.0482, 0.0222, 0.4413, 0.0233, 0.1341, 0.0682], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0073, 0.0160, 0.0106, 0.0215, 0.0084, 0.0153, 0.0139], device='cuda:0'), out_proj_covar=tensor([7.4898e-05, 6.4952e-05, 1.2863e-04, 8.0620e-05, 1.5685e-04, 7.2275e-05, 1.1891e-04, 1.0706e-04], device='cuda:0') 2023-03-07 13:43:15,560 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8099.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:43:21,306 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.756e+02 4.455e+02 5.328e+02 6.355e+02 1.061e+03, threshold=1.066e+03, percent-clipped=2.0 2023-03-07 13:43:33,805 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8111.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:43:47,462 INFO [train2.py:809] (0/4) Epoch 3, batch 150, loss[ctc_loss=0.1944, att_loss=0.3002, loss=0.2791, over 16408.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007253, over 44.00 utterances.], tot_loss[ctc_loss=0.2208, att_loss=0.3122, loss=0.2939, over 1738280.52 frames. utt_duration=1298 frames, utt_pad_proportion=0.04135, over 5361.77 utterances.], batch size: 44, lr: 3.37e-02, grad_scale: 16.0 2023-03-07 13:44:02,636 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-07 13:44:23,372 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7343, 2.0484, 2.6537, 3.9219, 3.9923, 4.0971, 2.5994, 1.8242], device='cuda:0'), covar=tensor([0.0308, 0.2511, 0.1330, 0.0548, 0.0250, 0.0135, 0.1873, 0.2664], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0154, 0.0143, 0.0097, 0.0075, 0.0083, 0.0153, 0.0146], device='cuda:0'), out_proj_covar=tensor([9.0160e-05, 1.4209e-04, 1.3545e-04, 1.0966e-04, 7.6597e-05, 7.7178e-05, 1.4731e-04, 1.3438e-04], device='cuda:0') 2023-03-07 13:44:50,034 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8158.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:45:07,247 INFO [train2.py:809] (0/4) Epoch 3, batch 200, loss[ctc_loss=0.1821, att_loss=0.2736, loss=0.2553, over 15962.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006622, over 41.00 utterances.], tot_loss[ctc_loss=0.2211, att_loss=0.3124, loss=0.2941, over 2078117.23 frames. utt_duration=1273 frames, utt_pad_proportion=0.04778, over 6536.29 utterances.], batch size: 41, lr: 3.37e-02, grad_scale: 16.0 2023-03-07 13:46:02,073 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.226e+02 3.842e+02 5.004e+02 6.502e+02 2.003e+03, threshold=1.001e+03, percent-clipped=2.0 2023-03-07 13:46:24,745 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-03-07 13:46:28,355 INFO [train2.py:809] (0/4) Epoch 3, batch 250, loss[ctc_loss=0.2232, att_loss=0.3022, loss=0.2864, over 16283.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.00696, over 43.00 utterances.], tot_loss[ctc_loss=0.2203, att_loss=0.3129, loss=0.2944, over 2350506.26 frames. utt_duration=1278 frames, utt_pad_proportion=0.0424, over 7364.74 utterances.], batch size: 43, lr: 3.36e-02, grad_scale: 16.0 2023-03-07 13:47:40,010 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8265.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:47:46,793 INFO [train2.py:809] (0/4) Epoch 3, batch 300, loss[ctc_loss=0.2401, att_loss=0.3308, loss=0.3127, over 16483.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006278, over 46.00 utterances.], tot_loss[ctc_loss=0.2206, att_loss=0.3131, loss=0.2946, over 2557762.77 frames. utt_duration=1311 frames, utt_pad_proportion=0.03531, over 7811.23 utterances.], batch size: 46, lr: 3.35e-02, grad_scale: 16.0 2023-03-07 13:48:03,754 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8280.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:48:25,614 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-07 13:48:41,921 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.817e+02 4.123e+02 5.343e+02 7.099e+02 1.515e+03, threshold=1.069e+03, percent-clipped=5.0 2023-03-07 13:49:05,807 INFO [train2.py:809] (0/4) Epoch 3, batch 350, loss[ctc_loss=0.1996, att_loss=0.3153, loss=0.2921, over 16761.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006984, over 48.00 utterances.], tot_loss[ctc_loss=0.2212, att_loss=0.3132, loss=0.2948, over 2718064.04 frames. utt_duration=1275 frames, utt_pad_proportion=0.04434, over 8537.02 utterances.], batch size: 48, lr: 3.34e-02, grad_scale: 8.0 2023-03-07 13:49:16,785 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8326.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:49:19,533 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8328.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:50:11,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([0.8110, 1.0091, 1.6192, 1.1041, 1.9496, 1.5435, 2.3307, 1.7005], device='cuda:0'), covar=tensor([0.1336, 0.1296, 0.0729, 0.0751, 0.0468, 0.0549, 0.0377, 0.0810], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0072, 0.0067, 0.0065, 0.0069, 0.0074, 0.0074, 0.0087], device='cuda:0'), out_proj_covar=tensor([4.5562e-05, 5.4173e-05, 4.7130e-05, 4.2226e-05, 4.1038e-05, 4.6886e-05, 4.3225e-05, 4.3411e-05], device='cuda:0') 2023-03-07 13:50:25,020 INFO [train2.py:809] (0/4) Epoch 3, batch 400, loss[ctc_loss=0.2221, att_loss=0.3213, loss=0.3015, over 16464.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006825, over 46.00 utterances.], tot_loss[ctc_loss=0.2203, att_loss=0.3129, loss=0.2944, over 2843068.47 frames. utt_duration=1283 frames, utt_pad_proportion=0.04321, over 8875.40 utterances.], batch size: 46, lr: 3.34e-02, grad_scale: 8.0 2023-03-07 13:51:12,555 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8399.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:51:20,127 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.563e+02 4.039e+02 5.098e+02 6.528e+02 1.596e+03, threshold=1.020e+03, percent-clipped=2.0 2023-03-07 13:51:44,894 INFO [train2.py:809] (0/4) Epoch 3, batch 450, loss[ctc_loss=0.2532, att_loss=0.342, loss=0.3243, over 16316.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006825, over 45.00 utterances.], tot_loss[ctc_loss=0.2192, att_loss=0.3118, loss=0.2932, over 2931633.84 frames. utt_duration=1275 frames, utt_pad_proportion=0.04782, over 9209.17 utterances.], batch size: 45, lr: 3.33e-02, grad_scale: 8.0 2023-03-07 13:52:30,018 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8447.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:52:44,332 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6969, 1.9783, 3.2603, 4.1772, 4.4338, 4.2282, 2.7346, 1.6070], device='cuda:0'), covar=tensor([0.0410, 0.2792, 0.1042, 0.0480, 0.0166, 0.0189, 0.1921, 0.2837], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0153, 0.0144, 0.0096, 0.0073, 0.0085, 0.0152, 0.0145], device='cuda:0'), out_proj_covar=tensor([9.3440e-05, 1.4172e-04, 1.3737e-04, 1.0753e-04, 7.6655e-05, 7.8658e-05, 1.4791e-04, 1.3464e-04], device='cuda:0') 2023-03-07 13:52:47,923 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8458.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:53:05,347 INFO [train2.py:809] (0/4) Epoch 3, batch 500, loss[ctc_loss=0.2161, att_loss=0.3116, loss=0.2925, over 16338.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005563, over 45.00 utterances.], tot_loss[ctc_loss=0.2193, att_loss=0.312, loss=0.2935, over 3012370.18 frames. utt_duration=1280 frames, utt_pad_proportion=0.04582, over 9426.55 utterances.], batch size: 45, lr: 3.32e-02, grad_scale: 8.0 2023-03-07 13:53:07,196 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5854, 4.5107, 4.5065, 3.6189, 4.4652, 3.8682, 4.0608, 2.2860], device='cuda:0'), covar=tensor([0.0138, 0.0084, 0.0172, 0.0311, 0.0099, 0.0175, 0.0199, 0.1362], device='cuda:0'), in_proj_covar=tensor([0.0035, 0.0035, 0.0032, 0.0049, 0.0035, 0.0040, 0.0046, 0.0083], device='cuda:0'), out_proj_covar=tensor([8.1564e-05, 9.5177e-05, 9.4154e-05, 1.1384e-04, 8.5880e-05, 1.1459e-04, 1.0682e-04, 1.8003e-04], device='cuda:0') 2023-03-07 13:54:00,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.491e+02 4.509e+02 5.776e+02 7.013e+02 1.879e+03, threshold=1.155e+03, percent-clipped=5.0 2023-03-07 13:54:04,271 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8506.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:54:24,638 INFO [train2.py:809] (0/4) Epoch 3, batch 550, loss[ctc_loss=0.2769, att_loss=0.3447, loss=0.3311, over 14092.00 frames. utt_duration=387.7 frames, utt_pad_proportion=0.3245, over 146.00 utterances.], tot_loss[ctc_loss=0.2192, att_loss=0.3114, loss=0.293, over 3065107.31 frames. utt_duration=1269 frames, utt_pad_proportion=0.05157, over 9672.03 utterances.], batch size: 146, lr: 3.31e-02, grad_scale: 8.0 2023-03-07 13:55:07,695 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-03-07 13:55:37,794 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4493, 2.7613, 3.5246, 2.4531, 3.5724, 4.5278, 4.2775, 3.4077], device='cuda:0'), covar=tensor([0.0224, 0.1380, 0.0885, 0.1729, 0.0908, 0.0247, 0.0359, 0.1350], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0136, 0.0118, 0.0140, 0.0148, 0.0078, 0.0094, 0.0148], device='cuda:0'), out_proj_covar=tensor([1.3135e-04, 1.3808e-04, 1.4350e-04, 1.4421e-04, 1.5859e-04, 9.2081e-05, 1.1517e-04, 1.5138e-04], device='cuda:0') 2023-03-07 13:55:45,136 INFO [train2.py:809] (0/4) Epoch 3, batch 600, loss[ctc_loss=0.2536, att_loss=0.3502, loss=0.3309, over 17295.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01248, over 55.00 utterances.], tot_loss[ctc_loss=0.2196, att_loss=0.3119, loss=0.2934, over 3110221.94 frames. utt_duration=1240 frames, utt_pad_proportion=0.05806, over 10043.95 utterances.], batch size: 55, lr: 3.31e-02, grad_scale: 8.0 2023-03-07 13:56:42,021 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.080e+02 4.122e+02 4.877e+02 6.146e+02 9.874e+02, threshold=9.754e+02, percent-clipped=0.0 2023-03-07 13:57:05,727 INFO [train2.py:809] (0/4) Epoch 3, batch 650, loss[ctc_loss=0.2767, att_loss=0.3455, loss=0.3318, over 17040.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.008686, over 52.00 utterances.], tot_loss[ctc_loss=0.2191, att_loss=0.3111, loss=0.2927, over 3146604.56 frames. utt_duration=1230 frames, utt_pad_proportion=0.06008, over 10247.96 utterances.], batch size: 52, lr: 3.30e-02, grad_scale: 8.0 2023-03-07 13:57:08,999 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=8621.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 13:57:34,847 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7975, 4.4283, 4.4152, 4.6432, 4.7374, 4.2617, 4.0377, 2.0991], device='cuda:0'), covar=tensor([0.0155, 0.0290, 0.0219, 0.0087, 0.1057, 0.0276, 0.0343, 0.3488], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0099, 0.0094, 0.0100, 0.0182, 0.0119, 0.0087, 0.0229], device='cuda:0'), out_proj_covar=tensor([8.9485e-05, 7.3812e-05, 7.6365e-05, 7.9477e-05, 1.5993e-04, 9.0049e-05, 7.4610e-05, 1.7720e-04], device='cuda:0') 2023-03-07 13:58:21,884 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.83 vs. limit=2.0 2023-03-07 13:58:25,322 INFO [train2.py:809] (0/4) Epoch 3, batch 700, loss[ctc_loss=0.2402, att_loss=0.3075, loss=0.294, over 16189.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.005752, over 41.00 utterances.], tot_loss[ctc_loss=0.2201, att_loss=0.3118, loss=0.2934, over 3174591.36 frames. utt_duration=1229 frames, utt_pad_proportion=0.06087, over 10342.27 utterances.], batch size: 41, lr: 3.29e-02, grad_scale: 8.0 2023-03-07 13:58:30,104 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2023-03-07 13:59:13,411 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.35 vs. limit=2.0 2023-03-07 13:59:21,219 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.427e+02 4.229e+02 5.521e+02 7.710e+02 1.687e+03, threshold=1.104e+03, percent-clipped=9.0 2023-03-07 13:59:44,323 INFO [train2.py:809] (0/4) Epoch 3, batch 750, loss[ctc_loss=0.2133, att_loss=0.3204, loss=0.299, over 17035.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009797, over 52.00 utterances.], tot_loss[ctc_loss=0.2197, att_loss=0.3121, loss=0.2936, over 3201921.43 frames. utt_duration=1245 frames, utt_pad_proportion=0.05573, over 10297.20 utterances.], batch size: 52, lr: 3.29e-02, grad_scale: 8.0 2023-03-07 14:00:50,972 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8760.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:00:58,041 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.38 vs. limit=5.0 2023-03-07 14:01:04,404 INFO [train2.py:809] (0/4) Epoch 3, batch 800, loss[ctc_loss=0.1768, att_loss=0.2915, loss=0.2686, over 16690.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006379, over 46.00 utterances.], tot_loss[ctc_loss=0.2187, att_loss=0.3112, loss=0.2927, over 3223690.27 frames. utt_duration=1261 frames, utt_pad_proportion=0.04966, over 10233.97 utterances.], batch size: 46, lr: 3.28e-02, grad_scale: 8.0 2023-03-07 14:01:43,287 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0701, 5.1759, 5.6866, 5.8729, 5.2406, 6.0176, 5.1571, 6.0025], device='cuda:0'), covar=tensor([0.0435, 0.0465, 0.0295, 0.0363, 0.1624, 0.0546, 0.0329, 0.0518], device='cuda:0'), in_proj_covar=tensor([0.0343, 0.0255, 0.0225, 0.0265, 0.0424, 0.0229, 0.0187, 0.0272], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 14:02:01,730 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.534e+02 3.936e+02 4.764e+02 5.876e+02 1.316e+03, threshold=9.528e+02, percent-clipped=1.0 2023-03-07 14:02:24,970 INFO [train2.py:809] (0/4) Epoch 3, batch 850, loss[ctc_loss=0.1848, att_loss=0.2824, loss=0.2629, over 15386.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.01027, over 35.00 utterances.], tot_loss[ctc_loss=0.2182, att_loss=0.3104, loss=0.292, over 3228951.42 frames. utt_duration=1248 frames, utt_pad_proportion=0.05401, over 10365.62 utterances.], batch size: 35, lr: 3.27e-02, grad_scale: 8.0 2023-03-07 14:02:29,327 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8821.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:02:52,925 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-03-07 14:03:43,842 INFO [train2.py:809] (0/4) Epoch 3, batch 900, loss[ctc_loss=0.1872, att_loss=0.2632, loss=0.248, over 15641.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008423, over 37.00 utterances.], tot_loss[ctc_loss=0.2184, att_loss=0.311, loss=0.2925, over 3244431.41 frames. utt_duration=1264 frames, utt_pad_proportion=0.0497, over 10276.98 utterances.], batch size: 37, lr: 3.26e-02, grad_scale: 8.0 2023-03-07 14:04:39,898 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.583e+02 4.192e+02 5.104e+02 6.098e+02 1.636e+03, threshold=1.021e+03, percent-clipped=5.0 2023-03-07 14:04:54,101 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1840, 4.7582, 4.9258, 4.4670, 1.8152, 3.9259, 2.3017, 2.9018], device='cuda:0'), covar=tensor([0.0211, 0.0130, 0.0425, 0.0305, 0.4424, 0.0283, 0.1553, 0.1091], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0078, 0.0176, 0.0112, 0.0224, 0.0087, 0.0165, 0.0157], device='cuda:0'), out_proj_covar=tensor([7.9540e-05, 7.0128e-05, 1.4163e-04, 8.6091e-05, 1.6673e-04, 7.5243e-05, 1.3061e-04, 1.2401e-04], device='cuda:0') 2023-03-07 14:05:03,059 INFO [train2.py:809] (0/4) Epoch 3, batch 950, loss[ctc_loss=0.2021, att_loss=0.3046, loss=0.2841, over 16973.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007159, over 50.00 utterances.], tot_loss[ctc_loss=0.2168, att_loss=0.3098, loss=0.2912, over 3245409.96 frames. utt_duration=1285 frames, utt_pad_proportion=0.0467, over 10114.22 utterances.], batch size: 50, lr: 3.26e-02, grad_scale: 8.0 2023-03-07 14:05:07,371 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=8921.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:05:16,809 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8927.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:06:18,696 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8754, 2.2367, 3.2130, 4.3823, 4.2392, 4.3533, 2.7671, 1.9495], device='cuda:0'), covar=tensor([0.0345, 0.2370, 0.1243, 0.0401, 0.0129, 0.0162, 0.1944, 0.2574], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0149, 0.0143, 0.0097, 0.0073, 0.0086, 0.0154, 0.0140], device='cuda:0'), out_proj_covar=tensor([9.1789e-05, 1.3855e-04, 1.3804e-04, 1.0955e-04, 7.6212e-05, 8.2514e-05, 1.4990e-04, 1.3024e-04], device='cuda:0') 2023-03-07 14:06:23,047 INFO [train2.py:809] (0/4) Epoch 3, batch 1000, loss[ctc_loss=0.1776, att_loss=0.2803, loss=0.2598, over 16405.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006729, over 44.00 utterances.], tot_loss[ctc_loss=0.2149, att_loss=0.3087, loss=0.2899, over 3253598.25 frames. utt_duration=1287 frames, utt_pad_proportion=0.04562, over 10124.18 utterances.], batch size: 44, lr: 3.25e-02, grad_scale: 8.0 2023-03-07 14:06:23,156 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=8969.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:06:50,808 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=8986.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:06:53,978 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=8988.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:07:19,491 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.658e+02 4.100e+02 4.728e+02 5.740e+02 1.456e+03, threshold=9.457e+02, percent-clipped=2.0 2023-03-07 14:07:20,913 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-03-07 14:07:34,312 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9013.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:07:43,209 INFO [train2.py:809] (0/4) Epoch 3, batch 1050, loss[ctc_loss=0.1958, att_loss=0.2945, loss=0.2747, over 15961.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006715, over 41.00 utterances.], tot_loss[ctc_loss=0.2144, att_loss=0.3089, loss=0.29, over 3256658.96 frames. utt_duration=1285 frames, utt_pad_proportion=0.04665, over 10145.28 utterances.], batch size: 41, lr: 3.24e-02, grad_scale: 8.0 2023-03-07 14:07:58,319 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5864, 4.3005, 4.2986, 4.5108, 1.8810, 3.9810, 2.2037, 2.8037], device='cuda:0'), covar=tensor([0.0516, 0.0238, 0.0617, 0.0240, 0.4555, 0.0241, 0.1811, 0.1078], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0079, 0.0178, 0.0109, 0.0225, 0.0088, 0.0170, 0.0159], device='cuda:0'), out_proj_covar=tensor([8.1782e-05, 7.1952e-05, 1.4327e-04, 8.5375e-05, 1.6796e-04, 7.6126e-05, 1.3436e-04, 1.2594e-04], device='cuda:0') 2023-03-07 14:08:27,852 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9047.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:09:02,451 INFO [train2.py:809] (0/4) Epoch 3, batch 1100, loss[ctc_loss=0.209, att_loss=0.3198, loss=0.2977, over 16863.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.008486, over 49.00 utterances.], tot_loss[ctc_loss=0.2139, att_loss=0.3088, loss=0.2898, over 3258713.55 frames. utt_duration=1293 frames, utt_pad_proportion=0.04473, over 10093.60 utterances.], batch size: 49, lr: 3.24e-02, grad_scale: 8.0 2023-03-07 14:09:11,149 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9074.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:09:26,773 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4119, 4.8564, 4.7426, 4.9995, 4.4186, 4.6672, 5.1593, 4.9282], device='cuda:0'), covar=tensor([0.0336, 0.0261, 0.0367, 0.0139, 0.0337, 0.0205, 0.0251, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0112, 0.0130, 0.0085, 0.0126, 0.0098, 0.0112, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-03-07 14:09:36,011 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3741, 4.7745, 4.3761, 4.5775, 4.8150, 4.4470, 4.1992, 4.4560], device='cuda:0'), covar=tensor([0.0115, 0.0158, 0.0105, 0.0143, 0.0080, 0.0096, 0.0322, 0.0194], device='cuda:0'), in_proj_covar=tensor([0.0040, 0.0040, 0.0042, 0.0030, 0.0029, 0.0035, 0.0054, 0.0046], device='cuda:0'), out_proj_covar=tensor([9.3357e-05, 9.5883e-05, 1.1278e-04, 7.6773e-05, 6.8563e-05, 9.1478e-05, 1.2773e-04, 1.1633e-04], device='cuda:0') 2023-03-07 14:09:58,861 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.446e+02 3.857e+02 5.076e+02 6.311e+02 2.003e+03, threshold=1.015e+03, percent-clipped=7.0 2023-03-07 14:10:17,668 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9116.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:10:22,913 INFO [train2.py:809] (0/4) Epoch 3, batch 1150, loss[ctc_loss=0.2028, att_loss=0.3093, loss=0.288, over 16001.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.008251, over 40.00 utterances.], tot_loss[ctc_loss=0.2143, att_loss=0.309, loss=0.2901, over 3266344.54 frames. utt_duration=1273 frames, utt_pad_proportion=0.0484, over 10279.57 utterances.], batch size: 40, lr: 3.23e-02, grad_scale: 8.0 2023-03-07 14:10:23,221 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9203, 4.7722, 4.7757, 3.7022, 4.7775, 3.9109, 4.5164, 2.4780], device='cuda:0'), covar=tensor([0.0128, 0.0093, 0.0240, 0.0434, 0.0134, 0.0189, 0.0173, 0.1521], device='cuda:0'), in_proj_covar=tensor([0.0036, 0.0038, 0.0034, 0.0055, 0.0037, 0.0043, 0.0049, 0.0087], device='cuda:0'), out_proj_covar=tensor([8.7460e-05, 1.0774e-04, 1.0333e-04, 1.3397e-04, 9.7308e-05, 1.3016e-04, 1.2053e-04, 1.9768e-04], device='cuda:0') 2023-03-07 14:10:57,492 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-07 14:11:42,361 INFO [train2.py:809] (0/4) Epoch 3, batch 1200, loss[ctc_loss=0.303, att_loss=0.335, loss=0.3286, over 16268.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007904, over 43.00 utterances.], tot_loss[ctc_loss=0.2143, att_loss=0.3085, loss=0.2896, over 3267983.89 frames. utt_duration=1282 frames, utt_pad_proportion=0.04645, over 10210.49 utterances.], batch size: 43, lr: 3.22e-02, grad_scale: 8.0 2023-03-07 14:12:41,117 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.934e+02 4.184e+02 5.248e+02 6.761e+02 1.156e+03, threshold=1.050e+03, percent-clipped=4.0 2023-03-07 14:13:01,579 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7553, 4.7912, 4.6708, 4.7306, 5.0118, 4.6112, 4.0839, 2.1930], device='cuda:0'), covar=tensor([0.0177, 0.0284, 0.0228, 0.0212, 0.0887, 0.0194, 0.0417, 0.3777], device='cuda:0'), in_proj_covar=tensor([0.0114, 0.0103, 0.0099, 0.0104, 0.0191, 0.0121, 0.0090, 0.0236], device='cuda:0'), out_proj_covar=tensor([9.4387e-05, 7.9385e-05, 8.1708e-05, 8.4169e-05, 1.6931e-04, 9.4746e-05, 7.9067e-05, 1.8652e-04], device='cuda:0') 2023-03-07 14:13:06,017 INFO [train2.py:809] (0/4) Epoch 3, batch 1250, loss[ctc_loss=0.1807, att_loss=0.2827, loss=0.2623, over 15732.00 frames. utt_duration=1657 frames, utt_pad_proportion=0.009872, over 38.00 utterances.], tot_loss[ctc_loss=0.2155, att_loss=0.3092, loss=0.2904, over 3267188.15 frames. utt_duration=1284 frames, utt_pad_proportion=0.04723, over 10191.17 utterances.], batch size: 38, lr: 3.22e-02, grad_scale: 8.0 2023-03-07 14:14:30,174 INFO [train2.py:809] (0/4) Epoch 3, batch 1300, loss[ctc_loss=0.1974, att_loss=0.3166, loss=0.2927, over 17017.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.007831, over 51.00 utterances.], tot_loss[ctc_loss=0.2144, att_loss=0.3093, loss=0.2903, over 3277716.66 frames. utt_duration=1291 frames, utt_pad_proportion=0.04236, over 10166.74 utterances.], batch size: 51, lr: 3.21e-02, grad_scale: 8.0 2023-03-07 14:14:52,967 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9283.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:15:29,123 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.817e+02 3.943e+02 4.874e+02 6.147e+02 1.582e+03, threshold=9.748e+02, percent-clipped=2.0 2023-03-07 14:15:54,117 INFO [train2.py:809] (0/4) Epoch 3, batch 1350, loss[ctc_loss=0.2021, att_loss=0.3205, loss=0.2968, over 16771.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006333, over 48.00 utterances.], tot_loss[ctc_loss=0.2135, att_loss=0.3086, loss=0.2896, over 3275040.73 frames. utt_duration=1289 frames, utt_pad_proportion=0.04279, over 10176.39 utterances.], batch size: 48, lr: 3.20e-02, grad_scale: 8.0 2023-03-07 14:16:32,203 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9342.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:17:17,853 INFO [train2.py:809] (0/4) Epoch 3, batch 1400, loss[ctc_loss=0.1895, att_loss=0.2913, loss=0.2709, over 16784.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005504, over 48.00 utterances.], tot_loss[ctc_loss=0.2128, att_loss=0.3084, loss=0.2893, over 3277545.42 frames. utt_duration=1275 frames, utt_pad_proportion=0.04607, over 10292.49 utterances.], batch size: 48, lr: 3.20e-02, grad_scale: 8.0 2023-03-07 14:17:18,089 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9369.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:18:16,803 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.777e+02 3.704e+02 4.754e+02 6.457e+02 1.261e+03, threshold=9.507e+02, percent-clipped=3.0 2023-03-07 14:18:33,965 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9414.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:18:37,098 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9416.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:18:41,731 INFO [train2.py:809] (0/4) Epoch 3, batch 1450, loss[ctc_loss=0.1853, att_loss=0.3016, loss=0.2783, over 16416.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006724, over 44.00 utterances.], tot_loss[ctc_loss=0.211, att_loss=0.3072, loss=0.2879, over 3271364.85 frames. utt_duration=1284 frames, utt_pad_proportion=0.04371, over 10204.65 utterances.], batch size: 44, lr: 3.19e-02, grad_scale: 8.0 2023-03-07 14:18:53,621 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4177, 4.8926, 4.4698, 4.8529, 5.0224, 4.5402, 4.3899, 4.6866], device='cuda:0'), covar=tensor([0.0104, 0.0199, 0.0130, 0.0111, 0.0075, 0.0113, 0.0317, 0.0221], device='cuda:0'), in_proj_covar=tensor([0.0043, 0.0044, 0.0044, 0.0032, 0.0031, 0.0038, 0.0059, 0.0051], device='cuda:0'), out_proj_covar=tensor([1.0358e-04, 1.0809e-04, 1.2304e-04, 8.5311e-05, 7.4602e-05, 1.0021e-04, 1.4418e-04, 1.3101e-04], device='cuda:0') 2023-03-07 14:19:58,099 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9464.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:20:06,076 INFO [train2.py:809] (0/4) Epoch 3, batch 1500, loss[ctc_loss=0.1895, att_loss=0.2733, loss=0.2565, over 15503.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008402, over 36.00 utterances.], tot_loss[ctc_loss=0.2118, att_loss=0.3079, loss=0.2886, over 3271449.37 frames. utt_duration=1249 frames, utt_pad_proportion=0.05308, over 10486.76 utterances.], batch size: 36, lr: 3.18e-02, grad_scale: 8.0 2023-03-07 14:20:15,974 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9475.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:20:54,302 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2814, 2.9260, 3.5726, 2.3514, 3.3366, 4.3994, 4.1704, 3.3665], device='cuda:0'), covar=tensor([0.0405, 0.1247, 0.0735, 0.1566, 0.1093, 0.0246, 0.0476, 0.1114], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0151, 0.0129, 0.0148, 0.0164, 0.0091, 0.0105, 0.0156], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 14:21:03,452 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.770e+02 4.105e+02 5.004e+02 6.164e+02 1.572e+03, threshold=1.001e+03, percent-clipped=6.0 2023-03-07 14:21:18,652 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8565, 4.9671, 5.4136, 5.5745, 5.0021, 5.7860, 5.0085, 5.8491], device='cuda:0'), covar=tensor([0.0535, 0.0610, 0.0511, 0.0511, 0.2242, 0.0744, 0.0488, 0.0570], device='cuda:0'), in_proj_covar=tensor([0.0346, 0.0248, 0.0233, 0.0274, 0.0420, 0.0230, 0.0199, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 14:21:27,230 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-07 14:21:27,817 INFO [train2.py:809] (0/4) Epoch 3, batch 1550, loss[ctc_loss=0.2395, att_loss=0.3323, loss=0.3138, over 17016.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008006, over 51.00 utterances.], tot_loss[ctc_loss=0.2114, att_loss=0.3075, loss=0.2883, over 3271206.20 frames. utt_duration=1259 frames, utt_pad_proportion=0.05136, over 10401.90 utterances.], batch size: 51, lr: 3.18e-02, grad_scale: 8.0 2023-03-07 14:22:08,757 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7274, 5.1284, 4.9918, 4.9004, 5.3123, 5.0956, 4.9415, 4.7858], device='cuda:0'), covar=tensor([0.1080, 0.0353, 0.0204, 0.0402, 0.0219, 0.0250, 0.0215, 0.0266], device='cuda:0'), in_proj_covar=tensor([0.0277, 0.0176, 0.0118, 0.0140, 0.0186, 0.0200, 0.0160, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 14:22:36,635 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-03-07 14:22:50,219 INFO [train2.py:809] (0/4) Epoch 3, batch 1600, loss[ctc_loss=0.2159, att_loss=0.2826, loss=0.2693, over 15760.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.007629, over 38.00 utterances.], tot_loss[ctc_loss=0.2119, att_loss=0.3075, loss=0.2884, over 3272435.48 frames. utt_duration=1247 frames, utt_pad_proportion=0.05321, over 10511.06 utterances.], batch size: 38, lr: 3.17e-02, grad_scale: 8.0 2023-03-07 14:23:12,993 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9583.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:23:14,636 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9584.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:23:49,434 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.439e+02 4.014e+02 4.915e+02 6.528e+02 1.408e+03, threshold=9.829e+02, percent-clipped=8.0 2023-03-07 14:24:13,621 INFO [train2.py:809] (0/4) Epoch 3, batch 1650, loss[ctc_loss=0.2073, att_loss=0.3193, loss=0.2969, over 17382.00 frames. utt_duration=1105 frames, utt_pad_proportion=0.03471, over 63.00 utterances.], tot_loss[ctc_loss=0.2116, att_loss=0.3079, loss=0.2887, over 3273225.91 frames. utt_duration=1238 frames, utt_pad_proportion=0.05677, over 10584.94 utterances.], batch size: 63, lr: 3.16e-02, grad_scale: 8.0 2023-03-07 14:24:33,197 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9631.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:24:52,629 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9642.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:24:57,516 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9645.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:25:17,991 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9658.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:25:35,216 INFO [train2.py:809] (0/4) Epoch 3, batch 1700, loss[ctc_loss=0.2033, att_loss=0.3131, loss=0.2912, over 16393.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.006026, over 44.00 utterances.], tot_loss[ctc_loss=0.2107, att_loss=0.3073, loss=0.288, over 3276551.30 frames. utt_duration=1265 frames, utt_pad_proportion=0.04946, over 10371.48 utterances.], batch size: 44, lr: 3.16e-02, grad_scale: 8.0 2023-03-07 14:25:35,571 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=9669.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:25:52,078 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2023-03-07 14:26:08,865 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9690.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:26:33,375 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.300e+02 3.813e+02 4.618e+02 5.851e+02 1.256e+03, threshold=9.236e+02, percent-clipped=5.0 2023-03-07 14:26:54,228 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=9717.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:26:57,509 INFO [train2.py:809] (0/4) Epoch 3, batch 1750, loss[ctc_loss=0.3242, att_loss=0.3673, loss=0.3586, over 14277.00 frames. utt_duration=395.4 frames, utt_pad_proportion=0.3135, over 145.00 utterances.], tot_loss[ctc_loss=0.2105, att_loss=0.3071, loss=0.2878, over 3279193.29 frames. utt_duration=1256 frames, utt_pad_proportion=0.05028, over 10456.83 utterances.], batch size: 145, lr: 3.15e-02, grad_scale: 8.0 2023-03-07 14:26:57,937 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9719.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:27:37,610 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=9743.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:27:38,301 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2023-03-07 14:27:49,245 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7192, 2.8920, 4.9831, 3.8840, 3.2728, 4.2554, 4.6163, 4.6140], device='cuda:0'), covar=tensor([0.0105, 0.1445, 0.0213, 0.1179, 0.2521, 0.0520, 0.0281, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0211, 0.0116, 0.0266, 0.0309, 0.0159, 0.0098, 0.0110], device='cuda:0'), out_proj_covar=tensor([8.5051e-05, 1.5485e-04, 8.9361e-05, 2.0875e-04, 2.2432e-04, 1.2979e-04, 8.0533e-05, 9.1916e-05], device='cuda:0') 2023-03-07 14:27:58,409 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.67 vs. limit=2.0 2023-03-07 14:28:19,374 INFO [train2.py:809] (0/4) Epoch 3, batch 1800, loss[ctc_loss=0.1992, att_loss=0.3106, loss=0.2883, over 16776.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005859, over 48.00 utterances.], tot_loss[ctc_loss=0.2091, att_loss=0.3064, loss=0.287, over 3280837.55 frames. utt_duration=1257 frames, utt_pad_proportion=0.04963, over 10449.46 utterances.], batch size: 48, lr: 3.14e-02, grad_scale: 8.0 2023-03-07 14:28:21,088 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9770.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:29:17,431 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 4.029e+02 5.153e+02 6.488e+02 1.407e+03, threshold=1.031e+03, percent-clipped=9.0 2023-03-07 14:29:17,862 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=9804.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:29:41,462 INFO [train2.py:809] (0/4) Epoch 3, batch 1850, loss[ctc_loss=0.3147, att_loss=0.3704, loss=0.3593, over 14611.00 frames. utt_duration=404.5 frames, utt_pad_proportion=0.2965, over 145.00 utterances.], tot_loss[ctc_loss=0.2106, att_loss=0.3076, loss=0.2882, over 3278284.58 frames. utt_duration=1249 frames, utt_pad_proportion=0.05247, over 10515.91 utterances.], batch size: 145, lr: 3.14e-02, grad_scale: 8.0 2023-03-07 14:30:46,083 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6827, 2.1400, 2.3163, 3.9173, 4.0280, 4.0618, 2.6322, 1.7109], device='cuda:0'), covar=tensor([0.0252, 0.2696, 0.1554, 0.0490, 0.0123, 0.0118, 0.1716, 0.2869], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0167, 0.0157, 0.0109, 0.0082, 0.0089, 0.0162, 0.0151], device='cuda:0'), out_proj_covar=tensor([1.0365e-04, 1.5830e-04, 1.5347e-04, 1.2312e-04, 8.5876e-05, 8.5779e-05, 1.5913e-04, 1.4371e-04], device='cuda:0') 2023-03-07 14:31:02,970 INFO [train2.py:809] (0/4) Epoch 3, batch 1900, loss[ctc_loss=0.2214, att_loss=0.3119, loss=0.2938, over 17291.00 frames. utt_duration=1099 frames, utt_pad_proportion=0.03662, over 63.00 utterances.], tot_loss[ctc_loss=0.2092, att_loss=0.3059, loss=0.2866, over 3274074.79 frames. utt_duration=1270 frames, utt_pad_proportion=0.0489, over 10325.75 utterances.], batch size: 63, lr: 3.13e-02, grad_scale: 8.0 2023-03-07 14:31:42,847 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8063, 4.6432, 4.5861, 3.4594, 4.7053, 4.0639, 4.2166, 2.3372], device='cuda:0'), covar=tensor([0.0097, 0.0090, 0.0160, 0.0445, 0.0094, 0.0151, 0.0198, 0.1363], device='cuda:0'), in_proj_covar=tensor([0.0039, 0.0041, 0.0036, 0.0062, 0.0039, 0.0046, 0.0053, 0.0090], device='cuda:0'), out_proj_covar=tensor([9.8400e-05, 1.2045e-04, 1.1446e-04, 1.5825e-04, 1.0496e-04, 1.4727e-04, 1.3993e-04, 2.1698e-04], device='cuda:0') 2023-03-07 14:32:00,231 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.657e+02 4.209e+02 5.043e+02 6.488e+02 1.113e+03, threshold=1.009e+03, percent-clipped=3.0 2023-03-07 14:32:23,701 INFO [train2.py:809] (0/4) Epoch 3, batch 1950, loss[ctc_loss=0.1891, att_loss=0.2964, loss=0.275, over 16192.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.006101, over 41.00 utterances.], tot_loss[ctc_loss=0.2075, att_loss=0.3049, loss=0.2854, over 3275388.08 frames. utt_duration=1270 frames, utt_pad_proportion=0.04869, over 10325.97 utterances.], batch size: 41, lr: 3.13e-02, grad_scale: 8.0 2023-03-07 14:32:58,030 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=9940.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:33:25,919 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.9537, 2.1539, 1.6567, 1.9718, 1.1529, 3.3373, 1.5745, 1.3602], device='cuda:0'), covar=tensor([0.0651, 0.0853, 0.1136, 0.1538, 0.2270, 0.0276, 0.2333, 0.3492], device='cuda:0'), in_proj_covar=tensor([0.0060, 0.0050, 0.0050, 0.0068, 0.0058, 0.0052, 0.0061, 0.0082], device='cuda:0'), out_proj_covar=tensor([3.6361e-05, 3.3121e-05, 3.3180e-05, 4.1868e-05, 3.8581e-05, 2.8785e-05, 3.8394e-05, 5.3392e-05], device='cuda:0') 2023-03-07 14:33:44,090 INFO [train2.py:809] (0/4) Epoch 3, batch 2000, loss[ctc_loss=0.1809, att_loss=0.289, loss=0.2674, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007228, over 44.00 utterances.], tot_loss[ctc_loss=0.207, att_loss=0.3048, loss=0.2853, over 3279835.61 frames. utt_duration=1274 frames, utt_pad_proportion=0.04643, over 10307.79 utterances.], batch size: 44, lr: 3.12e-02, grad_scale: 8.0 2023-03-07 14:34:19,191 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2206, 0.6258, 1.5936, 2.2130, 1.0152, 2.0934, 1.3219, 1.8496], device='cuda:0'), covar=tensor([0.0150, 0.1338, 0.1336, 0.0546, 0.0905, 0.0543, 0.0835, 0.0540], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0070, 0.0075, 0.0069, 0.0070, 0.0068, 0.0073, 0.0076], device='cuda:0'), out_proj_covar=tensor([3.4359e-05, 4.6976e-05, 4.7213e-05, 3.9309e-05, 4.1111e-05, 3.9396e-05, 4.2572e-05, 4.0803e-05], device='cuda:0') 2023-03-07 14:34:35,658 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-03-07 14:34:36,539 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-10000.pt 2023-03-07 14:34:46,865 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.957e+02 4.346e+02 5.150e+02 6.489e+02 1.727e+03, threshold=1.030e+03, percent-clipped=5.0 2023-03-07 14:35:03,012 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10014.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:35:09,378 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2118, 2.6722, 3.3208, 2.7862, 3.3862, 4.4979, 4.2245, 3.3075], device='cuda:0'), covar=tensor([0.0339, 0.1180, 0.0886, 0.1155, 0.0799, 0.0133, 0.0374, 0.1077], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0157, 0.0133, 0.0153, 0.0169, 0.0098, 0.0113, 0.0159], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 14:35:10,649 INFO [train2.py:809] (0/4) Epoch 3, batch 2050, loss[ctc_loss=0.2152, att_loss=0.309, loss=0.2902, over 16486.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005318, over 46.00 utterances.], tot_loss[ctc_loss=0.2072, att_loss=0.3054, loss=0.2858, over 3286954.32 frames. utt_duration=1284 frames, utt_pad_proportion=0.04293, over 10250.22 utterances.], batch size: 46, lr: 3.11e-02, grad_scale: 8.0 2023-03-07 14:35:26,693 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5045, 4.9827, 4.6948, 4.6433, 5.0697, 4.9767, 4.7233, 4.5364], device='cuda:0'), covar=tensor([0.0904, 0.0318, 0.0242, 0.0476, 0.0270, 0.0216, 0.0260, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0291, 0.0184, 0.0122, 0.0147, 0.0194, 0.0210, 0.0166, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 14:35:44,002 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10039.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:36:21,337 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10062.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 14:36:27,563 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3506, 4.7493, 4.5931, 4.7742, 4.2697, 4.6174, 4.9947, 4.7306], device='cuda:0'), covar=tensor([0.0295, 0.0200, 0.0299, 0.0144, 0.0311, 0.0175, 0.0175, 0.0146], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0118, 0.0137, 0.0086, 0.0133, 0.0100, 0.0114, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-03-07 14:36:32,065 INFO [train2.py:809] (0/4) Epoch 3, batch 2100, loss[ctc_loss=0.1912, att_loss=0.286, loss=0.2671, over 15861.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01058, over 39.00 utterances.], tot_loss[ctc_loss=0.2076, att_loss=0.3059, loss=0.2862, over 3285295.74 frames. utt_duration=1289 frames, utt_pad_proportion=0.04187, over 10204.18 utterances.], batch size: 39, lr: 3.11e-02, grad_scale: 8.0 2023-03-07 14:36:33,984 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10070.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:37:21,239 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10099.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:37:22,933 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10100.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 14:37:28,709 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.690e+02 4.150e+02 4.955e+02 6.007e+02 1.097e+03, threshold=9.911e+02, percent-clipped=1.0 2023-03-07 14:37:50,888 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10118.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:37:52,384 INFO [train2.py:809] (0/4) Epoch 3, batch 2150, loss[ctc_loss=0.219, att_loss=0.3225, loss=0.3018, over 17133.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01419, over 56.00 utterances.], tot_loss[ctc_loss=0.2081, att_loss=0.3061, loss=0.2865, over 3280602.66 frames. utt_duration=1293 frames, utt_pad_proportion=0.04336, over 10158.56 utterances.], batch size: 56, lr: 3.10e-02, grad_scale: 8.0 2023-03-07 14:37:58,677 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10123.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 14:38:09,860 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7910, 5.8791, 5.3738, 5.9350, 5.6060, 5.4466, 5.4652, 5.3457], device='cuda:0'), covar=tensor([0.1336, 0.0970, 0.0671, 0.0630, 0.0490, 0.1229, 0.2485, 0.2150], device='cuda:0'), in_proj_covar=tensor([0.0266, 0.0308, 0.0248, 0.0239, 0.0215, 0.0311, 0.0336, 0.0317], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 14:38:20,389 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5089, 5.0852, 5.0319, 4.9165, 2.0595, 2.8379, 5.1063, 3.7794], device='cuda:0'), covar=tensor([0.0463, 0.0211, 0.0170, 0.0339, 0.8169, 0.2362, 0.0271, 0.2511], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0141, 0.0162, 0.0183, 0.0412, 0.0299, 0.0156, 0.0249], device='cuda:0'), out_proj_covar=tensor([1.3641e-04, 7.3976e-05, 8.6039e-05, 9.0359e-05, 2.0395e-04, 1.5019e-04, 7.9930e-05, 1.4047e-04], device='cuda:0') 2023-03-07 14:38:57,435 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.6353, 1.4396, 1.1293, 1.6745, 0.1345, 2.7277, 1.7752, 1.2907], device='cuda:0'), covar=tensor([0.0582, 0.1356, 0.1769, 0.1894, 0.2823, 0.0440, 0.1234, 0.3110], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0057, 0.0058, 0.0073, 0.0058, 0.0059, 0.0062, 0.0086], device='cuda:0'), out_proj_covar=tensor([3.7297e-05, 3.6097e-05, 3.7575e-05, 4.4351e-05, 3.9031e-05, 3.1652e-05, 3.8725e-05, 5.6325e-05], device='cuda:0') 2023-03-07 14:39:12,157 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-03-07 14:39:12,831 INFO [train2.py:809] (0/4) Epoch 3, batch 2200, loss[ctc_loss=0.2002, att_loss=0.3215, loss=0.2973, over 17326.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.0105, over 55.00 utterances.], tot_loss[ctc_loss=0.2091, att_loss=0.3065, loss=0.287, over 3281075.67 frames. utt_duration=1299 frames, utt_pad_proportion=0.04148, over 10112.88 utterances.], batch size: 55, lr: 3.09e-02, grad_scale: 8.0 2023-03-07 14:39:17,976 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0593, 1.6573, 1.6194, 1.5221, 1.1156, 1.4937, 1.5911, 2.2670], device='cuda:0'), covar=tensor([0.0186, 0.0714, 0.0695, 0.0819, 0.1127, 0.0779, 0.0572, 0.0353], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0071, 0.0073, 0.0072, 0.0070, 0.0070, 0.0073, 0.0079], device='cuda:0'), out_proj_covar=tensor([3.5837e-05, 4.6378e-05, 4.5011e-05, 4.0693e-05, 4.1386e-05, 4.1512e-05, 4.1368e-05, 4.1463e-05], device='cuda:0') 2023-03-07 14:40:09,927 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.374e+02 4.147e+02 5.296e+02 6.795e+02 2.410e+03, threshold=1.059e+03, percent-clipped=6.0 2023-03-07 14:40:10,383 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5517, 4.5024, 4.6300, 4.2901, 1.6524, 4.6080, 2.3511, 2.5514], device='cuda:0'), covar=tensor([0.0151, 0.0163, 0.0498, 0.0375, 0.4385, 0.0188, 0.1653, 0.1387], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0085, 0.0193, 0.0113, 0.0226, 0.0090, 0.0179, 0.0166], device='cuda:0'), out_proj_covar=tensor([8.1755e-05, 7.8451e-05, 1.5914e-04, 8.8669e-05, 1.7390e-04, 7.9743e-05, 1.4235e-04, 1.3368e-04], device='cuda:0') 2023-03-07 14:40:33,279 INFO [train2.py:809] (0/4) Epoch 3, batch 2250, loss[ctc_loss=0.1993, att_loss=0.2937, loss=0.2748, over 15895.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.007377, over 39.00 utterances.], tot_loss[ctc_loss=0.2088, att_loss=0.3062, loss=0.2868, over 3276770.47 frames. utt_duration=1263 frames, utt_pad_proportion=0.05195, over 10389.72 utterances.], batch size: 39, lr: 3.09e-02, grad_scale: 8.0 2023-03-07 14:41:07,768 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10240.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:41:52,772 INFO [train2.py:809] (0/4) Epoch 3, batch 2300, loss[ctc_loss=0.1699, att_loss=0.2792, loss=0.2573, over 15992.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008225, over 40.00 utterances.], tot_loss[ctc_loss=0.207, att_loss=0.305, loss=0.2854, over 3274482.87 frames. utt_duration=1272 frames, utt_pad_proportion=0.04969, over 10306.19 utterances.], batch size: 40, lr: 3.08e-02, grad_scale: 8.0 2023-03-07 14:42:15,042 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-03-07 14:42:23,456 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10288.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:42:49,025 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 3.999e+02 4.937e+02 6.652e+02 1.498e+03, threshold=9.874e+02, percent-clipped=4.0 2023-03-07 14:43:04,881 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10314.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:43:12,499 INFO [train2.py:809] (0/4) Epoch 3, batch 2350, loss[ctc_loss=0.2212, att_loss=0.3255, loss=0.3047, over 17113.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.0153, over 56.00 utterances.], tot_loss[ctc_loss=0.2078, att_loss=0.3062, loss=0.2866, over 3279773.77 frames. utt_duration=1262 frames, utt_pad_proportion=0.0506, over 10406.60 utterances.], batch size: 56, lr: 3.08e-02, grad_scale: 16.0 2023-03-07 14:43:32,642 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1943, 5.1154, 5.0699, 4.6158, 1.9108, 2.6871, 5.2535, 3.8519], device='cuda:0'), covar=tensor([0.0621, 0.0164, 0.0147, 0.0424, 0.8029, 0.2460, 0.0151, 0.2196], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0139, 0.0162, 0.0188, 0.0406, 0.0300, 0.0154, 0.0255], device='cuda:0'), out_proj_covar=tensor([1.3933e-04, 7.4081e-05, 8.6408e-05, 9.2309e-05, 2.0131e-04, 1.5022e-04, 7.9701e-05, 1.4318e-04], device='cuda:0') 2023-03-07 14:44:21,301 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10362.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:44:32,345 INFO [train2.py:809] (0/4) Epoch 3, batch 2400, loss[ctc_loss=0.2498, att_loss=0.3485, loss=0.3288, over 17398.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03389, over 63.00 utterances.], tot_loss[ctc_loss=0.2094, att_loss=0.3073, loss=0.2877, over 3275612.38 frames. utt_duration=1231 frames, utt_pad_proportion=0.05934, over 10656.57 utterances.], batch size: 63, lr: 3.07e-02, grad_scale: 16.0 2023-03-07 14:45:05,443 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2734, 3.6580, 3.6854, 3.6946, 1.9529, 3.7424, 2.5462, 2.6640], device='cuda:0'), covar=tensor([0.0409, 0.0240, 0.0685, 0.0301, 0.3901, 0.0196, 0.1481, 0.1158], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0085, 0.0198, 0.0109, 0.0225, 0.0090, 0.0186, 0.0172], device='cuda:0'), out_proj_covar=tensor([8.3487e-05, 7.8150e-05, 1.6297e-04, 8.6553e-05, 1.7511e-04, 8.0474e-05, 1.4791e-04, 1.3879e-04], device='cuda:0') 2023-03-07 14:45:14,385 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10395.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 14:45:20,846 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10399.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:45:28,357 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.513e+02 3.987e+02 4.935e+02 6.057e+02 1.238e+03, threshold=9.870e+02, percent-clipped=1.0 2023-03-07 14:45:36,568 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10409.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:45:41,428 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2504, 4.1966, 3.9221, 4.3664, 4.3406, 4.0816, 3.5203, 2.2975], device='cuda:0'), covar=tensor([0.0263, 0.0450, 0.0368, 0.0141, 0.1066, 0.0286, 0.0671, 0.3319], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0114, 0.0107, 0.0108, 0.0211, 0.0125, 0.0099, 0.0247], device='cuda:0'), out_proj_covar=tensor([1.0753e-04, 9.1432e-05, 9.0597e-05, 9.4018e-05, 1.9157e-04, 1.0146e-04, 8.8628e-05, 2.0298e-04], device='cuda:0') 2023-03-07 14:45:50,451 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10418.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 14:45:51,946 INFO [train2.py:809] (0/4) Epoch 3, batch 2450, loss[ctc_loss=0.1861, att_loss=0.2946, loss=0.2729, over 16129.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005941, over 42.00 utterances.], tot_loss[ctc_loss=0.2085, att_loss=0.3068, loss=0.2871, over 3277686.19 frames. utt_duration=1219 frames, utt_pad_proportion=0.06165, over 10766.69 utterances.], batch size: 42, lr: 3.06e-02, grad_scale: 16.0 2023-03-07 14:45:55,051 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7284, 5.8375, 5.2743, 5.8917, 5.5108, 5.3448, 5.2776, 5.1627], device='cuda:0'), covar=tensor([0.0909, 0.0796, 0.0618, 0.0526, 0.0599, 0.0941, 0.1969, 0.2070], device='cuda:0'), in_proj_covar=tensor([0.0264, 0.0307, 0.0248, 0.0235, 0.0215, 0.0305, 0.0331, 0.0322], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 14:46:38,227 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10447.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:46:49,263 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10454.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:47:12,262 INFO [train2.py:809] (0/4) Epoch 3, batch 2500, loss[ctc_loss=0.1656, att_loss=0.2776, loss=0.2552, over 15956.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006284, over 41.00 utterances.], tot_loss[ctc_loss=0.2067, att_loss=0.3056, loss=0.2858, over 3277291.50 frames. utt_duration=1232 frames, utt_pad_proportion=0.05864, over 10656.33 utterances.], batch size: 41, lr: 3.06e-02, grad_scale: 16.0 2023-03-07 14:47:14,150 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10470.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:48:09,698 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.331e+02 4.171e+02 5.202e+02 6.320e+02 1.110e+03, threshold=1.040e+03, percent-clipped=6.0 2023-03-07 14:48:27,666 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10515.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 14:48:33,424 INFO [train2.py:809] (0/4) Epoch 3, batch 2550, loss[ctc_loss=0.2686, att_loss=0.3368, loss=0.3232, over 17124.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01377, over 56.00 utterances.], tot_loss[ctc_loss=0.206, att_loss=0.3054, loss=0.2855, over 3286423.20 frames. utt_duration=1251 frames, utt_pad_proportion=0.0513, over 10524.06 utterances.], batch size: 56, lr: 3.05e-02, grad_scale: 16.0 2023-03-07 14:49:35,928 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2014, 5.0962, 5.0668, 4.4188, 1.8673, 2.8337, 5.1262, 3.7449], device='cuda:0'), covar=tensor([0.0579, 0.0162, 0.0127, 0.0516, 0.8106, 0.2237, 0.0168, 0.2105], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0141, 0.0161, 0.0185, 0.0403, 0.0296, 0.0149, 0.0251], device='cuda:0'), out_proj_covar=tensor([1.3676e-04, 7.4144e-05, 8.6504e-05, 9.1330e-05, 1.9989e-04, 1.4907e-04, 7.7327e-05, 1.4061e-04], device='cuda:0') 2023-03-07 14:49:53,598 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-03-07 14:49:54,160 INFO [train2.py:809] (0/4) Epoch 3, batch 2600, loss[ctc_loss=0.1824, att_loss=0.2881, loss=0.2669, over 16697.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005992, over 46.00 utterances.], tot_loss[ctc_loss=0.2047, att_loss=0.3047, loss=0.2847, over 3291654.63 frames. utt_duration=1255 frames, utt_pad_proportion=0.04891, over 10501.39 utterances.], batch size: 46, lr: 3.05e-02, grad_scale: 16.0 2023-03-07 14:50:50,041 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.348e+02 3.926e+02 4.748e+02 6.116e+02 1.074e+03, threshold=9.496e+02, percent-clipped=2.0 2023-03-07 14:51:13,367 INFO [train2.py:809] (0/4) Epoch 3, batch 2650, loss[ctc_loss=0.1917, att_loss=0.3013, loss=0.2794, over 16326.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006369, over 45.00 utterances.], tot_loss[ctc_loss=0.206, att_loss=0.3048, loss=0.2851, over 3274489.86 frames. utt_duration=1207 frames, utt_pad_proportion=0.06559, over 10864.06 utterances.], batch size: 45, lr: 3.04e-02, grad_scale: 16.0 2023-03-07 14:51:54,025 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4162, 5.1524, 5.1929, 5.0363, 2.0304, 2.8717, 5.1893, 3.8466], device='cuda:0'), covar=tensor([0.0466, 0.0177, 0.0137, 0.0295, 0.8667, 0.2274, 0.0162, 0.2236], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0142, 0.0161, 0.0186, 0.0411, 0.0298, 0.0148, 0.0255], device='cuda:0'), out_proj_covar=tensor([1.3847e-04, 7.4409e-05, 8.6333e-05, 9.0699e-05, 2.0385e-04, 1.5010e-04, 7.8190e-05, 1.4271e-04], device='cuda:0') 2023-03-07 14:52:32,126 INFO [train2.py:809] (0/4) Epoch 3, batch 2700, loss[ctc_loss=0.1666, att_loss=0.3117, loss=0.2827, over 17285.00 frames. utt_duration=1004 frames, utt_pad_proportion=0.05232, over 69.00 utterances.], tot_loss[ctc_loss=0.2061, att_loss=0.3041, loss=0.2845, over 3256866.24 frames. utt_duration=1200 frames, utt_pad_proportion=0.07179, over 10868.20 utterances.], batch size: 69, lr: 3.03e-02, grad_scale: 16.0 2023-03-07 14:53:14,697 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10695.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:53:28,794 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.793e+02 4.160e+02 5.458e+02 6.683e+02 1.246e+03, threshold=1.092e+03, percent-clipped=5.0 2023-03-07 14:53:35,434 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9061, 4.7922, 4.6880, 3.1132, 4.7945, 3.9332, 4.0224, 2.4992], device='cuda:0'), covar=tensor([0.0111, 0.0087, 0.0220, 0.0628, 0.0076, 0.0226, 0.0258, 0.1465], device='cuda:0'), in_proj_covar=tensor([0.0041, 0.0041, 0.0037, 0.0066, 0.0039, 0.0050, 0.0056, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 14:53:51,026 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=10718.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 14:53:52,187 INFO [train2.py:809] (0/4) Epoch 3, batch 2750, loss[ctc_loss=0.2788, att_loss=0.3424, loss=0.3297, over 17250.00 frames. utt_duration=1171 frames, utt_pad_proportion=0.02492, over 59.00 utterances.], tot_loss[ctc_loss=0.205, att_loss=0.304, loss=0.2842, over 3269632.55 frames. utt_duration=1217 frames, utt_pad_proportion=0.06442, over 10763.85 utterances.], batch size: 59, lr: 3.03e-02, grad_scale: 16.0 2023-03-07 14:54:07,736 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7645, 5.2509, 5.0740, 5.2085, 4.8170, 5.0356, 5.5084, 5.2751], device='cuda:0'), covar=tensor([0.0250, 0.0221, 0.0385, 0.0123, 0.0285, 0.0112, 0.0185, 0.0120], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0119, 0.0143, 0.0090, 0.0131, 0.0099, 0.0116, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-03-07 14:54:08,375 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-07 14:54:20,966 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3882, 1.4550, 1.5611, 1.2466, 0.9084, 1.9401, 2.5734, 2.0566], device='cuda:0'), covar=tensor([0.0425, 0.1373, 0.0952, 0.1257, 0.1400, 0.0781, 0.0417, 0.0609], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0082, 0.0080, 0.0074, 0.0072, 0.0071, 0.0080, 0.0088], device='cuda:0'), out_proj_covar=tensor([3.8124e-05, 5.0425e-05, 4.6398e-05, 4.1793e-05, 4.5913e-05, 3.8802e-05, 4.2259e-05, 4.1405e-05], device='cuda:0') 2023-03-07 14:54:32,776 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10743.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:54:34,620 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10744.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:55:03,931 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0269, 4.9653, 4.8261, 3.1012, 4.8419, 4.2258, 4.3062, 2.3978], device='cuda:0'), covar=tensor([0.0124, 0.0064, 0.0175, 0.0630, 0.0062, 0.0188, 0.0223, 0.1492], device='cuda:0'), in_proj_covar=tensor([0.0041, 0.0042, 0.0038, 0.0067, 0.0040, 0.0050, 0.0057, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 14:55:06,761 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10765.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:55:08,185 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=10766.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 14:55:12,964 INFO [train2.py:809] (0/4) Epoch 3, batch 2800, loss[ctc_loss=0.2718, att_loss=0.3455, loss=0.3308, over 17366.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03469, over 63.00 utterances.], tot_loss[ctc_loss=0.204, att_loss=0.304, loss=0.284, over 3278018.29 frames. utt_duration=1240 frames, utt_pad_proportion=0.05586, over 10585.66 utterances.], batch size: 63, lr: 3.02e-02, grad_scale: 16.0 2023-03-07 14:55:21,937 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10774.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:55:57,909 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5909, 4.0216, 3.3342, 3.7023, 3.9975, 3.7804, 3.3752, 2.4887], device='cuda:0'), covar=tensor([0.0329, 0.0208, 0.0385, 0.0272, 0.0454, 0.0244, 0.0425, 0.2544], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0109, 0.0103, 0.0102, 0.0205, 0.0123, 0.0094, 0.0233], device='cuda:0'), out_proj_covar=tensor([1.0643e-04, 8.9667e-05, 8.6821e-05, 8.9965e-05, 1.8690e-04, 1.0268e-04, 8.5801e-05, 1.9479e-04], device='cuda:0') 2023-03-07 14:56:02,762 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2669, 3.5471, 3.1998, 3.3633, 3.7589, 3.2334, 1.9799, 4.2851], device='cuda:0'), covar=tensor([0.1254, 0.0313, 0.0891, 0.0563, 0.0416, 0.0679, 0.1156, 0.0149], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0089, 0.0140, 0.0114, 0.0098, 0.0135, 0.0121, 0.0073], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-03-07 14:56:10,552 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.287e+02 4.175e+02 5.006e+02 5.883e+02 1.707e+03, threshold=1.001e+03, percent-clipped=3.0 2023-03-07 14:56:12,568 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10805.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:56:20,271 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=10810.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 14:56:25,160 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5993, 1.9249, 4.8930, 3.9305, 3.2717, 4.3655, 4.4704, 4.4381], device='cuda:0'), covar=tensor([0.0122, 0.1811, 0.0115, 0.1090, 0.2424, 0.0324, 0.0254, 0.0325], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0210, 0.0113, 0.0257, 0.0312, 0.0163, 0.0105, 0.0116], device='cuda:0'), out_proj_covar=tensor([9.4527e-05, 1.5750e-04, 9.2303e-05, 2.0386e-04, 2.3044e-04, 1.3089e-04, 8.6534e-05, 9.5818e-05], device='cuda:0') 2023-03-07 14:56:33,912 INFO [train2.py:809] (0/4) Epoch 3, batch 2850, loss[ctc_loss=0.1529, att_loss=0.2727, loss=0.2487, over 16016.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006784, over 40.00 utterances.], tot_loss[ctc_loss=0.2026, att_loss=0.303, loss=0.2829, over 3271542.81 frames. utt_duration=1247 frames, utt_pad_proportion=0.05608, over 10503.19 utterances.], batch size: 40, lr: 3.02e-02, grad_scale: 16.0 2023-03-07 14:57:00,140 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10835.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 14:57:54,307 INFO [train2.py:809] (0/4) Epoch 3, batch 2900, loss[ctc_loss=0.2084, att_loss=0.3112, loss=0.2907, over 16678.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.005822, over 46.00 utterances.], tot_loss[ctc_loss=0.2043, att_loss=0.3043, loss=0.2843, over 3279333.87 frames. utt_duration=1257 frames, utt_pad_proportion=0.05185, over 10446.14 utterances.], batch size: 46, lr: 3.01e-02, grad_scale: 4.0 2023-03-07 14:58:21,006 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=10885.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 14:58:54,580 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.371e+02 4.021e+02 4.974e+02 6.236e+02 1.292e+03, threshold=9.948e+02, percent-clipped=6.0 2023-03-07 14:59:14,724 INFO [train2.py:809] (0/4) Epoch 3, batch 2950, loss[ctc_loss=0.2435, att_loss=0.3375, loss=0.3187, over 17309.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01068, over 55.00 utterances.], tot_loss[ctc_loss=0.2025, att_loss=0.3031, loss=0.283, over 3272045.08 frames. utt_duration=1259 frames, utt_pad_proportion=0.05282, over 10406.93 utterances.], batch size: 55, lr: 3.01e-02, grad_scale: 4.0 2023-03-07 14:59:59,106 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=10946.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 15:00:34,064 INFO [train2.py:809] (0/4) Epoch 3, batch 3000, loss[ctc_loss=0.2046, att_loss=0.3127, loss=0.2911, over 17389.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.0475, over 69.00 utterances.], tot_loss[ctc_loss=0.2027, att_loss=0.3034, loss=0.2832, over 3278129.27 frames. utt_duration=1265 frames, utt_pad_proportion=0.04986, over 10381.72 utterances.], batch size: 69, lr: 3.00e-02, grad_scale: 4.0 2023-03-07 15:00:34,066 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 15:00:40,985 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3288, 2.9158, 3.7380, 2.5198, 3.4207, 4.4882, 4.3041, 3.3810], device='cuda:0'), covar=tensor([0.0354, 0.1446, 0.0671, 0.1694, 0.1061, 0.0268, 0.0362, 0.1228], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0169, 0.0146, 0.0158, 0.0175, 0.0109, 0.0116, 0.0167], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 15:00:45,289 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5820, 4.0929, 3.6851, 3.9040, 4.0982, 3.8402, 3.7264, 1.8779], device='cuda:0'), covar=tensor([0.0356, 0.0359, 0.0337, 0.0182, 0.0962, 0.0275, 0.0523, 0.3812], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0113, 0.0106, 0.0105, 0.0210, 0.0127, 0.0102, 0.0239], device='cuda:0'), out_proj_covar=tensor([1.1006e-04, 9.3144e-05, 8.9289e-05, 9.3188e-05, 1.9153e-04, 1.0605e-04, 9.1528e-05, 2.0106e-04], device='cuda:0') 2023-03-07 15:00:47,744 INFO [train2.py:843] (0/4) Epoch 3, validation: ctc_loss=0.1004, att_loss=0.2657, loss=0.2327, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 15:00:47,745 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 15:01:48,348 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.705e+02 4.292e+02 5.307e+02 6.516e+02 1.303e+03, threshold=1.061e+03, percent-clipped=4.0 2023-03-07 15:02:09,267 INFO [train2.py:809] (0/4) Epoch 3, batch 3050, loss[ctc_loss=0.1611, att_loss=0.2692, loss=0.2476, over 16016.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006986, over 40.00 utterances.], tot_loss[ctc_loss=0.2017, att_loss=0.3026, loss=0.2824, over 3273961.72 frames. utt_duration=1248 frames, utt_pad_proportion=0.05557, over 10505.71 utterances.], batch size: 40, lr: 2.99e-02, grad_scale: 4.0 2023-03-07 15:02:45,137 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0769, 4.9344, 4.8124, 3.1170, 4.8741, 4.1885, 3.9658, 2.3974], device='cuda:0'), covar=tensor([0.0110, 0.0065, 0.0185, 0.0657, 0.0070, 0.0137, 0.0263, 0.1413], device='cuda:0'), in_proj_covar=tensor([0.0041, 0.0043, 0.0039, 0.0069, 0.0041, 0.0052, 0.0058, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 15:03:23,368 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11065.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:03:29,050 INFO [train2.py:809] (0/4) Epoch 3, batch 3100, loss[ctc_loss=0.2317, att_loss=0.3257, loss=0.3069, over 17075.00 frames. utt_duration=1221 frames, utt_pad_proportion=0.01756, over 56.00 utterances.], tot_loss[ctc_loss=0.2005, att_loss=0.3013, loss=0.2812, over 3260486.85 frames. utt_duration=1273 frames, utt_pad_proportion=0.05039, over 10257.78 utterances.], batch size: 56, lr: 2.99e-02, grad_scale: 4.0 2023-03-07 15:04:20,577 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11100.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:04:29,505 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.231e+02 3.861e+02 4.979e+02 6.138e+02 1.576e+03, threshold=9.958e+02, percent-clipped=5.0 2023-03-07 15:04:35,978 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11110.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 15:04:40,459 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11113.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:04:50,174 INFO [train2.py:809] (0/4) Epoch 3, batch 3150, loss[ctc_loss=0.2831, att_loss=0.3412, loss=0.3296, over 16392.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.008357, over 44.00 utterances.], tot_loss[ctc_loss=0.1998, att_loss=0.3007, loss=0.2806, over 3254640.54 frames. utt_duration=1298 frames, utt_pad_proportion=0.045, over 10039.60 utterances.], batch size: 44, lr: 2.98e-02, grad_scale: 4.0 2023-03-07 15:04:50,509 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.8086, 2.3638, 2.6020, 1.7666, 1.4506, 2.7533, 2.0515, 1.4808], device='cuda:0'), covar=tensor([0.1048, 0.0888, 0.0933, 0.2418, 0.2151, 0.1046, 0.1614, 0.4405], device='cuda:0'), in_proj_covar=tensor([0.0060, 0.0047, 0.0052, 0.0065, 0.0048, 0.0056, 0.0055, 0.0077], device='cuda:0'), out_proj_covar=tensor([3.5549e-05, 3.0338e-05, 3.3885e-05, 4.0940e-05, 3.3530e-05, 3.3016e-05, 3.5876e-05, 5.1915e-05], device='cuda:0') 2023-03-07 15:05:08,073 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11130.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:05:31,819 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-07 15:05:53,205 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11158.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:06:11,074 INFO [train2.py:809] (0/4) Epoch 3, batch 3200, loss[ctc_loss=0.1846, att_loss=0.3058, loss=0.2816, over 17346.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03678, over 63.00 utterances.], tot_loss[ctc_loss=0.1999, att_loss=0.3011, loss=0.2809, over 3260768.42 frames. utt_duration=1275 frames, utt_pad_proportion=0.04864, over 10241.51 utterances.], batch size: 63, lr: 2.98e-02, grad_scale: 8.0 2023-03-07 15:06:19,876 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11174.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:06:48,279 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-03-07 15:07:10,782 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.395e+02 3.760e+02 4.677e+02 5.867e+02 1.602e+03, threshold=9.354e+02, percent-clipped=2.0 2023-03-07 15:07:31,879 INFO [train2.py:809] (0/4) Epoch 3, batch 3250, loss[ctc_loss=0.1839, att_loss=0.3042, loss=0.2802, over 16951.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.00843, over 50.00 utterances.], tot_loss[ctc_loss=0.2005, att_loss=0.3017, loss=0.2815, over 3261093.24 frames. utt_duration=1266 frames, utt_pad_proportion=0.05097, over 10319.32 utterances.], batch size: 50, lr: 2.97e-02, grad_scale: 8.0 2023-03-07 15:07:58,313 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11235.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:08:07,814 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11241.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 15:08:52,863 INFO [train2.py:809] (0/4) Epoch 3, batch 3300, loss[ctc_loss=0.1538, att_loss=0.2661, loss=0.2436, over 15387.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.009551, over 35.00 utterances.], tot_loss[ctc_loss=0.1992, att_loss=0.301, loss=0.2806, over 3270163.01 frames. utt_duration=1278 frames, utt_pad_proportion=0.04566, over 10245.14 utterances.], batch size: 35, lr: 2.97e-02, grad_scale: 8.0 2023-03-07 15:09:52,441 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.248e+02 3.797e+02 4.904e+02 6.059e+02 1.982e+03, threshold=9.808e+02, percent-clipped=6.0 2023-03-07 15:10:12,979 INFO [train2.py:809] (0/4) Epoch 3, batch 3350, loss[ctc_loss=0.1557, att_loss=0.277, loss=0.2528, over 16184.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006121, over 41.00 utterances.], tot_loss[ctc_loss=0.1999, att_loss=0.3017, loss=0.2813, over 3272739.26 frames. utt_duration=1251 frames, utt_pad_proportion=0.05092, over 10474.98 utterances.], batch size: 41, lr: 2.96e-02, grad_scale: 8.0 2023-03-07 15:10:29,642 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7159, 2.2175, 4.9335, 3.7793, 3.3077, 4.4681, 4.6929, 4.5424], device='cuda:0'), covar=tensor([0.0119, 0.1687, 0.0148, 0.1194, 0.2243, 0.0306, 0.0157, 0.0247], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0217, 0.0114, 0.0262, 0.0313, 0.0165, 0.0106, 0.0115], device='cuda:0'), out_proj_covar=tensor([9.7522e-05, 1.6256e-04, 9.2110e-05, 2.0903e-04, 2.3254e-04, 1.3201e-04, 8.8022e-05, 9.7157e-05], device='cuda:0') 2023-03-07 15:11:14,838 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1992, 4.6942, 4.7756, 3.9392, 1.8796, 2.5775, 4.1376, 3.6693], device='cuda:0'), covar=tensor([0.0501, 0.0141, 0.0113, 0.0642, 0.8099, 0.2486, 0.0357, 0.1794], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0144, 0.0156, 0.0180, 0.0399, 0.0299, 0.0151, 0.0258], device='cuda:0'), out_proj_covar=tensor([1.3545e-04, 7.5493e-05, 8.3877e-05, 8.7036e-05, 1.9846e-04, 1.4974e-04, 7.9082e-05, 1.4432e-04], device='cuda:0') 2023-03-07 15:11:23,760 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7734, 5.3315, 5.1637, 5.0659, 5.4679, 5.2466, 5.1601, 5.0772], device='cuda:0'), covar=tensor([0.1245, 0.0245, 0.0157, 0.0332, 0.0176, 0.0241, 0.0184, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0312, 0.0189, 0.0130, 0.0158, 0.0203, 0.0224, 0.0174, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:11:34,093 INFO [train2.py:809] (0/4) Epoch 3, batch 3400, loss[ctc_loss=0.2034, att_loss=0.3246, loss=0.3004, over 17400.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.04698, over 69.00 utterances.], tot_loss[ctc_loss=0.2015, att_loss=0.3029, loss=0.2826, over 3270991.45 frames. utt_duration=1211 frames, utt_pad_proportion=0.06319, over 10821.66 utterances.], batch size: 69, lr: 2.96e-02, grad_scale: 8.0 2023-03-07 15:12:25,245 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11400.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:12:34,185 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.436e+02 3.676e+02 4.769e+02 6.355e+02 1.428e+03, threshold=9.539e+02, percent-clipped=8.0 2023-03-07 15:12:55,340 INFO [train2.py:809] (0/4) Epoch 3, batch 3450, loss[ctc_loss=0.2159, att_loss=0.3285, loss=0.306, over 17289.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01192, over 55.00 utterances.], tot_loss[ctc_loss=0.1992, att_loss=0.301, loss=0.2806, over 3265572.78 frames. utt_duration=1227 frames, utt_pad_proportion=0.06141, over 10656.37 utterances.], batch size: 55, lr: 2.95e-02, grad_scale: 8.0 2023-03-07 15:13:12,833 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11430.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:13:41,671 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11448.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:14:15,451 INFO [train2.py:809] (0/4) Epoch 3, batch 3500, loss[ctc_loss=0.1928, att_loss=0.3068, loss=0.284, over 16952.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008079, over 50.00 utterances.], tot_loss[ctc_loss=0.1984, att_loss=0.3011, loss=0.2806, over 3268743.19 frames. utt_duration=1236 frames, utt_pad_proportion=0.05896, over 10592.17 utterances.], batch size: 50, lr: 2.95e-02, grad_scale: 8.0 2023-03-07 15:14:29,619 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11478.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:14:57,982 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9275, 1.4152, 1.9990, 0.9701, 1.5524, 2.0944, 2.1479, 2.5078], device='cuda:0'), covar=tensor([0.0358, 0.1668, 0.1177, 0.1539, 0.1199, 0.0750, 0.0786, 0.0537], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0087, 0.0090, 0.0080, 0.0073, 0.0076, 0.0090, 0.0099], device='cuda:0'), out_proj_covar=tensor([3.8679e-05, 5.0088e-05, 4.8255e-05, 4.5736e-05, 4.5570e-05, 4.1166e-05, 4.2022e-05, 4.1487e-05], device='cuda:0') 2023-03-07 15:15:14,773 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.985e+02 3.561e+02 4.473e+02 5.619e+02 1.545e+03, threshold=8.947e+02, percent-clipped=3.0 2023-03-07 15:15:20,549 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.09 vs. limit=5.0 2023-03-07 15:15:36,469 INFO [train2.py:809] (0/4) Epoch 3, batch 3550, loss[ctc_loss=0.1485, att_loss=0.2581, loss=0.2362, over 15864.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.00914, over 39.00 utterances.], tot_loss[ctc_loss=0.1965, att_loss=0.3001, loss=0.2794, over 3273800.83 frames. utt_duration=1253 frames, utt_pad_proportion=0.05203, over 10466.58 utterances.], batch size: 39, lr: 2.94e-02, grad_scale: 8.0 2023-03-07 15:15:53,466 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11530.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:15:56,788 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11532.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:16:12,289 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11541.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 15:16:56,927 INFO [train2.py:809] (0/4) Epoch 3, batch 3600, loss[ctc_loss=0.1825, att_loss=0.2748, loss=0.2563, over 15593.00 frames. utt_duration=1687 frames, utt_pad_proportion=0.01048, over 37.00 utterances.], tot_loss[ctc_loss=0.1983, att_loss=0.3008, loss=0.2803, over 3272351.16 frames. utt_duration=1235 frames, utt_pad_proportion=0.05682, over 10608.37 utterances.], batch size: 37, lr: 2.93e-02, grad_scale: 8.0 2023-03-07 15:17:01,635 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3776, 4.6879, 4.3226, 4.6687, 4.6867, 4.4416, 4.2704, 4.6010], device='cuda:0'), covar=tensor([0.0127, 0.0167, 0.0143, 0.0116, 0.0125, 0.0126, 0.0282, 0.0230], device='cuda:0'), in_proj_covar=tensor([0.0047, 0.0047, 0.0048, 0.0036, 0.0034, 0.0043, 0.0063, 0.0058], device='cuda:0'), out_proj_covar=tensor([1.3380e-04, 1.3394e-04, 1.5938e-04, 1.0586e-04, 9.8122e-05, 1.2840e-04, 1.7930e-04, 1.6921e-04], device='cuda:0') 2023-03-07 15:17:29,000 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11589.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 15:17:36,035 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11593.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:17:43,588 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8810, 6.0573, 5.5244, 6.0660, 5.7204, 5.5855, 5.5566, 5.3910], device='cuda:0'), covar=tensor([0.1141, 0.0898, 0.0724, 0.0673, 0.0628, 0.1162, 0.2255, 0.2555], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0322, 0.0260, 0.0254, 0.0232, 0.0323, 0.0349, 0.0331], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:17:48,730 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9336, 4.7767, 5.0488, 3.5254, 4.6749, 4.1891, 4.2618, 2.6828], device='cuda:0'), covar=tensor([0.0227, 0.0104, 0.0150, 0.0714, 0.0122, 0.0176, 0.0278, 0.1717], device='cuda:0'), in_proj_covar=tensor([0.0043, 0.0046, 0.0039, 0.0074, 0.0043, 0.0054, 0.0062, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 15:17:56,230 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.467e+02 4.035e+02 5.369e+02 6.421e+02 1.595e+03, threshold=1.074e+03, percent-clipped=5.0 2023-03-07 15:17:58,852 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.49 vs. limit=5.0 2023-03-07 15:18:18,104 INFO [train2.py:809] (0/4) Epoch 3, batch 3650, loss[ctc_loss=0.1828, att_loss=0.2715, loss=0.2538, over 15992.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008132, over 40.00 utterances.], tot_loss[ctc_loss=0.1991, att_loss=0.3013, loss=0.2808, over 3273988.96 frames. utt_duration=1209 frames, utt_pad_proportion=0.06254, over 10844.22 utterances.], batch size: 40, lr: 2.93e-02, grad_scale: 8.0 2023-03-07 15:18:43,253 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11635.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:19:38,109 INFO [train2.py:809] (0/4) Epoch 3, batch 3700, loss[ctc_loss=0.2631, att_loss=0.3345, loss=0.3202, over 14326.00 frames. utt_duration=391.4 frames, utt_pad_proportion=0.3145, over 147.00 utterances.], tot_loss[ctc_loss=0.1976, att_loss=0.3007, loss=0.2801, over 3270728.66 frames. utt_duration=1226 frames, utt_pad_proportion=0.06081, over 10687.19 utterances.], batch size: 147, lr: 2.92e-02, grad_scale: 8.0 2023-03-07 15:20:18,465 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-03-07 15:20:19,414 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.6393, 1.6348, 2.2947, 1.9358, 1.3362, 2.2453, 1.8646, 1.4137], device='cuda:0'), covar=tensor([0.0830, 0.1733, 0.1133, 0.2402, 0.3449, 0.0877, 0.1146, 0.4016], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0058, 0.0057, 0.0071, 0.0057, 0.0062, 0.0057, 0.0083], device='cuda:0'), out_proj_covar=tensor([3.8325e-05, 3.5610e-05, 3.5715e-05, 4.4653e-05, 3.9558e-05, 3.7718e-05, 3.8041e-05, 5.6795e-05], device='cuda:0') 2023-03-07 15:20:22,409 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11696.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:20:27,160 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11699.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:20:28,594 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7061, 4.5933, 4.6362, 3.1923, 4.5170, 3.9739, 4.1326, 2.6249], device='cuda:0'), covar=tensor([0.0097, 0.0102, 0.0166, 0.0619, 0.0093, 0.0178, 0.0220, 0.1310], device='cuda:0'), in_proj_covar=tensor([0.0044, 0.0046, 0.0040, 0.0075, 0.0045, 0.0055, 0.0063, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 15:20:37,359 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.856e+02 3.737e+02 4.615e+02 5.715e+02 1.108e+03, threshold=9.229e+02, percent-clipped=1.0 2023-03-07 15:20:58,425 INFO [train2.py:809] (0/4) Epoch 3, batch 3750, loss[ctc_loss=0.1712, att_loss=0.286, loss=0.263, over 16286.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006853, over 43.00 utterances.], tot_loss[ctc_loss=0.1968, att_loss=0.3, loss=0.2794, over 3274795.29 frames. utt_duration=1243 frames, utt_pad_proportion=0.05609, over 10555.25 utterances.], batch size: 43, lr: 2.92e-02, grad_scale: 8.0 2023-03-07 15:21:04,850 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5842, 3.8641, 3.1560, 3.2674, 3.9911, 3.4962, 2.0062, 4.6200], device='cuda:0'), covar=tensor([0.1079, 0.0411, 0.1142, 0.0626, 0.0481, 0.0844, 0.1198, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0095, 0.0147, 0.0117, 0.0112, 0.0143, 0.0125, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001], device='cuda:0') 2023-03-07 15:22:04,550 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11760.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:22:18,742 INFO [train2.py:809] (0/4) Epoch 3, batch 3800, loss[ctc_loss=0.1354, att_loss=0.2543, loss=0.2305, over 15405.00 frames. utt_duration=1762 frames, utt_pad_proportion=0.008666, over 35.00 utterances.], tot_loss[ctc_loss=0.1967, att_loss=0.3001, loss=0.2794, over 3274982.09 frames. utt_duration=1239 frames, utt_pad_proportion=0.05665, over 10586.07 utterances.], batch size: 35, lr: 2.91e-02, grad_scale: 8.0 2023-03-07 15:22:44,247 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9823, 4.3219, 4.7225, 4.9358, 1.9822, 4.6108, 2.9000, 2.9034], device='cuda:0'), covar=tensor([0.0334, 0.0215, 0.0602, 0.0116, 0.3307, 0.0145, 0.1470, 0.1226], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0091, 0.0203, 0.0104, 0.0219, 0.0093, 0.0183, 0.0179], device='cuda:0'), out_proj_covar=tensor([8.7843e-05, 8.4070e-05, 1.6980e-04, 8.5975e-05, 1.7450e-04, 8.2237e-05, 1.5006e-04, 1.4727e-04], device='cuda:0') 2023-03-07 15:23:02,454 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.11 vs. limit=2.0 2023-03-07 15:23:04,012 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.38 vs. limit=2.0 2023-03-07 15:23:12,507 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-07 15:23:18,126 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.859e+02 4.239e+02 5.484e+02 6.646e+02 1.603e+03, threshold=1.097e+03, percent-clipped=8.0 2023-03-07 15:23:33,650 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11816.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:23:37,905 INFO [train2.py:809] (0/4) Epoch 3, batch 3850, loss[ctc_loss=0.2123, att_loss=0.3092, loss=0.2898, over 17073.00 frames. utt_duration=1315 frames, utt_pad_proportion=0.006943, over 52.00 utterances.], tot_loss[ctc_loss=0.1959, att_loss=0.2997, loss=0.2789, over 3273632.50 frames. utt_duration=1241 frames, utt_pad_proportion=0.05664, over 10561.58 utterances.], batch size: 52, lr: 2.91e-02, grad_scale: 8.0 2023-03-07 15:23:55,912 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=11830.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:24:20,870 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11846.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:24:55,936 INFO [train2.py:809] (0/4) Epoch 3, batch 3900, loss[ctc_loss=0.2082, att_loss=0.3096, loss=0.2893, over 16609.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006544, over 47.00 utterances.], tot_loss[ctc_loss=0.1937, att_loss=0.2983, loss=0.2774, over 3275808.27 frames. utt_duration=1261 frames, utt_pad_proportion=0.05217, over 10406.20 utterances.], batch size: 47, lr: 2.90e-02, grad_scale: 8.0 2023-03-07 15:25:08,377 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1937, 5.3456, 4.6468, 5.3215, 5.0390, 4.7463, 4.8803, 4.6581], device='cuda:0'), covar=tensor([0.0785, 0.0740, 0.0687, 0.0582, 0.0632, 0.1150, 0.1589, 0.1756], device='cuda:0'), in_proj_covar=tensor([0.0276, 0.0305, 0.0255, 0.0248, 0.0230, 0.0315, 0.0335, 0.0329], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:25:08,641 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11877.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:25:09,898 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=11878.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:25:25,508 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11888.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:25:53,042 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.007e+02 3.699e+02 4.427e+02 5.534e+02 1.749e+03, threshold=8.855e+02, percent-clipped=3.0 2023-03-07 15:25:54,868 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11907.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:26:13,112 INFO [train2.py:809] (0/4) Epoch 3, batch 3950, loss[ctc_loss=0.2064, att_loss=0.3199, loss=0.2972, over 17058.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009369, over 53.00 utterances.], tot_loss[ctc_loss=0.1934, att_loss=0.2981, loss=0.2772, over 3256993.00 frames. utt_duration=1249 frames, utt_pad_proportion=0.05712, over 10442.21 utterances.], batch size: 53, lr: 2.90e-02, grad_scale: 8.0 2023-03-07 15:26:14,872 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11920.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:26:27,235 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0824, 4.3583, 4.8796, 4.7200, 1.9630, 4.6849, 2.7853, 3.3303], device='cuda:0'), covar=tensor([0.0292, 0.0273, 0.0553, 0.0168, 0.3335, 0.0133, 0.1544, 0.0986], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0093, 0.0207, 0.0107, 0.0223, 0.0094, 0.0184, 0.0182], device='cuda:0'), out_proj_covar=tensor([8.7842e-05, 8.5030e-05, 1.7331e-04, 8.8630e-05, 1.7790e-04, 8.2071e-05, 1.5206e-04, 1.4980e-04], device='cuda:0') 2023-03-07 15:26:28,594 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7778, 2.0196, 2.4083, 4.3006, 4.3402, 4.3945, 2.8293, 1.5823], device='cuda:0'), covar=tensor([0.0325, 0.2719, 0.2117, 0.0477, 0.0187, 0.0174, 0.1652, 0.3060], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0175, 0.0172, 0.0119, 0.0099, 0.0098, 0.0173, 0.0163], device='cuda:0'), out_proj_covar=tensor([1.2493e-04, 1.6859e-04, 1.7178e-04, 1.3311e-04, 1.0753e-04, 9.8815e-05, 1.7314e-04, 1.5994e-04], device='cuda:0') 2023-03-07 15:26:51,436 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8034, 5.1197, 5.5160, 5.6541, 5.0884, 5.7330, 5.0281, 5.9120], device='cuda:0'), covar=tensor([0.0504, 0.0473, 0.0413, 0.0412, 0.1779, 0.0678, 0.0464, 0.0411], device='cuda:0'), in_proj_covar=tensor([0.0390, 0.0266, 0.0260, 0.0310, 0.0459, 0.0255, 0.0227, 0.0291], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 15:26:53,621 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.66 vs. limit=2.0 2023-03-07 15:26:57,660 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11948.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:27:04,404 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-3.pt 2023-03-07 15:27:26,996 INFO [train2.py:809] (0/4) Epoch 4, batch 0, loss[ctc_loss=0.1911, att_loss=0.3039, loss=0.2814, over 17328.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.009705, over 55.00 utterances.], tot_loss[ctc_loss=0.1911, att_loss=0.3039, loss=0.2814, over 17328.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.009705, over 55.00 utterances.], batch size: 55, lr: 2.71e-02, grad_scale: 8.0 2023-03-07 15:27:26,998 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 15:27:39,231 INFO [train2.py:843] (0/4) Epoch 4, validation: ctc_loss=0.101, att_loss=0.2624, loss=0.2301, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 15:27:39,232 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 15:28:24,353 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=11981.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:28:28,831 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=11984.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:28:39,301 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=11991.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:28:53,382 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-12000.pt 2023-03-07 15:29:03,780 INFO [train2.py:809] (0/4) Epoch 4, batch 50, loss[ctc_loss=0.183, att_loss=0.2781, loss=0.259, over 13277.00 frames. utt_duration=1832 frames, utt_pad_proportion=0.1017, over 29.00 utterances.], tot_loss[ctc_loss=0.196, att_loss=0.299, loss=0.2784, over 735090.04 frames. utt_duration=1250 frames, utt_pad_proportion=0.05631, over 2355.02 utterances.], batch size: 29, lr: 2.70e-02, grad_scale: 8.0 2023-03-07 15:29:08,345 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.811e+02 4.015e+02 4.845e+02 5.994e+02 1.313e+03, threshold=9.689e+02, percent-clipped=6.0 2023-03-07 15:29:13,792 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12009.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:30:10,845 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12045.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:30:24,238 INFO [train2.py:809] (0/4) Epoch 4, batch 100, loss[ctc_loss=0.1896, att_loss=0.3168, loss=0.2913, over 17411.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.02809, over 63.00 utterances.], tot_loss[ctc_loss=0.186, att_loss=0.2929, loss=0.2715, over 1286382.06 frames. utt_duration=1325 frames, utt_pad_proportion=0.04421, over 3886.75 utterances.], batch size: 63, lr: 2.70e-02, grad_scale: 8.0 2023-03-07 15:30:27,568 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12055.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:31:01,989 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12076.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:31:33,474 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.03 vs. limit=2.0 2023-03-07 15:31:44,684 INFO [train2.py:809] (0/4) Epoch 4, batch 150, loss[ctc_loss=0.2076, att_loss=0.3104, loss=0.2899, over 16983.00 frames. utt_duration=687.8 frames, utt_pad_proportion=0.137, over 99.00 utterances.], tot_loss[ctc_loss=0.1902, att_loss=0.2959, loss=0.2748, over 1732464.34 frames. utt_duration=1256 frames, utt_pad_proportion=0.05396, over 5524.92 utterances.], batch size: 99, lr: 2.69e-02, grad_scale: 8.0 2023-03-07 15:31:49,483 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.353e+02 4.093e+02 5.170e+02 6.831e+02 1.205e+03, threshold=1.034e+03, percent-clipped=5.0 2023-03-07 15:32:01,242 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9908, 5.3112, 4.9328, 5.2224, 5.4756, 5.3501, 5.2773, 5.0158], device='cuda:0'), covar=tensor([0.0914, 0.0349, 0.0343, 0.0411, 0.0250, 0.0224, 0.0192, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0317, 0.0191, 0.0132, 0.0166, 0.0207, 0.0225, 0.0180, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:32:38,618 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12137.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:33:04,150 INFO [train2.py:809] (0/4) Epoch 4, batch 200, loss[ctc_loss=0.2152, att_loss=0.3107, loss=0.2916, over 16473.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006358, over 46.00 utterances.], tot_loss[ctc_loss=0.1889, att_loss=0.2949, loss=0.2737, over 2077467.02 frames. utt_duration=1267 frames, utt_pad_proportion=0.04958, over 6566.67 utterances.], batch size: 46, lr: 2.69e-02, grad_scale: 8.0 2023-03-07 15:33:04,526 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8284, 1.9223, 2.0364, 0.9480, 1.3451, 1.6244, 1.7325, 1.6666], device='cuda:0'), covar=tensor([0.0232, 0.0944, 0.1360, 0.1119, 0.0916, 0.0906, 0.0975, 0.0965], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0084, 0.0088, 0.0073, 0.0071, 0.0074, 0.0087, 0.0092], device='cuda:0'), out_proj_covar=tensor([3.6643e-05, 4.6416e-05, 4.6060e-05, 4.1055e-05, 4.0619e-05, 4.0004e-05, 4.0340e-05, 4.1029e-05], device='cuda:0') 2023-03-07 15:33:22,016 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5697, 4.4462, 4.4476, 4.8277, 4.8447, 4.5303, 4.3064, 1.9131], device='cuda:0'), covar=tensor([0.0348, 0.0434, 0.0244, 0.0150, 0.1333, 0.0322, 0.0406, 0.3980], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0116, 0.0107, 0.0111, 0.0229, 0.0136, 0.0103, 0.0251], device='cuda:0'), out_proj_covar=tensor([1.1449e-04, 9.9086e-05, 9.4257e-05, 1.0200e-04, 2.0984e-04, 1.1776e-04, 9.5303e-05, 2.1650e-04], device='cuda:0') 2023-03-07 15:33:34,660 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12172.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:33:51,666 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3024, 4.4406, 4.3567, 4.7452, 4.9090, 4.4631, 4.2686, 1.9377], device='cuda:0'), covar=tensor([0.0311, 0.0351, 0.0221, 0.0101, 0.0678, 0.0262, 0.0349, 0.3689], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0111, 0.0104, 0.0107, 0.0222, 0.0133, 0.0100, 0.0244], device='cuda:0'), out_proj_covar=tensor([1.1165e-04, 9.5703e-05, 9.1969e-05, 9.9158e-05, 2.0379e-04, 1.1539e-04, 9.3114e-05, 2.1059e-04], device='cuda:0') 2023-03-07 15:33:59,102 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12188.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:34:21,516 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12202.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:34:23,024 INFO [train2.py:809] (0/4) Epoch 4, batch 250, loss[ctc_loss=0.1626, att_loss=0.274, loss=0.2517, over 15940.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.00609, over 41.00 utterances.], tot_loss[ctc_loss=0.1902, att_loss=0.2964, loss=0.2751, over 2349249.53 frames. utt_duration=1251 frames, utt_pad_proportion=0.05147, over 7518.37 utterances.], batch size: 41, lr: 2.68e-02, grad_scale: 8.0 2023-03-07 15:34:28,452 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.253e+02 3.783e+02 4.490e+02 5.232e+02 8.659e+02, threshold=8.979e+02, percent-clipped=0.0 2023-03-07 15:34:41,749 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.08 vs. limit=2.0 2023-03-07 15:34:54,094 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3925, 2.1770, 2.8484, 4.2721, 3.8931, 4.1799, 2.6573, 1.6877], device='cuda:0'), covar=tensor([0.0511, 0.2632, 0.1436, 0.0385, 0.0353, 0.0299, 0.2041, 0.2970], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0176, 0.0173, 0.0117, 0.0105, 0.0099, 0.0175, 0.0163], device='cuda:0'), out_proj_covar=tensor([1.2642e-04, 1.6967e-04, 1.7276e-04, 1.3190e-04, 1.1267e-04, 9.9583e-05, 1.7595e-04, 1.6055e-04], device='cuda:0') 2023-03-07 15:34:58,911 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0733, 4.9445, 5.0578, 3.3530, 4.8976, 4.1823, 4.1406, 2.3487], device='cuda:0'), covar=tensor([0.0134, 0.0071, 0.0121, 0.0615, 0.0086, 0.0169, 0.0266, 0.1576], device='cuda:0'), in_proj_covar=tensor([0.0043, 0.0046, 0.0040, 0.0077, 0.0047, 0.0056, 0.0064, 0.0094], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 15:35:15,623 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12236.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:35:42,722 INFO [train2.py:809] (0/4) Epoch 4, batch 300, loss[ctc_loss=0.1238, att_loss=0.2371, loss=0.2144, over 14591.00 frames. utt_duration=1826 frames, utt_pad_proportion=0.03866, over 32.00 utterances.], tot_loss[ctc_loss=0.1909, att_loss=0.2974, loss=0.2761, over 2564636.91 frames. utt_duration=1241 frames, utt_pad_proportion=0.05008, over 8278.39 utterances.], batch size: 32, lr: 2.68e-02, grad_scale: 8.0 2023-03-07 15:36:19,383 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12276.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:36:42,496 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12291.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:36:47,996 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-03-07 15:37:00,290 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0471, 5.2598, 5.5247, 5.6456, 5.3888, 6.0004, 5.1320, 6.0952], device='cuda:0'), covar=tensor([0.0446, 0.0646, 0.0457, 0.0526, 0.1481, 0.0508, 0.0430, 0.0340], device='cuda:0'), in_proj_covar=tensor([0.0404, 0.0277, 0.0271, 0.0323, 0.0478, 0.0276, 0.0233, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 15:37:01,516 INFO [train2.py:809] (0/4) Epoch 4, batch 350, loss[ctc_loss=0.1776, att_loss=0.2718, loss=0.2529, over 15892.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008724, over 39.00 utterances.], tot_loss[ctc_loss=0.1898, att_loss=0.2958, loss=0.2746, over 2716959.32 frames. utt_duration=1250 frames, utt_pad_proportion=0.05058, over 8702.19 utterances.], batch size: 39, lr: 2.67e-02, grad_scale: 8.0 2023-03-07 15:37:03,263 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12304.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:37:06,085 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.120e+02 4.027e+02 4.741e+02 6.835e+02 1.897e+03, threshold=9.483e+02, percent-clipped=6.0 2023-03-07 15:37:55,405 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8869, 2.1132, 2.8718, 4.3815, 4.1163, 4.3445, 2.7821, 1.7792], device='cuda:0'), covar=tensor([0.0409, 0.3098, 0.1679, 0.0383, 0.0275, 0.0220, 0.1925, 0.3057], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0183, 0.0177, 0.0121, 0.0105, 0.0101, 0.0174, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 15:37:58,214 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12339.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:37:59,867 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12340.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:38:20,460 INFO [train2.py:809] (0/4) Epoch 4, batch 400, loss[ctc_loss=0.2437, att_loss=0.3295, loss=0.3124, over 14072.00 frames. utt_duration=387.1 frames, utt_pad_proportion=0.3232, over 146.00 utterances.], tot_loss[ctc_loss=0.1891, att_loss=0.2955, loss=0.2742, over 2840647.37 frames. utt_duration=1225 frames, utt_pad_proportion=0.05801, over 9288.92 utterances.], batch size: 146, lr: 2.67e-02, grad_scale: 8.0 2023-03-07 15:38:23,705 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12355.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:38:31,945 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6336, 2.0182, 2.5052, 3.9730, 3.8364, 4.0919, 2.8634, 1.9282], device='cuda:0'), covar=tensor([0.0306, 0.2447, 0.1629, 0.0574, 0.0296, 0.0145, 0.1330, 0.2300], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0180, 0.0175, 0.0120, 0.0104, 0.0100, 0.0170, 0.0165], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 15:38:57,420 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.06 vs. limit=2.0 2023-03-07 15:38:59,153 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2023-03-07 15:39:07,531 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9926, 4.6629, 4.3439, 4.6913, 4.5009, 4.3993, 4.0183, 4.3295], device='cuda:0'), covar=tensor([0.0132, 0.0154, 0.0094, 0.0113, 0.0120, 0.0116, 0.0351, 0.0257], device='cuda:0'), in_proj_covar=tensor([0.0048, 0.0047, 0.0047, 0.0036, 0.0036, 0.0043, 0.0062, 0.0058], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 15:39:15,494 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.05 vs. limit=2.0 2023-03-07 15:39:19,352 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7744, 5.8723, 5.3391, 5.9308, 5.5090, 5.2854, 5.3510, 5.1691], device='cuda:0'), covar=tensor([0.1010, 0.1006, 0.0815, 0.0689, 0.0621, 0.1268, 0.2233, 0.2281], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0331, 0.0268, 0.0260, 0.0240, 0.0330, 0.0359, 0.0343], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:39:33,097 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9546, 6.0318, 5.6463, 6.1474, 5.7186, 5.4923, 5.5507, 5.5408], device='cuda:0'), covar=tensor([0.1044, 0.0874, 0.0706, 0.0584, 0.0578, 0.1243, 0.1885, 0.1672], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0334, 0.0267, 0.0259, 0.0239, 0.0329, 0.0358, 0.0343], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:39:39,637 INFO [train2.py:809] (0/4) Epoch 4, batch 450, loss[ctc_loss=0.1939, att_loss=0.3081, loss=0.2852, over 16639.00 frames. utt_duration=1418 frames, utt_pad_proportion=0.004482, over 47.00 utterances.], tot_loss[ctc_loss=0.1897, att_loss=0.2966, loss=0.2752, over 2946118.64 frames. utt_duration=1217 frames, utt_pad_proportion=0.05742, over 9695.26 utterances.], batch size: 47, lr: 2.66e-02, grad_scale: 8.0 2023-03-07 15:39:39,716 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12403.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:39:44,235 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.374e+02 3.731e+02 4.738e+02 5.685e+02 9.673e+02, threshold=9.477e+02, percent-clipped=1.0 2023-03-07 15:39:47,543 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2023-03-07 15:40:25,801 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12432.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:40:47,541 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12446.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:40:58,414 INFO [train2.py:809] (0/4) Epoch 4, batch 500, loss[ctc_loss=0.1628, att_loss=0.268, loss=0.247, over 15772.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.00862, over 38.00 utterances.], tot_loss[ctc_loss=0.1885, att_loss=0.2961, loss=0.2746, over 3023867.92 frames. utt_duration=1242 frames, utt_pad_proportion=0.05002, over 9749.87 utterances.], batch size: 38, lr: 2.66e-02, grad_scale: 8.0 2023-03-07 15:41:28,650 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12472.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:42:15,268 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12502.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:42:16,490 INFO [train2.py:809] (0/4) Epoch 4, batch 550, loss[ctc_loss=0.2217, att_loss=0.3181, loss=0.2988, over 17385.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.04578, over 69.00 utterances.], tot_loss[ctc_loss=0.1883, att_loss=0.2953, loss=0.2739, over 3081473.23 frames. utt_duration=1263 frames, utt_pad_proportion=0.04648, over 9772.17 utterances.], batch size: 69, lr: 2.65e-02, grad_scale: 8.0 2023-03-07 15:42:21,769 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.239e+02 3.520e+02 4.552e+02 5.513e+02 1.151e+03, threshold=9.104e+02, percent-clipped=4.0 2023-03-07 15:42:23,574 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12507.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:42:43,524 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12520.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:43:31,154 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12550.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:43:35,779 INFO [train2.py:809] (0/4) Epoch 4, batch 600, loss[ctc_loss=0.1678, att_loss=0.2867, loss=0.2629, over 17012.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008152, over 51.00 utterances.], tot_loss[ctc_loss=0.1871, att_loss=0.2942, loss=0.2728, over 3120762.37 frames. utt_duration=1273 frames, utt_pad_proportion=0.04552, over 9817.61 utterances.], batch size: 51, lr: 2.65e-02, grad_scale: 8.0 2023-03-07 15:44:07,509 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-03-07 15:44:11,394 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6786, 5.0441, 4.8624, 4.7541, 5.1184, 5.1104, 4.8459, 4.8036], device='cuda:0'), covar=tensor([0.1028, 0.0405, 0.0189, 0.0477, 0.0285, 0.0229, 0.0252, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0333, 0.0199, 0.0139, 0.0174, 0.0216, 0.0234, 0.0185, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:44:12,994 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12576.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:44:55,222 INFO [train2.py:809] (0/4) Epoch 4, batch 650, loss[ctc_loss=0.2007, att_loss=0.2844, loss=0.2677, over 16272.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007569, over 43.00 utterances.], tot_loss[ctc_loss=0.1857, att_loss=0.2934, loss=0.2719, over 3151286.88 frames. utt_duration=1295 frames, utt_pad_proportion=0.04219, over 9744.82 utterances.], batch size: 43, lr: 2.65e-02, grad_scale: 8.0 2023-03-07 15:44:57,665 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12604.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:44:59,833 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-07 15:45:00,416 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 3.647e+02 4.227e+02 5.409e+02 1.351e+03, threshold=8.454e+02, percent-clipped=4.0 2023-03-07 15:45:29,070 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12624.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:45:53,178 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12640.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:46:13,056 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12652.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:46:14,655 INFO [train2.py:809] (0/4) Epoch 4, batch 700, loss[ctc_loss=0.1503, att_loss=0.2821, loss=0.2557, over 16628.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005259, over 47.00 utterances.], tot_loss[ctc_loss=0.1876, att_loss=0.2947, loss=0.2733, over 3181149.34 frames. utt_duration=1274 frames, utt_pad_proportion=0.0475, over 10000.53 utterances.], batch size: 47, lr: 2.64e-02, grad_scale: 8.0 2023-03-07 15:47:09,119 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12688.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:47:34,088 INFO [train2.py:809] (0/4) Epoch 4, batch 750, loss[ctc_loss=0.2053, att_loss=0.3165, loss=0.2943, over 17019.00 frames. utt_duration=1286 frames, utt_pad_proportion=0.01083, over 53.00 utterances.], tot_loss[ctc_loss=0.186, att_loss=0.2935, loss=0.272, over 3197371.20 frames. utt_duration=1281 frames, utt_pad_proportion=0.04529, over 9992.64 utterances.], batch size: 53, lr: 2.64e-02, grad_scale: 8.0 2023-03-07 15:47:38,633 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.969e+02 3.712e+02 4.325e+02 6.002e+02 1.306e+03, threshold=8.650e+02, percent-clipped=6.0 2023-03-07 15:47:56,448 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.0110, 2.0611, 1.8729, 1.0656, 1.2866, 1.1071, 1.3884, 1.4818], device='cuda:0'), covar=tensor([0.0624, 0.1107, 0.1438, 0.1197, 0.0932, 0.1516, 0.1068, 0.1177], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0091, 0.0089, 0.0074, 0.0076, 0.0081, 0.0091, 0.0093], device='cuda:0'), out_proj_covar=tensor([3.9449e-05, 4.9174e-05, 4.6749e-05, 4.1279e-05, 4.1145e-05, 4.3686e-05, 4.2149e-05, 4.3516e-05], device='cuda:0') 2023-03-07 15:48:19,077 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=12732.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:48:52,751 INFO [train2.py:809] (0/4) Epoch 4, batch 800, loss[ctc_loss=0.1493, att_loss=0.2552, loss=0.234, over 15515.00 frames. utt_duration=1726 frames, utt_pad_proportion=0.007731, over 36.00 utterances.], tot_loss[ctc_loss=0.1841, att_loss=0.2919, loss=0.2704, over 3211321.62 frames. utt_duration=1310 frames, utt_pad_proportion=0.03971, over 9820.13 utterances.], batch size: 36, lr: 2.63e-02, grad_scale: 8.0 2023-03-07 15:48:54,569 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5347, 2.7573, 3.4915, 2.2773, 3.6720, 4.5992, 4.2638, 3.4085], device='cuda:0'), covar=tensor([0.0355, 0.1455, 0.1068, 0.1530, 0.0916, 0.0334, 0.0428, 0.1203], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0175, 0.0162, 0.0165, 0.0179, 0.0130, 0.0123, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 15:49:08,730 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=12763.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:49:35,417 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=12780.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:50:11,370 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=12802.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:50:12,734 INFO [train2.py:809] (0/4) Epoch 4, batch 850, loss[ctc_loss=0.2023, att_loss=0.3091, loss=0.2877, over 16905.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.005466, over 49.00 utterances.], tot_loss[ctc_loss=0.1841, att_loss=0.2922, loss=0.2706, over 3226589.48 frames. utt_duration=1325 frames, utt_pad_proportion=0.03518, over 9752.91 utterances.], batch size: 49, lr: 2.63e-02, grad_scale: 8.0 2023-03-07 15:50:17,379 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.255e+02 3.893e+02 4.938e+02 6.078e+02 1.799e+03, threshold=9.877e+02, percent-clipped=7.0 2023-03-07 15:50:46,134 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=12824.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 15:51:32,706 INFO [train2.py:809] (0/4) Epoch 4, batch 900, loss[ctc_loss=0.2046, att_loss=0.3113, loss=0.29, over 17330.00 frames. utt_duration=878.9 frames, utt_pad_proportion=0.07875, over 79.00 utterances.], tot_loss[ctc_loss=0.183, att_loss=0.2916, loss=0.2699, over 3237587.38 frames. utt_duration=1324 frames, utt_pad_proportion=0.03611, over 9794.29 utterances.], batch size: 79, lr: 2.62e-02, grad_scale: 16.0 2023-03-07 15:52:45,972 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0068, 2.6568, 3.3445, 2.1150, 3.2233, 4.0411, 3.9998, 2.9613], device='cuda:0'), covar=tensor([0.0448, 0.1714, 0.1151, 0.1716, 0.1006, 0.0473, 0.0387, 0.1586], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0186, 0.0169, 0.0171, 0.0185, 0.0141, 0.0130, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 15:52:53,773 INFO [train2.py:809] (0/4) Epoch 4, batch 950, loss[ctc_loss=0.1665, att_loss=0.2927, loss=0.2675, over 16607.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.00638, over 47.00 utterances.], tot_loss[ctc_loss=0.184, att_loss=0.2921, loss=0.2705, over 3237811.14 frames. utt_duration=1302 frames, utt_pad_proportion=0.04357, over 9959.30 utterances.], batch size: 47, lr: 2.62e-02, grad_scale: 16.0 2023-03-07 15:52:57,223 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6017, 5.1844, 4.7805, 5.1754, 4.5915, 4.8965, 5.4519, 5.1202], device='cuda:0'), covar=tensor([0.0383, 0.0318, 0.0496, 0.0154, 0.0364, 0.0165, 0.0178, 0.0128], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0141, 0.0176, 0.0107, 0.0157, 0.0111, 0.0140, 0.0128], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 15:52:58,561 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.008e+02 3.388e+02 4.239e+02 5.070e+02 1.281e+03, threshold=8.479e+02, percent-clipped=6.0 2023-03-07 15:54:15,086 INFO [train2.py:809] (0/4) Epoch 4, batch 1000, loss[ctc_loss=0.1366, att_loss=0.2584, loss=0.2341, over 15945.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.006891, over 41.00 utterances.], tot_loss[ctc_loss=0.1865, att_loss=0.2934, loss=0.272, over 3242189.31 frames. utt_duration=1277 frames, utt_pad_proportion=0.05172, over 10171.30 utterances.], batch size: 41, lr: 2.61e-02, grad_scale: 8.0 2023-03-07 15:55:37,225 INFO [train2.py:809] (0/4) Epoch 4, batch 1050, loss[ctc_loss=0.2279, att_loss=0.2948, loss=0.2814, over 16268.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007422, over 43.00 utterances.], tot_loss[ctc_loss=0.1866, att_loss=0.2938, loss=0.2724, over 3255260.46 frames. utt_duration=1277 frames, utt_pad_proportion=0.04891, over 10206.08 utterances.], batch size: 43, lr: 2.61e-02, grad_scale: 8.0 2023-03-07 15:55:43,242 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.497e+02 3.995e+02 4.964e+02 6.656e+02 1.651e+03, threshold=9.928e+02, percent-clipped=9.0 2023-03-07 15:55:43,440 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2817, 5.5652, 4.9073, 5.5916, 5.2189, 4.9661, 5.1177, 4.9438], device='cuda:0'), covar=tensor([0.1288, 0.0896, 0.0851, 0.0652, 0.0596, 0.1407, 0.2157, 0.2541], device='cuda:0'), in_proj_covar=tensor([0.0300, 0.0333, 0.0275, 0.0268, 0.0247, 0.0333, 0.0369, 0.0348], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:56:58,343 INFO [train2.py:809] (0/4) Epoch 4, batch 1100, loss[ctc_loss=0.2464, att_loss=0.3266, loss=0.3106, over 13703.00 frames. utt_duration=379.5 frames, utt_pad_proportion=0.3399, over 145.00 utterances.], tot_loss[ctc_loss=0.1865, att_loss=0.2937, loss=0.2722, over 3251530.59 frames. utt_duration=1251 frames, utt_pad_proportion=0.05767, over 10412.76 utterances.], batch size: 145, lr: 2.61e-02, grad_scale: 8.0 2023-03-07 15:57:35,949 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4139, 4.5607, 4.4696, 4.5485, 1.7550, 4.4784, 2.1580, 1.8824], device='cuda:0'), covar=tensor([0.0130, 0.0120, 0.0601, 0.0241, 0.3853, 0.0165, 0.1813, 0.1842], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0089, 0.0215, 0.0109, 0.0222, 0.0095, 0.0195, 0.0185], device='cuda:0'), out_proj_covar=tensor([8.6758e-05, 8.4084e-05, 1.8131e-04, 9.3406e-05, 1.8203e-04, 8.4158e-05, 1.6296e-04, 1.5482e-04], device='cuda:0') 2023-03-07 15:58:18,912 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13102.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:58:20,177 INFO [train2.py:809] (0/4) Epoch 4, batch 1150, loss[ctc_loss=0.1717, att_loss=0.2781, loss=0.2568, over 16275.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007478, over 43.00 utterances.], tot_loss[ctc_loss=0.1873, att_loss=0.2943, loss=0.2729, over 3263310.29 frames. utt_duration=1241 frames, utt_pad_proportion=0.0569, over 10528.77 utterances.], batch size: 43, lr: 2.60e-02, grad_scale: 8.0 2023-03-07 15:58:26,389 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.150e+02 3.646e+02 4.698e+02 5.581e+02 1.233e+03, threshold=9.396e+02, percent-clipped=4.0 2023-03-07 15:58:43,121 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7993, 4.4492, 5.0144, 5.0435, 2.3046, 4.5872, 3.0966, 3.1194], device='cuda:0'), covar=tensor([0.0106, 0.0154, 0.0425, 0.0154, 0.3039, 0.0168, 0.1240, 0.1259], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0089, 0.0216, 0.0110, 0.0223, 0.0096, 0.0197, 0.0184], device='cuda:0'), out_proj_covar=tensor([8.6589e-05, 8.3852e-05, 1.8214e-04, 9.4078e-05, 1.8255e-04, 8.4960e-05, 1.6463e-04, 1.5449e-04], device='cuda:0') 2023-03-07 15:58:46,172 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=13119.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 15:59:08,215 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.3395, 3.1282, 2.4874, 1.8924, 1.7795, 2.6394, 1.4094, 1.3307], device='cuda:0'), covar=tensor([0.0917, 0.0397, 0.0536, 0.1995, 0.1126, 0.1215, 0.1844, 0.5109], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0054, 0.0058, 0.0067, 0.0057, 0.0068, 0.0065, 0.0091], device='cuda:0'), out_proj_covar=tensor([3.9571e-05, 3.5215e-05, 3.5135e-05, 4.4526e-05, 3.8252e-05, 4.4079e-05, 4.1553e-05, 6.3726e-05], device='cuda:0') 2023-03-07 15:59:35,986 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13150.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 15:59:37,777 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0793, 4.5142, 4.4458, 4.3812, 4.6199, 4.4437, 4.3324, 4.1522], device='cuda:0'), covar=tensor([0.1203, 0.0396, 0.0202, 0.0410, 0.0248, 0.0302, 0.0287, 0.0343], device='cuda:0'), in_proj_covar=tensor([0.0332, 0.0199, 0.0141, 0.0174, 0.0217, 0.0241, 0.0186, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 15:59:40,601 INFO [train2.py:809] (0/4) Epoch 4, batch 1200, loss[ctc_loss=0.1888, att_loss=0.3091, loss=0.285, over 17046.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.009347, over 52.00 utterances.], tot_loss[ctc_loss=0.1872, att_loss=0.2939, loss=0.2726, over 3256127.31 frames. utt_duration=1242 frames, utt_pad_proportion=0.05809, over 10501.27 utterances.], batch size: 52, lr: 2.60e-02, grad_scale: 8.0 2023-03-07 15:59:52,147 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1211, 4.9787, 4.9452, 3.8054, 1.9533, 2.5924, 4.9673, 3.6388], device='cuda:0'), covar=tensor([0.0537, 0.0120, 0.0135, 0.0977, 0.7310, 0.2705, 0.0143, 0.2039], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0145, 0.0165, 0.0183, 0.0384, 0.0309, 0.0151, 0.0278], device='cuda:0'), out_proj_covar=tensor([1.3913e-04, 7.3286e-05, 8.6619e-05, 8.9238e-05, 1.9507e-04, 1.5368e-04, 7.6866e-05, 1.5194e-04], device='cuda:0') 2023-03-07 16:00:06,758 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.0719, 3.2055, 1.9479, 1.5255, 1.9711, 1.5343, 1.4929, 1.5129], device='cuda:0'), covar=tensor([0.0694, 0.0280, 0.1474, 0.1079, 0.0645, 0.0902, 0.0936, 0.1419], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0078, 0.0085, 0.0071, 0.0065, 0.0071, 0.0083, 0.0086], device='cuda:0'), out_proj_covar=tensor([3.7526e-05, 4.1462e-05, 4.3302e-05, 3.9131e-05, 3.5115e-05, 3.8376e-05, 3.8926e-05, 4.0025e-05], device='cuda:0') 2023-03-07 16:00:25,121 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4406, 4.8920, 4.5994, 4.9852, 4.3924, 4.7936, 5.1712, 4.8782], device='cuda:0'), covar=tensor([0.0417, 0.0288, 0.0518, 0.0156, 0.0457, 0.0197, 0.0248, 0.0167], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0142, 0.0174, 0.0109, 0.0159, 0.0111, 0.0138, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 16:01:01,006 INFO [train2.py:809] (0/4) Epoch 4, batch 1250, loss[ctc_loss=0.1873, att_loss=0.2924, loss=0.2714, over 16401.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007525, over 44.00 utterances.], tot_loss[ctc_loss=0.1888, att_loss=0.2953, loss=0.274, over 3250660.34 frames. utt_duration=1199 frames, utt_pad_proportion=0.06951, over 10855.17 utterances.], batch size: 44, lr: 2.59e-02, grad_scale: 8.0 2023-03-07 16:01:07,189 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.553e+02 3.747e+02 4.896e+02 5.785e+02 2.345e+03, threshold=9.791e+02, percent-clipped=4.0 2023-03-07 16:01:14,537 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6081, 5.2417, 4.7963, 5.0003, 5.0795, 4.7938, 4.5843, 5.1098], device='cuda:0'), covar=tensor([0.0090, 0.0084, 0.0089, 0.0090, 0.0083, 0.0084, 0.0247, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0050, 0.0049, 0.0049, 0.0036, 0.0037, 0.0045, 0.0064, 0.0061], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:02:21,494 INFO [train2.py:809] (0/4) Epoch 4, batch 1300, loss[ctc_loss=0.1192, att_loss=0.2512, loss=0.2248, over 15753.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009766, over 38.00 utterances.], tot_loss[ctc_loss=0.1888, att_loss=0.2959, loss=0.2745, over 3255964.04 frames. utt_duration=1180 frames, utt_pad_proportion=0.07265, over 11055.16 utterances.], batch size: 38, lr: 2.59e-02, grad_scale: 8.0 2023-03-07 16:03:41,849 INFO [train2.py:809] (0/4) Epoch 4, batch 1350, loss[ctc_loss=0.2064, att_loss=0.3054, loss=0.2856, over 17037.00 frames. utt_duration=1338 frames, utt_pad_proportion=0.007709, over 51.00 utterances.], tot_loss[ctc_loss=0.1868, att_loss=0.2944, loss=0.2729, over 3258008.47 frames. utt_duration=1193 frames, utt_pad_proportion=0.07094, over 10941.30 utterances.], batch size: 51, lr: 2.58e-02, grad_scale: 8.0 2023-03-07 16:03:48,891 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.865e+02 3.582e+02 4.373e+02 5.653e+02 1.802e+03, threshold=8.745e+02, percent-clipped=3.0 2023-03-07 16:04:59,729 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-03-07 16:05:02,334 INFO [train2.py:809] (0/4) Epoch 4, batch 1400, loss[ctc_loss=0.2058, att_loss=0.3151, loss=0.2933, over 17320.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02261, over 59.00 utterances.], tot_loss[ctc_loss=0.1861, att_loss=0.2945, loss=0.2729, over 3273942.41 frames. utt_duration=1193 frames, utt_pad_proportion=0.06569, over 10986.56 utterances.], batch size: 59, lr: 2.58e-02, grad_scale: 8.0 2023-03-07 16:05:26,490 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-07 16:06:21,015 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3414, 2.4089, 4.5727, 3.5987, 3.2269, 4.0967, 3.9013, 4.1957], device='cuda:0'), covar=tensor([0.0134, 0.1655, 0.0093, 0.1153, 0.2169, 0.0402, 0.0266, 0.0329], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0233, 0.0125, 0.0289, 0.0336, 0.0182, 0.0112, 0.0128], device='cuda:0'), out_proj_covar=tensor([1.0191e-04, 1.7898e-04, 1.0205e-04, 2.2835e-04, 2.5630e-04, 1.4915e-04, 9.3651e-05, 1.0884e-04], device='cuda:0') 2023-03-07 16:06:22,110 INFO [train2.py:809] (0/4) Epoch 4, batch 1450, loss[ctc_loss=0.1952, att_loss=0.3033, loss=0.2817, over 16911.00 frames. utt_duration=684.8 frames, utt_pad_proportion=0.1407, over 99.00 utterances.], tot_loss[ctc_loss=0.1877, att_loss=0.2954, loss=0.2739, over 3271543.39 frames. utt_duration=1175 frames, utt_pad_proportion=0.07261, over 11148.40 utterances.], batch size: 99, lr: 2.58e-02, grad_scale: 8.0 2023-03-07 16:06:28,372 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.701e+02 3.999e+02 4.753e+02 5.960e+02 1.214e+03, threshold=9.507e+02, percent-clipped=6.0 2023-03-07 16:06:33,574 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9772, 2.5211, 4.0813, 3.4313, 3.1102, 3.7157, 3.5285, 3.7960], device='cuda:0'), covar=tensor([0.0126, 0.1250, 0.0114, 0.0921, 0.1821, 0.0342, 0.0203, 0.0219], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0234, 0.0125, 0.0290, 0.0339, 0.0183, 0.0113, 0.0129], device='cuda:0'), out_proj_covar=tensor([1.0202e-04, 1.7978e-04, 1.0263e-04, 2.2923e-04, 2.5815e-04, 1.4990e-04, 9.4216e-05, 1.0948e-04], device='cuda:0') 2023-03-07 16:06:47,413 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=13419.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:07:33,424 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-03-07 16:07:42,522 INFO [train2.py:809] (0/4) Epoch 4, batch 1500, loss[ctc_loss=0.1637, att_loss=0.2871, loss=0.2625, over 16535.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006617, over 45.00 utterances.], tot_loss[ctc_loss=0.1868, att_loss=0.2951, loss=0.2734, over 3272477.39 frames. utt_duration=1180 frames, utt_pad_proportion=0.07185, over 11106.09 utterances.], batch size: 45, lr: 2.57e-02, grad_scale: 8.0 2023-03-07 16:08:04,755 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=13467.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:08:36,219 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.47 vs. limit=5.0 2023-03-07 16:09:03,078 INFO [train2.py:809] (0/4) Epoch 4, batch 1550, loss[ctc_loss=0.1853, att_loss=0.3131, loss=0.2876, over 17403.00 frames. utt_duration=1182 frames, utt_pad_proportion=0.01865, over 59.00 utterances.], tot_loss[ctc_loss=0.1889, att_loss=0.2964, loss=0.2749, over 3269660.28 frames. utt_duration=1159 frames, utt_pad_proportion=0.07756, over 11300.36 utterances.], batch size: 59, lr: 2.57e-02, grad_scale: 8.0 2023-03-07 16:09:09,261 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.195e+02 3.814e+02 4.608e+02 5.940e+02 1.228e+03, threshold=9.215e+02, percent-clipped=5.0 2023-03-07 16:10:23,312 INFO [train2.py:809] (0/4) Epoch 4, batch 1600, loss[ctc_loss=0.3421, att_loss=0.3687, loss=0.3634, over 13614.00 frames. utt_duration=376.8 frames, utt_pad_proportion=0.3458, over 145.00 utterances.], tot_loss[ctc_loss=0.188, att_loss=0.295, loss=0.2736, over 3258979.01 frames. utt_duration=1160 frames, utt_pad_proportion=0.08051, over 11256.09 utterances.], batch size: 145, lr: 2.56e-02, grad_scale: 8.0 2023-03-07 16:11:01,620 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-03-07 16:11:42,753 INFO [train2.py:809] (0/4) Epoch 4, batch 1650, loss[ctc_loss=0.1418, att_loss=0.2605, loss=0.2368, over 10568.00 frames. utt_duration=1839 frames, utt_pad_proportion=0.2349, over 23.00 utterances.], tot_loss[ctc_loss=0.186, att_loss=0.2929, loss=0.2715, over 3245841.77 frames. utt_duration=1192 frames, utt_pad_proportion=0.07464, over 10907.57 utterances.], batch size: 23, lr: 2.56e-02, grad_scale: 8.0 2023-03-07 16:11:48,979 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 3.919e+02 5.311e+02 6.584e+02 2.912e+03, threshold=1.062e+03, percent-clipped=12.0 2023-03-07 16:13:01,508 INFO [train2.py:809] (0/4) Epoch 4, batch 1700, loss[ctc_loss=0.1371, att_loss=0.2546, loss=0.2311, over 15999.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007884, over 40.00 utterances.], tot_loss[ctc_loss=0.1854, att_loss=0.2929, loss=0.2714, over 3254934.92 frames. utt_duration=1215 frames, utt_pad_proportion=0.06716, over 10726.03 utterances.], batch size: 40, lr: 2.55e-02, grad_scale: 8.0 2023-03-07 16:13:32,690 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7446, 1.9412, 2.9467, 4.3647, 4.1668, 4.5062, 2.7592, 1.7486], device='cuda:0'), covar=tensor([0.0417, 0.3134, 0.1416, 0.0485, 0.0313, 0.0095, 0.1635, 0.3001], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0187, 0.0177, 0.0131, 0.0116, 0.0102, 0.0180, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:14:20,724 INFO [train2.py:809] (0/4) Epoch 4, batch 1750, loss[ctc_loss=0.1731, att_loss=0.2972, loss=0.2724, over 16876.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.006964, over 49.00 utterances.], tot_loss[ctc_loss=0.1844, att_loss=0.2925, loss=0.2709, over 3262113.88 frames. utt_duration=1210 frames, utt_pad_proportion=0.06619, over 10801.64 utterances.], batch size: 49, lr: 2.55e-02, grad_scale: 8.0 2023-03-07 16:14:27,020 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.465e+02 3.709e+02 4.336e+02 5.273e+02 1.516e+03, threshold=8.672e+02, percent-clipped=2.0 2023-03-07 16:15:15,170 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=13737.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:15:23,326 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-07 16:15:39,471 INFO [train2.py:809] (0/4) Epoch 4, batch 1800, loss[ctc_loss=0.1364, att_loss=0.273, loss=0.2457, over 16420.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006588, over 44.00 utterances.], tot_loss[ctc_loss=0.1843, att_loss=0.293, loss=0.2713, over 3270975.39 frames. utt_duration=1225 frames, utt_pad_proportion=0.06025, over 10691.64 utterances.], batch size: 44, lr: 2.55e-02, grad_scale: 8.0 2023-03-07 16:15:59,998 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7750, 4.6885, 4.5220, 4.7571, 5.1235, 4.9260, 4.3475, 1.9134], device='cuda:0'), covar=tensor([0.0281, 0.0277, 0.0225, 0.0198, 0.0878, 0.0209, 0.0468, 0.3542], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0116, 0.0110, 0.0113, 0.0246, 0.0129, 0.0103, 0.0251], device='cuda:0'), out_proj_covar=tensor([1.2119e-04, 1.0335e-04, 9.9445e-05, 1.0718e-04, 2.2737e-04, 1.1868e-04, 9.8505e-05, 2.2395e-04], device='cuda:0') 2023-03-07 16:16:16,839 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3716, 2.0439, 2.8693, 4.0602, 3.8978, 3.7966, 2.6716, 1.4078], device='cuda:0'), covar=tensor([0.0577, 0.2913, 0.1384, 0.0656, 0.0554, 0.0341, 0.1812, 0.3284], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0183, 0.0175, 0.0132, 0.0117, 0.0104, 0.0177, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:16:51,933 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=13798.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 16:16:59,429 INFO [train2.py:809] (0/4) Epoch 4, batch 1850, loss[ctc_loss=0.1711, att_loss=0.2859, loss=0.263, over 16115.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006361, over 42.00 utterances.], tot_loss[ctc_loss=0.1855, att_loss=0.2944, loss=0.2726, over 3271760.59 frames. utt_duration=1216 frames, utt_pad_proportion=0.06194, over 10774.00 utterances.], batch size: 42, lr: 2.54e-02, grad_scale: 8.0 2023-03-07 16:17:05,503 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.641e+02 3.804e+02 4.653e+02 6.004e+02 1.107e+03, threshold=9.305e+02, percent-clipped=7.0 2023-03-07 16:18:19,889 INFO [train2.py:809] (0/4) Epoch 4, batch 1900, loss[ctc_loss=0.2005, att_loss=0.2933, loss=0.2747, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006858, over 43.00 utterances.], tot_loss[ctc_loss=0.1846, att_loss=0.294, loss=0.2721, over 3271418.26 frames. utt_duration=1217 frames, utt_pad_proportion=0.06203, over 10769.23 utterances.], batch size: 43, lr: 2.54e-02, grad_scale: 8.0 2023-03-07 16:19:39,444 INFO [train2.py:809] (0/4) Epoch 4, batch 1950, loss[ctc_loss=0.2027, att_loss=0.2879, loss=0.2709, over 16171.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.00745, over 41.00 utterances.], tot_loss[ctc_loss=0.1856, att_loss=0.2943, loss=0.2726, over 3274935.00 frames. utt_duration=1208 frames, utt_pad_proportion=0.06317, over 10857.22 utterances.], batch size: 41, lr: 2.53e-02, grad_scale: 8.0 2023-03-07 16:19:45,531 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.392e+02 3.636e+02 4.382e+02 5.668e+02 1.213e+03, threshold=8.764e+02, percent-clipped=2.0 2023-03-07 16:20:38,793 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.23 vs. limit=2.0 2023-03-07 16:20:44,995 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8772, 5.3103, 4.9553, 5.0831, 5.4086, 5.2811, 5.0082, 4.8698], device='cuda:0'), covar=tensor([0.1135, 0.0358, 0.0240, 0.0496, 0.0210, 0.0253, 0.0278, 0.0325], device='cuda:0'), in_proj_covar=tensor([0.0332, 0.0197, 0.0139, 0.0177, 0.0218, 0.0238, 0.0187, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 16:21:00,025 INFO [train2.py:809] (0/4) Epoch 4, batch 2000, loss[ctc_loss=0.1371, att_loss=0.2689, loss=0.2426, over 15955.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006949, over 41.00 utterances.], tot_loss[ctc_loss=0.1847, att_loss=0.2935, loss=0.2717, over 3266175.08 frames. utt_duration=1205 frames, utt_pad_proportion=0.06622, over 10859.17 utterances.], batch size: 41, lr: 2.53e-02, grad_scale: 8.0 2023-03-07 16:22:15,586 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-14000.pt 2023-03-07 16:22:23,899 INFO [train2.py:809] (0/4) Epoch 4, batch 2050, loss[ctc_loss=0.1886, att_loss=0.3045, loss=0.2813, over 16267.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.00798, over 43.00 utterances.], tot_loss[ctc_loss=0.1842, att_loss=0.2936, loss=0.2717, over 3274461.38 frames. utt_duration=1211 frames, utt_pad_proportion=0.06174, over 10827.64 utterances.], batch size: 43, lr: 2.53e-02, grad_scale: 8.0 2023-03-07 16:22:30,198 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.271e+02 3.689e+02 4.518e+02 6.015e+02 1.676e+03, threshold=9.035e+02, percent-clipped=8.0 2023-03-07 16:23:23,908 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3412, 4.9051, 4.5630, 4.5986, 4.8379, 4.6857, 4.2918, 4.8914], device='cuda:0'), covar=tensor([0.0107, 0.0117, 0.0104, 0.0114, 0.0106, 0.0080, 0.0309, 0.0190], device='cuda:0'), in_proj_covar=tensor([0.0050, 0.0049, 0.0050, 0.0037, 0.0037, 0.0045, 0.0067, 0.0063], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:23:43,386 INFO [train2.py:809] (0/4) Epoch 4, batch 2100, loss[ctc_loss=0.1802, att_loss=0.2997, loss=0.2758, over 16957.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007091, over 50.00 utterances.], tot_loss[ctc_loss=0.1834, att_loss=0.293, loss=0.2711, over 3271633.18 frames. utt_duration=1239 frames, utt_pad_proportion=0.05524, over 10576.17 utterances.], batch size: 50, lr: 2.52e-02, grad_scale: 8.0 2023-03-07 16:24:14,205 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14072.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:24:47,827 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14093.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 16:24:47,970 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7092, 4.0721, 3.3730, 3.5716, 4.0314, 3.7280, 2.3203, 4.5792], device='cuda:0'), covar=tensor([0.1098, 0.0324, 0.1090, 0.0694, 0.0440, 0.0679, 0.1154, 0.0227], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0113, 0.0168, 0.0133, 0.0133, 0.0162, 0.0138, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:24:55,769 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6016, 2.8260, 2.1017, 1.1649, 1.5197, 2.0327, 1.3351, 2.2643], device='cuda:0'), covar=tensor([0.1090, 0.1254, 0.2366, 0.3117, 0.2158, 0.2095, 0.1982, 0.1302], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0088, 0.0096, 0.0080, 0.0080, 0.0080, 0.0086, 0.0088], device='cuda:0'), out_proj_covar=tensor([3.9211e-05, 4.4606e-05, 4.8320e-05, 4.3392e-05, 4.0756e-05, 4.3645e-05, 4.0556e-05, 4.2414e-05], device='cuda:0') 2023-03-07 16:25:03,076 INFO [train2.py:809] (0/4) Epoch 4, batch 2150, loss[ctc_loss=0.1596, att_loss=0.2727, loss=0.2501, over 16466.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006, over 46.00 utterances.], tot_loss[ctc_loss=0.1811, att_loss=0.2917, loss=0.2695, over 3270827.34 frames. utt_duration=1263 frames, utt_pad_proportion=0.0508, over 10375.31 utterances.], batch size: 46, lr: 2.52e-02, grad_scale: 8.0 2023-03-07 16:25:03,414 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14103.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:25:09,252 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.138e+02 3.723e+02 4.436e+02 6.346e+02 3.042e+03, threshold=8.872e+02, percent-clipped=8.0 2023-03-07 16:25:15,639 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7536, 2.1468, 1.8837, 0.9050, 1.4544, 1.5062, 1.0446, 2.0479], device='cuda:0'), covar=tensor([0.0539, 0.0903, 0.1676, 0.1588, 0.0973, 0.1224, 0.1138, 0.0663], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0090, 0.0098, 0.0081, 0.0082, 0.0082, 0.0088, 0.0089], device='cuda:0'), out_proj_covar=tensor([3.9730e-05, 4.5483e-05, 4.9557e-05, 4.3932e-05, 4.1257e-05, 4.4543e-05, 4.1284e-05, 4.2870e-05], device='cuda:0') 2023-03-07 16:25:30,156 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14120.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 16:25:51,844 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14133.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:26:23,574 INFO [train2.py:809] (0/4) Epoch 4, batch 2200, loss[ctc_loss=0.1456, att_loss=0.281, loss=0.2539, over 16331.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006293, over 45.00 utterances.], tot_loss[ctc_loss=0.1805, att_loss=0.2907, loss=0.2687, over 3265384.23 frames. utt_duration=1258 frames, utt_pad_proportion=0.0551, over 10393.53 utterances.], batch size: 45, lr: 2.51e-02, grad_scale: 8.0 2023-03-07 16:26:41,879 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14164.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:27:09,521 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14181.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 16:27:44,781 INFO [train2.py:809] (0/4) Epoch 4, batch 2250, loss[ctc_loss=0.1611, att_loss=0.2968, loss=0.2697, over 17052.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.00796, over 52.00 utterances.], tot_loss[ctc_loss=0.1798, att_loss=0.29, loss=0.268, over 3258084.93 frames. utt_duration=1254 frames, utt_pad_proportion=0.05783, over 10408.80 utterances.], batch size: 52, lr: 2.51e-02, grad_scale: 8.0 2023-03-07 16:27:50,862 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.200e+02 3.354e+02 4.208e+02 5.126e+02 1.091e+03, threshold=8.417e+02, percent-clipped=2.0 2023-03-07 16:27:54,364 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3194, 2.3865, 4.8232, 3.6271, 3.1522, 4.2924, 4.3704, 4.4923], device='cuda:0'), covar=tensor([0.0137, 0.1799, 0.0096, 0.1171, 0.2303, 0.0302, 0.0170, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0233, 0.0121, 0.0286, 0.0330, 0.0178, 0.0108, 0.0130], device='cuda:0'), out_proj_covar=tensor([1.0607e-04, 1.7910e-04, 9.9414e-05, 2.2798e-04, 2.5493e-04, 1.4532e-04, 9.1648e-05, 1.0900e-04], device='cuda:0') 2023-03-07 16:28:44,455 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7098, 4.5955, 4.3067, 4.7997, 2.1505, 4.6370, 2.2777, 1.9503], device='cuda:0'), covar=tensor([0.0103, 0.0133, 0.0959, 0.0155, 0.3359, 0.0141, 0.1837, 0.1886], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0093, 0.0229, 0.0111, 0.0230, 0.0102, 0.0207, 0.0196], device='cuda:0'), out_proj_covar=tensor([9.1689e-05, 8.8965e-05, 1.9594e-04, 9.6753e-05, 1.9182e-04, 9.3171e-05, 1.7428e-04, 1.6505e-04], device='cuda:0') 2023-03-07 16:29:04,198 INFO [train2.py:809] (0/4) Epoch 4, batch 2300, loss[ctc_loss=0.1582, att_loss=0.2731, loss=0.2501, over 14504.00 frames. utt_duration=1815 frames, utt_pad_proportion=0.03011, over 32.00 utterances.], tot_loss[ctc_loss=0.1813, att_loss=0.2914, loss=0.2694, over 3259756.42 frames. utt_duration=1219 frames, utt_pad_proportion=0.06545, over 10713.32 utterances.], batch size: 32, lr: 2.51e-02, grad_scale: 8.0 2023-03-07 16:30:00,746 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-03-07 16:30:12,167 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7417, 5.0437, 4.8828, 4.8665, 5.2575, 5.2200, 4.9326, 4.8784], device='cuda:0'), covar=tensor([0.1254, 0.0466, 0.0247, 0.0574, 0.0309, 0.0258, 0.0248, 0.0274], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0210, 0.0142, 0.0186, 0.0229, 0.0246, 0.0191, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 16:30:24,652 INFO [train2.py:809] (0/4) Epoch 4, batch 2350, loss[ctc_loss=0.186, att_loss=0.2905, loss=0.2696, over 17440.00 frames. utt_duration=706 frames, utt_pad_proportion=0.1153, over 99.00 utterances.], tot_loss[ctc_loss=0.1812, att_loss=0.2918, loss=0.2697, over 3263773.54 frames. utt_duration=1189 frames, utt_pad_proportion=0.0705, over 10998.36 utterances.], batch size: 99, lr: 2.50e-02, grad_scale: 8.0 2023-03-07 16:30:31,150 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.697e+02 3.860e+02 4.546e+02 5.797e+02 1.491e+03, threshold=9.092e+02, percent-clipped=6.0 2023-03-07 16:31:44,323 INFO [train2.py:809] (0/4) Epoch 4, batch 2400, loss[ctc_loss=0.1876, att_loss=0.3118, loss=0.287, over 17307.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01072, over 55.00 utterances.], tot_loss[ctc_loss=0.182, att_loss=0.2922, loss=0.2702, over 3269651.22 frames. utt_duration=1212 frames, utt_pad_proportion=0.06392, over 10804.62 utterances.], batch size: 55, lr: 2.50e-02, grad_scale: 8.0 2023-03-07 16:32:48,075 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14393.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:33:03,735 INFO [train2.py:809] (0/4) Epoch 4, batch 2450, loss[ctc_loss=0.1505, att_loss=0.2589, loss=0.2372, over 15764.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009075, over 38.00 utterances.], tot_loss[ctc_loss=0.1814, att_loss=0.2915, loss=0.2695, over 3268586.95 frames. utt_duration=1234 frames, utt_pad_proportion=0.06012, over 10608.31 utterances.], batch size: 38, lr: 2.49e-02, grad_scale: 8.0 2023-03-07 16:33:09,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.088e+02 3.936e+02 4.992e+02 6.745e+02 1.784e+03, threshold=9.985e+02, percent-clipped=4.0 2023-03-07 16:33:20,219 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14413.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:33:36,204 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5055, 4.9754, 4.5489, 5.1085, 4.4835, 4.7695, 5.1924, 4.9710], device='cuda:0'), covar=tensor([0.0308, 0.0238, 0.0589, 0.0112, 0.0384, 0.0184, 0.0191, 0.0132], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0150, 0.0188, 0.0117, 0.0168, 0.0118, 0.0147, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 16:33:44,299 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14428.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:34:04,968 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14441.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:34:23,285 INFO [train2.py:809] (0/4) Epoch 4, batch 2500, loss[ctc_loss=0.1542, att_loss=0.2675, loss=0.2448, over 15985.00 frames. utt_duration=1600 frames, utt_pad_proportion=0.008612, over 40.00 utterances.], tot_loss[ctc_loss=0.1799, att_loss=0.2906, loss=0.2685, over 3266500.09 frames. utt_duration=1245 frames, utt_pad_proportion=0.05744, over 10510.50 utterances.], batch size: 40, lr: 2.49e-02, grad_scale: 8.0 2023-03-07 16:34:28,828 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4280, 3.7451, 3.0680, 3.3828, 3.8716, 3.4842, 2.5471, 4.4384], device='cuda:0'), covar=tensor([0.1365, 0.0342, 0.1310, 0.0608, 0.0465, 0.0631, 0.1063, 0.0220], device='cuda:0'), in_proj_covar=tensor([0.0142, 0.0113, 0.0170, 0.0134, 0.0135, 0.0162, 0.0145, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:34:33,331 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14459.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:34:57,621 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14474.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:35:00,385 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14476.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 16:35:17,976 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14487.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:35:42,399 INFO [train2.py:809] (0/4) Epoch 4, batch 2550, loss[ctc_loss=0.2539, att_loss=0.3307, loss=0.3153, over 17194.00 frames. utt_duration=872.1 frames, utt_pad_proportion=0.08683, over 79.00 utterances.], tot_loss[ctc_loss=0.1803, att_loss=0.2908, loss=0.2687, over 3273756.21 frames. utt_duration=1247 frames, utt_pad_proportion=0.05455, over 10511.74 utterances.], batch size: 79, lr: 2.49e-02, grad_scale: 8.0 2023-03-07 16:35:48,916 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.665e+02 3.493e+02 4.541e+02 5.557e+02 1.503e+03, threshold=9.083e+02, percent-clipped=2.0 2023-03-07 16:36:54,442 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14548.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:37:02,528 INFO [train2.py:809] (0/4) Epoch 4, batch 2600, loss[ctc_loss=0.1534, att_loss=0.261, loss=0.2395, over 15944.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007249, over 41.00 utterances.], tot_loss[ctc_loss=0.1804, att_loss=0.2908, loss=0.2687, over 3280549.22 frames. utt_duration=1241 frames, utt_pad_proportion=0.05358, over 10582.63 utterances.], batch size: 41, lr: 2.48e-02, grad_scale: 8.0 2023-03-07 16:37:04,376 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0240, 4.8867, 4.8303, 2.9890, 4.8770, 4.3282, 3.9917, 2.4643], device='cuda:0'), covar=tensor([0.0150, 0.0088, 0.0209, 0.0879, 0.0083, 0.0158, 0.0311, 0.1517], device='cuda:0'), in_proj_covar=tensor([0.0048, 0.0054, 0.0045, 0.0086, 0.0051, 0.0063, 0.0072, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 16:38:21,821 INFO [train2.py:809] (0/4) Epoch 4, batch 2650, loss[ctc_loss=0.1452, att_loss=0.2671, loss=0.2427, over 16016.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006691, over 40.00 utterances.], tot_loss[ctc_loss=0.1804, att_loss=0.2918, loss=0.2695, over 3294001.34 frames. utt_duration=1244 frames, utt_pad_proportion=0.04874, over 10602.49 utterances.], batch size: 40, lr: 2.48e-02, grad_scale: 8.0 2023-03-07 16:38:28,429 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.304e+02 3.793e+02 4.783e+02 6.009e+02 1.165e+03, threshold=9.567e+02, percent-clipped=2.0 2023-03-07 16:38:58,119 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2537, 1.6847, 2.1055, 1.7754, 1.6370, 0.9357, 1.0736, 1.4102], device='cuda:0'), covar=tensor([0.0283, 0.1455, 0.1492, 0.1011, 0.1122, 0.1619, 0.1347, 0.1134], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0090, 0.0094, 0.0080, 0.0081, 0.0079, 0.0091, 0.0092], device='cuda:0'), out_proj_covar=tensor([3.8564e-05, 4.6196e-05, 4.8049e-05, 4.4168e-05, 4.1997e-05, 4.3633e-05, 4.3009e-05, 4.4654e-05], device='cuda:0') 2023-03-07 16:39:41,968 INFO [train2.py:809] (0/4) Epoch 4, batch 2700, loss[ctc_loss=0.1621, att_loss=0.2953, loss=0.2687, over 17019.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.007875, over 51.00 utterances.], tot_loss[ctc_loss=0.1793, att_loss=0.2905, loss=0.2682, over 3276634.59 frames. utt_duration=1260 frames, utt_pad_proportion=0.04829, over 10418.33 utterances.], batch size: 51, lr: 2.48e-02, grad_scale: 8.0 2023-03-07 16:40:37,759 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3545, 2.1811, 2.9959, 2.3761, 3.0101, 3.3803, 3.2963, 2.6407], device='cuda:0'), covar=tensor([0.0405, 0.1688, 0.1140, 0.1178, 0.0740, 0.0627, 0.0559, 0.1193], device='cuda:0'), in_proj_covar=tensor([0.0184, 0.0196, 0.0184, 0.0173, 0.0189, 0.0162, 0.0142, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:41:01,184 INFO [train2.py:809] (0/4) Epoch 4, batch 2750, loss[ctc_loss=0.1261, att_loss=0.2507, loss=0.2258, over 15502.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008023, over 36.00 utterances.], tot_loss[ctc_loss=0.1807, att_loss=0.2915, loss=0.2693, over 3276817.56 frames. utt_duration=1213 frames, utt_pad_proportion=0.05938, over 10814.98 utterances.], batch size: 36, lr: 2.47e-02, grad_scale: 8.0 2023-03-07 16:41:02,570 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-07 16:41:07,374 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.333e+02 3.898e+02 4.978e+02 5.848e+02 1.042e+03, threshold=9.955e+02, percent-clipped=2.0 2023-03-07 16:41:08,335 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-03-07 16:41:42,290 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14728.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:41:57,544 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6932, 2.3295, 3.0591, 4.2405, 4.0744, 4.3177, 2.7634, 1.8557], device='cuda:0'), covar=tensor([0.0548, 0.2759, 0.1439, 0.0558, 0.0364, 0.0246, 0.1854, 0.2871], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0188, 0.0184, 0.0140, 0.0123, 0.0107, 0.0186, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:42:02,324 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=14740.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:42:13,377 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.8868, 2.5178, 2.8348, 2.4858, 2.7130, 3.1783, 2.4875, 2.0273], device='cuda:0'), covar=tensor([0.1412, 0.1261, 0.1157, 0.2830, 0.1483, 0.2169, 0.1201, 0.7243], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0060, 0.0060, 0.0075, 0.0060, 0.0074, 0.0062, 0.0095], device='cuda:0'), out_proj_covar=tensor([4.1683e-05, 3.7573e-05, 3.7894e-05, 5.0362e-05, 3.9524e-05, 5.2070e-05, 4.2162e-05, 6.9197e-05], device='cuda:0') 2023-03-07 16:42:23,125 INFO [train2.py:809] (0/4) Epoch 4, batch 2800, loss[ctc_loss=0.1492, att_loss=0.2914, loss=0.2629, over 16325.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006445, over 45.00 utterances.], tot_loss[ctc_loss=0.1787, att_loss=0.2906, loss=0.2682, over 3281384.90 frames. utt_duration=1225 frames, utt_pad_proportion=0.05536, over 10723.85 utterances.], batch size: 45, lr: 2.47e-02, grad_scale: 8.0 2023-03-07 16:42:32,819 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14759.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:42:48,957 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14769.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:43:00,440 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14776.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:43:00,661 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=14776.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 16:43:23,474 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4265, 4.9435, 4.5977, 5.0698, 5.0657, 4.7173, 4.3608, 4.8407], device='cuda:0'), covar=tensor([0.0095, 0.0128, 0.0101, 0.0069, 0.0081, 0.0100, 0.0350, 0.0190], device='cuda:0'), in_proj_covar=tensor([0.0048, 0.0047, 0.0049, 0.0036, 0.0036, 0.0044, 0.0064, 0.0062], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0001, 0.0001, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 16:43:41,664 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=14801.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:43:44,570 INFO [train2.py:809] (0/4) Epoch 4, batch 2850, loss[ctc_loss=0.2183, att_loss=0.3133, loss=0.2943, over 16864.00 frames. utt_duration=682.9 frames, utt_pad_proportion=0.1442, over 99.00 utterances.], tot_loss[ctc_loss=0.1791, att_loss=0.2905, loss=0.2682, over 3267364.48 frames. utt_duration=1193 frames, utt_pad_proportion=0.06856, over 10964.56 utterances.], batch size: 99, lr: 2.46e-02, grad_scale: 8.0 2023-03-07 16:43:50,724 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.067e+02 3.258e+02 4.280e+02 5.701e+02 1.148e+03, threshold=8.559e+02, percent-clipped=5.0 2023-03-07 16:43:50,870 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14807.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:44:18,601 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=14824.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 16:44:48,677 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=14843.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:45:04,287 INFO [train2.py:809] (0/4) Epoch 4, batch 2900, loss[ctc_loss=0.1422, att_loss=0.2675, loss=0.2424, over 16331.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006126, over 45.00 utterances.], tot_loss[ctc_loss=0.1793, att_loss=0.2902, loss=0.2681, over 3263595.49 frames. utt_duration=1210 frames, utt_pad_proportion=0.06573, over 10804.86 utterances.], batch size: 45, lr: 2.46e-02, grad_scale: 8.0 2023-03-07 16:46:15,629 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8672, 6.0116, 5.3622, 5.9468, 5.6688, 5.4066, 5.5045, 5.3067], device='cuda:0'), covar=tensor([0.0896, 0.0704, 0.0714, 0.0650, 0.0598, 0.1150, 0.2061, 0.2004], device='cuda:0'), in_proj_covar=tensor([0.0306, 0.0332, 0.0278, 0.0269, 0.0253, 0.0344, 0.0379, 0.0351], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 16:46:25,732 INFO [train2.py:809] (0/4) Epoch 4, batch 2950, loss[ctc_loss=0.1583, att_loss=0.2538, loss=0.2347, over 15618.00 frames. utt_duration=1690 frames, utt_pad_proportion=0.01046, over 37.00 utterances.], tot_loss[ctc_loss=0.1791, att_loss=0.2903, loss=0.2681, over 3274733.81 frames. utt_duration=1212 frames, utt_pad_proportion=0.06198, over 10822.95 utterances.], batch size: 37, lr: 2.46e-02, grad_scale: 8.0 2023-03-07 16:46:32,080 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.413e+02 3.902e+02 4.882e+02 6.670e+02 1.203e+03, threshold=9.763e+02, percent-clipped=9.0 2023-03-07 16:47:46,086 INFO [train2.py:809] (0/4) Epoch 4, batch 3000, loss[ctc_loss=0.1059, att_loss=0.2358, loss=0.2098, over 15372.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01047, over 35.00 utterances.], tot_loss[ctc_loss=0.178, att_loss=0.2894, loss=0.2671, over 3267309.47 frames. utt_duration=1210 frames, utt_pad_proportion=0.06495, over 10810.58 utterances.], batch size: 35, lr: 2.45e-02, grad_scale: 16.0 2023-03-07 16:47:46,088 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 16:48:00,419 INFO [train2.py:843] (0/4) Epoch 4, validation: ctc_loss=0.08782, att_loss=0.2564, loss=0.2227, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 16:48:00,419 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 16:49:20,424 INFO [train2.py:809] (0/4) Epoch 4, batch 3050, loss[ctc_loss=0.1612, att_loss=0.2661, loss=0.2451, over 16417.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006679, over 44.00 utterances.], tot_loss[ctc_loss=0.1783, att_loss=0.2897, loss=0.2674, over 3272098.69 frames. utt_duration=1204 frames, utt_pad_proportion=0.0664, over 10885.99 utterances.], batch size: 44, lr: 2.45e-02, grad_scale: 16.0 2023-03-07 16:49:26,544 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.237e+02 3.805e+02 4.579e+02 6.197e+02 1.095e+03, threshold=9.158e+02, percent-clipped=2.0 2023-03-07 16:50:19,382 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0124, 5.1241, 5.4707, 5.6790, 5.2302, 5.7979, 5.0043, 5.9986], device='cuda:0'), covar=tensor([0.0563, 0.0624, 0.0460, 0.0602, 0.1821, 0.0855, 0.0415, 0.0499], device='cuda:0'), in_proj_covar=tensor([0.0465, 0.0294, 0.0297, 0.0353, 0.0513, 0.0301, 0.0247, 0.0316], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0002, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 16:50:40,675 INFO [train2.py:809] (0/4) Epoch 4, batch 3100, loss[ctc_loss=0.221, att_loss=0.3178, loss=0.2984, over 14248.00 frames. utt_duration=394.3 frames, utt_pad_proportion=0.3154, over 145.00 utterances.], tot_loss[ctc_loss=0.1775, att_loss=0.2892, loss=0.2669, over 3271279.74 frames. utt_duration=1221 frames, utt_pad_proportion=0.06336, over 10727.08 utterances.], batch size: 145, lr: 2.45e-02, grad_scale: 16.0 2023-03-07 16:51:06,270 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15069.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:51:24,936 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-03-07 16:51:50,416 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15096.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:52:01,875 INFO [train2.py:809] (0/4) Epoch 4, batch 3150, loss[ctc_loss=0.1814, att_loss=0.2911, loss=0.2692, over 17479.00 frames. utt_duration=707.8 frames, utt_pad_proportion=0.1119, over 99.00 utterances.], tot_loss[ctc_loss=0.1753, att_loss=0.2878, loss=0.2653, over 3262834.82 frames. utt_duration=1246 frames, utt_pad_proportion=0.05738, over 10486.86 utterances.], batch size: 99, lr: 2.44e-02, grad_scale: 16.0 2023-03-07 16:52:07,995 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.047e+02 3.829e+02 4.318e+02 5.478e+02 1.412e+03, threshold=8.637e+02, percent-clipped=4.0 2023-03-07 16:52:23,549 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15117.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:53:05,042 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15143.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:53:17,086 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9462, 6.0779, 5.4675, 5.9433, 5.7705, 5.5762, 5.6697, 5.2984], device='cuda:0'), covar=tensor([0.1181, 0.0750, 0.0743, 0.0807, 0.0628, 0.1213, 0.2004, 0.2306], device='cuda:0'), in_proj_covar=tensor([0.0306, 0.0343, 0.0279, 0.0269, 0.0256, 0.0339, 0.0374, 0.0350], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 16:53:21,411 INFO [train2.py:809] (0/4) Epoch 4, batch 3200, loss[ctc_loss=0.1683, att_loss=0.2614, loss=0.2427, over 14520.00 frames. utt_duration=1817 frames, utt_pad_proportion=0.04737, over 32.00 utterances.], tot_loss[ctc_loss=0.1756, att_loss=0.2876, loss=0.2652, over 3260130.74 frames. utt_duration=1245 frames, utt_pad_proportion=0.05882, over 10491.15 utterances.], batch size: 32, lr: 2.44e-02, grad_scale: 16.0 2023-03-07 16:53:46,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-03-07 16:54:21,462 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15191.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:54:28,170 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15195.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 16:54:42,169 INFO [train2.py:809] (0/4) Epoch 4, batch 3250, loss[ctc_loss=0.1442, att_loss=0.2688, loss=0.2439, over 16122.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.005883, over 42.00 utterances.], tot_loss[ctc_loss=0.1745, att_loss=0.2864, loss=0.264, over 3249227.02 frames. utt_duration=1252 frames, utt_pad_proportion=0.06044, over 10389.36 utterances.], batch size: 42, lr: 2.44e-02, grad_scale: 16.0 2023-03-07 16:54:48,522 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.387e+02 3.575e+02 4.506e+02 5.643e+02 1.593e+03, threshold=9.012e+02, percent-clipped=6.0 2023-03-07 16:54:59,833 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2023-03-07 16:56:01,887 INFO [train2.py:809] (0/4) Epoch 4, batch 3300, loss[ctc_loss=0.1735, att_loss=0.2912, loss=0.2677, over 17363.00 frames. utt_duration=880.7 frames, utt_pad_proportion=0.07584, over 79.00 utterances.], tot_loss[ctc_loss=0.1757, att_loss=0.2877, loss=0.2653, over 3253042.73 frames. utt_duration=1247 frames, utt_pad_proportion=0.06018, over 10449.32 utterances.], batch size: 79, lr: 2.43e-02, grad_scale: 16.0 2023-03-07 16:56:06,941 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15256.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 16:56:31,230 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-03-07 16:57:22,461 INFO [train2.py:809] (0/4) Epoch 4, batch 3350, loss[ctc_loss=0.1787, att_loss=0.3031, loss=0.2782, over 17162.00 frames. utt_duration=1227 frames, utt_pad_proportion=0.01273, over 56.00 utterances.], tot_loss[ctc_loss=0.1764, att_loss=0.2883, loss=0.266, over 3253125.22 frames. utt_duration=1220 frames, utt_pad_proportion=0.06695, over 10680.67 utterances.], batch size: 56, lr: 2.43e-02, grad_scale: 16.0 2023-03-07 16:57:28,539 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.491e+02 3.835e+02 4.773e+02 5.805e+02 1.964e+03, threshold=9.546e+02, percent-clipped=8.0 2023-03-07 16:58:42,390 INFO [train2.py:809] (0/4) Epoch 4, batch 3400, loss[ctc_loss=0.168, att_loss=0.2819, loss=0.2591, over 16544.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005957, over 45.00 utterances.], tot_loss[ctc_loss=0.175, att_loss=0.2878, loss=0.2652, over 3261526.61 frames. utt_duration=1240 frames, utt_pad_proportion=0.06045, over 10536.83 utterances.], batch size: 45, lr: 2.42e-02, grad_scale: 16.0 2023-03-07 16:59:40,653 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-03-07 16:59:50,435 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15396.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:00:02,646 INFO [train2.py:809] (0/4) Epoch 4, batch 3450, loss[ctc_loss=0.1384, att_loss=0.2524, loss=0.2296, over 16180.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006289, over 41.00 utterances.], tot_loss[ctc_loss=0.1734, att_loss=0.287, loss=0.2642, over 3265876.33 frames. utt_duration=1240 frames, utt_pad_proportion=0.05988, over 10544.62 utterances.], batch size: 41, lr: 2.42e-02, grad_scale: 8.0 2023-03-07 17:00:10,266 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.330e+02 3.236e+02 3.892e+02 5.378e+02 1.574e+03, threshold=7.784e+02, percent-clipped=3.0 2023-03-07 17:00:29,338 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9081, 4.6035, 4.7701, 3.0898, 4.6809, 4.0437, 3.8349, 2.7511], device='cuda:0'), covar=tensor([0.0096, 0.0079, 0.0209, 0.0766, 0.0077, 0.0213, 0.0307, 0.1238], device='cuda:0'), in_proj_covar=tensor([0.0049, 0.0054, 0.0045, 0.0087, 0.0052, 0.0064, 0.0073, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-07 17:00:43,778 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-03-07 17:01:07,803 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15444.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:01:22,738 INFO [train2.py:809] (0/4) Epoch 4, batch 3500, loss[ctc_loss=0.1506, att_loss=0.2802, loss=0.2543, over 16489.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006172, over 46.00 utterances.], tot_loss[ctc_loss=0.1751, att_loss=0.2877, loss=0.2652, over 3267266.78 frames. utt_duration=1238 frames, utt_pad_proportion=0.05857, over 10565.93 utterances.], batch size: 46, lr: 2.42e-02, grad_scale: 8.0 2023-03-07 17:02:43,141 INFO [train2.py:809] (0/4) Epoch 4, batch 3550, loss[ctc_loss=0.1571, att_loss=0.2716, loss=0.2487, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007447, over 43.00 utterances.], tot_loss[ctc_loss=0.1759, att_loss=0.2881, loss=0.2656, over 3261972.60 frames. utt_duration=1210 frames, utt_pad_proportion=0.06557, over 10793.80 utterances.], batch size: 43, lr: 2.41e-02, grad_scale: 8.0 2023-03-07 17:02:48,857 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-07 17:02:50,734 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.279e+02 3.812e+02 4.413e+02 5.065e+02 1.151e+03, threshold=8.826e+02, percent-clipped=3.0 2023-03-07 17:04:00,618 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=15551.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 17:04:03,541 INFO [train2.py:809] (0/4) Epoch 4, batch 3600, loss[ctc_loss=0.1263, att_loss=0.2621, loss=0.2349, over 16266.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.006847, over 43.00 utterances.], tot_loss[ctc_loss=0.1758, att_loss=0.2882, loss=0.2657, over 3258307.36 frames. utt_duration=1212 frames, utt_pad_proportion=0.06644, over 10765.31 utterances.], batch size: 43, lr: 2.41e-02, grad_scale: 8.0 2023-03-07 17:05:24,926 INFO [train2.py:809] (0/4) Epoch 4, batch 3650, loss[ctc_loss=0.176, att_loss=0.2903, loss=0.2674, over 16874.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007899, over 49.00 utterances.], tot_loss[ctc_loss=0.1734, att_loss=0.2873, loss=0.2645, over 3270641.21 frames. utt_duration=1233 frames, utt_pad_proportion=0.05815, over 10619.11 utterances.], batch size: 49, lr: 2.41e-02, grad_scale: 8.0 2023-03-07 17:05:32,895 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.227e+02 3.822e+02 4.406e+02 5.174e+02 1.169e+03, threshold=8.811e+02, percent-clipped=2.0 2023-03-07 17:06:23,675 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4027, 4.8216, 4.3158, 4.8889, 4.2451, 4.5489, 4.9271, 4.8151], device='cuda:0'), covar=tensor([0.0341, 0.0236, 0.0743, 0.0143, 0.0428, 0.0228, 0.0284, 0.0169], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0160, 0.0209, 0.0129, 0.0175, 0.0128, 0.0154, 0.0144], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0005, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-07 17:06:44,838 INFO [train2.py:809] (0/4) Epoch 4, batch 3700, loss[ctc_loss=0.2554, att_loss=0.3347, loss=0.3188, over 14376.00 frames. utt_duration=395.3 frames, utt_pad_proportion=0.3101, over 146.00 utterances.], tot_loss[ctc_loss=0.1733, att_loss=0.287, loss=0.2642, over 3267018.89 frames. utt_duration=1226 frames, utt_pad_proportion=0.06136, over 10674.80 utterances.], batch size: 146, lr: 2.40e-02, grad_scale: 8.0 2023-03-07 17:07:50,328 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1209, 5.2433, 5.4882, 5.6073, 5.3323, 5.9940, 5.1341, 6.0224], device='cuda:0'), covar=tensor([0.0505, 0.0777, 0.0557, 0.0754, 0.1899, 0.0674, 0.0459, 0.0553], device='cuda:0'), in_proj_covar=tensor([0.0465, 0.0305, 0.0316, 0.0371, 0.0530, 0.0313, 0.0261, 0.0323], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 17:08:05,619 INFO [train2.py:809] (0/4) Epoch 4, batch 3750, loss[ctc_loss=0.1651, att_loss=0.2667, loss=0.2464, over 15660.00 frames. utt_duration=1695 frames, utt_pad_proportion=0.007157, over 37.00 utterances.], tot_loss[ctc_loss=0.1731, att_loss=0.2866, loss=0.2639, over 3264776.53 frames. utt_duration=1230 frames, utt_pad_proportion=0.06089, over 10627.42 utterances.], batch size: 37, lr: 2.40e-02, grad_scale: 8.0 2023-03-07 17:08:13,131 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.429e+02 3.564e+02 4.340e+02 5.643e+02 1.443e+03, threshold=8.680e+02, percent-clipped=5.0 2023-03-07 17:08:36,411 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6185, 1.2966, 2.0177, 1.3548, 2.7590, 1.4174, 1.1100, 2.7858], device='cuda:0'), covar=tensor([0.0997, 0.2699, 0.2525, 0.1847, 0.0319, 0.1750, 0.2124, 0.0761], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0091, 0.0092, 0.0079, 0.0074, 0.0078, 0.0089, 0.0082], device='cuda:0'), out_proj_covar=tensor([3.8301e-05, 4.6371e-05, 4.8591e-05, 4.4201e-05, 3.7577e-05, 4.1431e-05, 4.4396e-05, 4.2281e-05], device='cuda:0') 2023-03-07 17:09:25,260 INFO [train2.py:809] (0/4) Epoch 4, batch 3800, loss[ctc_loss=0.1628, att_loss=0.2891, loss=0.2638, over 16768.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006407, over 48.00 utterances.], tot_loss[ctc_loss=0.1747, att_loss=0.2876, loss=0.265, over 3270435.97 frames. utt_duration=1228 frames, utt_pad_proportion=0.06014, over 10663.56 utterances.], batch size: 48, lr: 2.40e-02, grad_scale: 8.0 2023-03-07 17:09:40,154 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.42 vs. limit=5.0 2023-03-07 17:10:46,099 INFO [train2.py:809] (0/4) Epoch 4, batch 3850, loss[ctc_loss=0.1908, att_loss=0.274, loss=0.2573, over 11854.00 frames. utt_duration=1826 frames, utt_pad_proportion=0.1633, over 26.00 utterances.], tot_loss[ctc_loss=0.1749, att_loss=0.2881, loss=0.2655, over 3269863.51 frames. utt_duration=1226 frames, utt_pad_proportion=0.06027, over 10678.56 utterances.], batch size: 26, lr: 2.39e-02, grad_scale: 8.0 2023-03-07 17:10:53,742 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.413e+02 3.550e+02 4.549e+02 5.409e+02 1.209e+03, threshold=9.099e+02, percent-clipped=5.0 2023-03-07 17:11:36,588 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-07 17:12:00,201 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=15851.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:12:02,982 INFO [train2.py:809] (0/4) Epoch 4, batch 3900, loss[ctc_loss=0.2397, att_loss=0.3258, loss=0.3086, over 13894.00 frames. utt_duration=385 frames, utt_pad_proportion=0.3316, over 145.00 utterances.], tot_loss[ctc_loss=0.1743, att_loss=0.2874, loss=0.2648, over 3258768.19 frames. utt_duration=1208 frames, utt_pad_proportion=0.06889, over 10801.06 utterances.], batch size: 145, lr: 2.39e-02, grad_scale: 8.0 2023-03-07 17:12:29,950 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-07 17:12:49,882 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=15883.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:13:13,702 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=15899.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:13:13,867 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8689, 5.2520, 4.7718, 5.3204, 4.6590, 5.0859, 5.4555, 5.2285], device='cuda:0'), covar=tensor([0.0323, 0.0240, 0.0605, 0.0136, 0.0391, 0.0137, 0.0182, 0.0130], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0157, 0.0202, 0.0131, 0.0173, 0.0126, 0.0149, 0.0145], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-07 17:13:19,709 INFO [train2.py:809] (0/4) Epoch 4, batch 3950, loss[ctc_loss=0.139, att_loss=0.2441, loss=0.223, over 15900.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.00835, over 39.00 utterances.], tot_loss[ctc_loss=0.1732, att_loss=0.286, loss=0.2634, over 3260053.90 frames. utt_duration=1232 frames, utt_pad_proportion=0.0628, over 10601.00 utterances.], batch size: 39, lr: 2.39e-02, grad_scale: 8.0 2023-03-07 17:13:27,825 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.270e+02 3.629e+02 4.356e+02 5.723e+02 1.612e+03, threshold=8.711e+02, percent-clipped=3.0 2023-03-07 17:13:50,963 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.1764, 3.7806, 3.0842, 3.2411, 3.7878, 3.4392, 2.6434, 4.4377], device='cuda:0'), covar=tensor([0.1401, 0.0361, 0.1187, 0.0762, 0.0506, 0.0751, 0.0951, 0.0286], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0121, 0.0171, 0.0138, 0.0147, 0.0161, 0.0145, 0.0122], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 17:14:10,050 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-4.pt 2023-03-07 17:14:39,243 INFO [train2.py:809] (0/4) Epoch 5, batch 0, loss[ctc_loss=0.1736, att_loss=0.2903, loss=0.267, over 17309.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01083, over 55.00 utterances.], tot_loss[ctc_loss=0.1736, att_loss=0.2903, loss=0.267, over 17309.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01083, over 55.00 utterances.], batch size: 55, lr: 2.22e-02, grad_scale: 8.0 2023-03-07 17:14:39,246 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 17:14:51,958 INFO [train2.py:843] (0/4) Epoch 5, validation: ctc_loss=0.08303, att_loss=0.2543, loss=0.22, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 17:14:51,959 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 17:15:06,148 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=15944.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:16:13,044 INFO [train2.py:809] (0/4) Epoch 5, batch 50, loss[ctc_loss=0.1382, att_loss=0.272, loss=0.2452, over 16124.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006587, over 42.00 utterances.], tot_loss[ctc_loss=0.1691, att_loss=0.2878, loss=0.264, over 742644.57 frames. utt_duration=1232 frames, utt_pad_proportion=0.04869, over 2413.68 utterances.], batch size: 42, lr: 2.22e-02, grad_scale: 8.0 2023-03-07 17:16:35,811 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-16000.pt 2023-03-07 17:16:41,374 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16001.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:16:51,838 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.119e+02 3.417e+02 4.421e+02 5.634e+02 1.181e+03, threshold=8.842e+02, percent-clipped=4.0 2023-03-07 17:17:36,497 INFO [train2.py:809] (0/4) Epoch 5, batch 100, loss[ctc_loss=0.1813, att_loss=0.3044, loss=0.2798, over 17316.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02289, over 59.00 utterances.], tot_loss[ctc_loss=0.169, att_loss=0.2855, loss=0.2622, over 1309439.01 frames. utt_duration=1297 frames, utt_pad_proportion=0.03427, over 4043.59 utterances.], batch size: 59, lr: 2.21e-02, grad_scale: 8.0 2023-03-07 17:18:18,119 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16062.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:18:19,602 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3163, 3.6601, 2.9765, 3.3045, 3.8106, 3.4968, 2.3230, 4.4652], device='cuda:0'), covar=tensor([0.1191, 0.0394, 0.1099, 0.0670, 0.0479, 0.0624, 0.1180, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0121, 0.0173, 0.0140, 0.0147, 0.0162, 0.0145, 0.0124], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 17:18:55,093 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16085.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:18:56,281 INFO [train2.py:809] (0/4) Epoch 5, batch 150, loss[ctc_loss=0.1616, att_loss=0.293, loss=0.2667, over 16642.00 frames. utt_duration=1418 frames, utt_pad_proportion=0.004363, over 47.00 utterances.], tot_loss[ctc_loss=0.1671, att_loss=0.2848, loss=0.2613, over 1748542.22 frames. utt_duration=1303 frames, utt_pad_proportion=0.0355, over 5373.41 utterances.], batch size: 47, lr: 2.21e-02, grad_scale: 8.0 2023-03-07 17:19:31,235 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.071e+02 3.158e+02 4.095e+02 5.546e+02 1.058e+03, threshold=8.189e+02, percent-clipped=5.0 2023-03-07 17:19:57,737 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6301, 5.8971, 5.1403, 5.8458, 5.5149, 5.2129, 5.2470, 5.1755], device='cuda:0'), covar=tensor([0.1048, 0.0733, 0.0819, 0.0532, 0.0621, 0.1128, 0.1792, 0.1751], device='cuda:0'), in_proj_covar=tensor([0.0325, 0.0373, 0.0292, 0.0281, 0.0265, 0.0354, 0.0393, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 17:20:16,133 INFO [train2.py:809] (0/4) Epoch 5, batch 200, loss[ctc_loss=0.1979, att_loss=0.3013, loss=0.2806, over 16878.00 frames. utt_duration=690.4 frames, utt_pad_proportion=0.1348, over 98.00 utterances.], tot_loss[ctc_loss=0.1691, att_loss=0.2851, loss=0.2619, over 2083641.56 frames. utt_duration=1255 frames, utt_pad_proportion=0.05094, over 6651.49 utterances.], batch size: 98, lr: 2.21e-02, grad_scale: 8.0 2023-03-07 17:20:29,306 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6813, 5.9300, 5.2113, 5.8918, 5.5332, 5.2976, 5.3187, 5.1486], device='cuda:0'), covar=tensor([0.1093, 0.0791, 0.0770, 0.0526, 0.0626, 0.1046, 0.2017, 0.2023], device='cuda:0'), in_proj_covar=tensor([0.0325, 0.0372, 0.0294, 0.0281, 0.0266, 0.0353, 0.0394, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 17:20:32,638 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16146.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:20:37,370 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16149.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:21:02,222 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7335, 3.9067, 3.1493, 3.6814, 3.9065, 3.7563, 2.7900, 4.6327], device='cuda:0'), covar=tensor([0.1023, 0.0341, 0.1076, 0.0530, 0.0486, 0.0602, 0.0940, 0.0408], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0119, 0.0171, 0.0140, 0.0146, 0.0163, 0.0145, 0.0126], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 17:21:35,464 INFO [train2.py:809] (0/4) Epoch 5, batch 250, loss[ctc_loss=0.1456, att_loss=0.26, loss=0.2372, over 15996.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.00744, over 40.00 utterances.], tot_loss[ctc_loss=0.1707, att_loss=0.2867, loss=0.2635, over 2345268.93 frames. utt_duration=1209 frames, utt_pad_proportion=0.06339, over 7767.55 utterances.], batch size: 40, lr: 2.20e-02, grad_scale: 8.0 2023-03-07 17:22:01,863 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-07 17:22:11,996 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.315e+02 3.409e+02 3.964e+02 4.827e+02 7.562e+02, threshold=7.928e+02, percent-clipped=0.0 2023-03-07 17:22:15,455 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16210.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:22:57,404 INFO [train2.py:809] (0/4) Epoch 5, batch 300, loss[ctc_loss=0.1559, att_loss=0.2827, loss=0.2574, over 16339.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005472, over 45.00 utterances.], tot_loss[ctc_loss=0.1695, att_loss=0.2858, loss=0.2625, over 2560640.74 frames. utt_duration=1210 frames, utt_pad_proportion=0.05918, over 8474.98 utterances.], batch size: 45, lr: 2.20e-02, grad_scale: 8.0 2023-03-07 17:23:02,216 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16239.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:24:18,582 INFO [train2.py:809] (0/4) Epoch 5, batch 350, loss[ctc_loss=0.135, att_loss=0.2575, loss=0.233, over 16418.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.005973, over 44.00 utterances.], tot_loss[ctc_loss=0.1685, att_loss=0.2851, loss=0.2618, over 2713350.26 frames. utt_duration=1200 frames, utt_pad_proportion=0.0635, over 9059.10 utterances.], batch size: 44, lr: 2.20e-02, grad_scale: 8.0 2023-03-07 17:24:56,573 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.150e+02 3.423e+02 4.365e+02 5.343e+02 1.409e+03, threshold=8.730e+02, percent-clipped=9.0 2023-03-07 17:25:41,308 INFO [train2.py:809] (0/4) Epoch 5, batch 400, loss[ctc_loss=0.1143, att_loss=0.2635, loss=0.2337, over 16405.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006699, over 44.00 utterances.], tot_loss[ctc_loss=0.1675, att_loss=0.2843, loss=0.261, over 2844773.83 frames. utt_duration=1238 frames, utt_pad_proportion=0.05287, over 9205.64 utterances.], batch size: 44, lr: 2.20e-02, grad_scale: 8.0 2023-03-07 17:26:15,244 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16357.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:26:37,581 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16371.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:27:02,335 INFO [train2.py:809] (0/4) Epoch 5, batch 450, loss[ctc_loss=0.1729, att_loss=0.2939, loss=0.2697, over 16462.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.007573, over 46.00 utterances.], tot_loss[ctc_loss=0.1677, att_loss=0.2837, loss=0.2605, over 2938521.08 frames. utt_duration=1251 frames, utt_pad_proportion=0.0512, over 9406.60 utterances.], batch size: 46, lr: 2.19e-02, grad_scale: 8.0 2023-03-07 17:27:40,287 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.521e+02 3.844e+02 4.538e+02 5.770e+02 9.247e+02, threshold=9.077e+02, percent-clipped=1.0 2023-03-07 17:27:48,472 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4079, 2.6675, 3.5322, 2.4329, 3.3357, 4.5596, 4.3259, 3.1791], device='cuda:0'), covar=tensor([0.0430, 0.1845, 0.1082, 0.1820, 0.1254, 0.0471, 0.0453, 0.1644], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0194, 0.0188, 0.0184, 0.0202, 0.0179, 0.0151, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 17:28:17,585 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16432.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:28:24,061 INFO [train2.py:809] (0/4) Epoch 5, batch 500, loss[ctc_loss=0.1653, att_loss=0.2703, loss=0.2493, over 15999.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007853, over 40.00 utterances.], tot_loss[ctc_loss=0.1663, att_loss=0.2829, loss=0.2596, over 3017233.26 frames. utt_duration=1288 frames, utt_pad_proportion=0.04267, over 9377.83 utterances.], batch size: 40, lr: 2.19e-02, grad_scale: 8.0 2023-03-07 17:28:32,492 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16441.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:28:49,428 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.18 vs. limit=5.0 2023-03-07 17:29:39,249 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-03-07 17:29:45,247 INFO [train2.py:809] (0/4) Epoch 5, batch 550, loss[ctc_loss=0.1349, att_loss=0.2592, loss=0.2343, over 16964.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007655, over 50.00 utterances.], tot_loss[ctc_loss=0.1663, att_loss=0.2831, loss=0.2598, over 3075289.62 frames. utt_duration=1274 frames, utt_pad_proportion=0.04617, over 9667.53 utterances.], batch size: 50, lr: 2.19e-02, grad_scale: 8.0 2023-03-07 17:30:09,267 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3826, 4.2246, 4.6951, 4.6838, 2.3450, 4.9263, 2.5360, 2.6345], device='cuda:0'), covar=tensor([0.0179, 0.0211, 0.0722, 0.0315, 0.2838, 0.0169, 0.2050, 0.1578], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0095, 0.0239, 0.0112, 0.0226, 0.0094, 0.0218, 0.0197], device='cuda:0'), out_proj_covar=tensor([9.5325e-05, 9.3032e-05, 2.0738e-04, 9.7777e-05, 1.9587e-04, 8.7956e-05, 1.8610e-04, 1.7003e-04], device='cuda:0') 2023-03-07 17:30:16,758 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16505.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:30:22,784 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.254e+02 3.300e+02 4.154e+02 5.468e+02 1.131e+03, threshold=8.309e+02, percent-clipped=7.0 2023-03-07 17:31:06,244 INFO [train2.py:809] (0/4) Epoch 5, batch 600, loss[ctc_loss=0.1595, att_loss=0.283, loss=0.2583, over 16400.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.00754, over 44.00 utterances.], tot_loss[ctc_loss=0.1651, att_loss=0.2824, loss=0.259, over 3122975.66 frames. utt_duration=1300 frames, utt_pad_proportion=0.03971, over 9620.13 utterances.], batch size: 44, lr: 2.18e-02, grad_scale: 8.0 2023-03-07 17:31:12,032 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16539.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:32:27,518 INFO [train2.py:809] (0/4) Epoch 5, batch 650, loss[ctc_loss=0.138, att_loss=0.2632, loss=0.2381, over 16263.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.007696, over 43.00 utterances.], tot_loss[ctc_loss=0.1636, att_loss=0.2813, loss=0.2578, over 3156251.69 frames. utt_duration=1312 frames, utt_pad_proportion=0.03827, over 9632.38 utterances.], batch size: 43, lr: 2.18e-02, grad_scale: 8.0 2023-03-07 17:32:29,904 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16587.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:33:07,235 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.993e+02 3.372e+02 4.098e+02 5.133e+02 8.733e+02, threshold=8.196e+02, percent-clipped=2.0 2023-03-07 17:33:33,157 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8877, 4.1815, 4.2639, 4.4061, 2.0133, 4.5900, 2.5683, 2.4381], device='cuda:0'), covar=tensor([0.0251, 0.0149, 0.0717, 0.0252, 0.2921, 0.0165, 0.1774, 0.1659], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0096, 0.0239, 0.0110, 0.0226, 0.0094, 0.0218, 0.0201], device='cuda:0'), out_proj_covar=tensor([9.7100e-05, 9.3641e-05, 2.0781e-04, 9.7654e-05, 1.9595e-04, 8.9076e-05, 1.8710e-04, 1.7255e-04], device='cuda:0') 2023-03-07 17:33:51,404 INFO [train2.py:809] (0/4) Epoch 5, batch 700, loss[ctc_loss=0.1697, att_loss=0.2957, loss=0.2705, over 17438.00 frames. utt_duration=1109 frames, utt_pad_proportion=0.03089, over 63.00 utterances.], tot_loss[ctc_loss=0.1633, att_loss=0.2815, loss=0.2579, over 3191480.48 frames. utt_duration=1291 frames, utt_pad_proportion=0.04027, over 9897.22 utterances.], batch size: 63, lr: 2.18e-02, grad_scale: 8.0 2023-03-07 17:34:26,843 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16657.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:35:14,451 INFO [train2.py:809] (0/4) Epoch 5, batch 750, loss[ctc_loss=0.1592, att_loss=0.2844, loss=0.2594, over 16326.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006369, over 45.00 utterances.], tot_loss[ctc_loss=0.1633, att_loss=0.2814, loss=0.2578, over 3212618.10 frames. utt_duration=1279 frames, utt_pad_proportion=0.04269, over 10056.76 utterances.], batch size: 45, lr: 2.17e-02, grad_scale: 8.0 2023-03-07 17:35:45,783 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16705.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:35:51,953 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.143e+02 3.355e+02 3.924e+02 5.306e+02 2.112e+03, threshold=7.847e+02, percent-clipped=6.0 2023-03-07 17:36:21,539 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=16727.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:36:36,912 INFO [train2.py:809] (0/4) Epoch 5, batch 800, loss[ctc_loss=0.1753, att_loss=0.2936, loss=0.27, over 16876.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007111, over 49.00 utterances.], tot_loss[ctc_loss=0.1644, att_loss=0.282, loss=0.2585, over 3231911.32 frames. utt_duration=1275 frames, utt_pad_proportion=0.04286, over 10147.23 utterances.], batch size: 49, lr: 2.17e-02, grad_scale: 8.0 2023-03-07 17:36:45,951 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16741.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:37:59,756 INFO [train2.py:809] (0/4) Epoch 5, batch 850, loss[ctc_loss=0.1523, att_loss=0.2819, loss=0.256, over 17062.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009253, over 53.00 utterances.], tot_loss[ctc_loss=0.1658, att_loss=0.2827, loss=0.2593, over 3244180.32 frames. utt_duration=1258 frames, utt_pad_proportion=0.04691, over 10327.01 utterances.], batch size: 53, lr: 2.17e-02, grad_scale: 8.0 2023-03-07 17:38:04,492 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16789.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:38:30,534 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=16805.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:38:36,319 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 3.471e+02 4.444e+02 5.932e+02 1.311e+03, threshold=8.889e+02, percent-clipped=13.0 2023-03-07 17:39:22,342 INFO [train2.py:809] (0/4) Epoch 5, batch 900, loss[ctc_loss=0.1794, att_loss=0.2961, loss=0.2728, over 16975.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.00636, over 50.00 utterances.], tot_loss[ctc_loss=0.1653, att_loss=0.282, loss=0.2587, over 3244990.36 frames. utt_duration=1254 frames, utt_pad_proportion=0.05067, over 10367.29 utterances.], batch size: 50, lr: 2.16e-02, grad_scale: 8.0 2023-03-07 17:39:48,823 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=16853.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:40:13,277 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16868.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 17:40:42,556 INFO [train2.py:809] (0/4) Epoch 5, batch 950, loss[ctc_loss=0.1781, att_loss=0.2783, loss=0.2583, over 15628.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009858, over 37.00 utterances.], tot_loss[ctc_loss=0.1661, att_loss=0.2825, loss=0.2592, over 3246205.70 frames. utt_duration=1244 frames, utt_pad_proportion=0.05566, over 10454.64 utterances.], batch size: 37, lr: 2.16e-02, grad_scale: 8.0 2023-03-07 17:41:14,652 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=16906.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:41:18,939 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.139e+02 3.605e+02 4.438e+02 5.826e+02 1.823e+03, threshold=8.877e+02, percent-clipped=6.0 2023-03-07 17:41:52,655 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16929.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 17:42:04,888 INFO [train2.py:809] (0/4) Epoch 5, batch 1000, loss[ctc_loss=0.1643, att_loss=0.2822, loss=0.2586, over 17413.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.04348, over 69.00 utterances.], tot_loss[ctc_loss=0.1659, att_loss=0.2826, loss=0.2593, over 3241308.83 frames. utt_duration=1238 frames, utt_pad_proportion=0.06026, over 10481.25 utterances.], batch size: 69, lr: 2.16e-02, grad_scale: 8.0 2023-03-07 17:42:54,630 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=16967.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 17:43:25,865 INFO [train2.py:809] (0/4) Epoch 5, batch 1050, loss[ctc_loss=0.1355, att_loss=0.2521, loss=0.2288, over 16114.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006987, over 42.00 utterances.], tot_loss[ctc_loss=0.1652, att_loss=0.2826, loss=0.2592, over 3245780.44 frames. utt_duration=1237 frames, utt_pad_proportion=0.06012, over 10509.22 utterances.], batch size: 42, lr: 2.16e-02, grad_scale: 8.0 2023-03-07 17:44:02,520 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.326e+02 3.274e+02 4.192e+02 5.406e+02 1.067e+03, threshold=8.384e+02, percent-clipped=5.0 2023-03-07 17:44:31,996 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17027.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:44:47,445 INFO [train2.py:809] (0/4) Epoch 5, batch 1100, loss[ctc_loss=0.1824, att_loss=0.2946, loss=0.2722, over 17402.00 frames. utt_duration=882.6 frames, utt_pad_proportion=0.0758, over 79.00 utterances.], tot_loss[ctc_loss=0.1647, att_loss=0.2822, loss=0.2587, over 3253599.84 frames. utt_duration=1238 frames, utt_pad_proportion=0.05917, over 10526.93 utterances.], batch size: 79, lr: 2.15e-02, grad_scale: 8.0 2023-03-07 17:45:50,220 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17075.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:46:07,600 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-03-07 17:46:09,757 INFO [train2.py:809] (0/4) Epoch 5, batch 1150, loss[ctc_loss=0.1363, att_loss=0.2499, loss=0.2272, over 15406.00 frames. utt_duration=1762 frames, utt_pad_proportion=0.008425, over 35.00 utterances.], tot_loss[ctc_loss=0.1649, att_loss=0.2829, loss=0.2593, over 3263998.88 frames. utt_duration=1242 frames, utt_pad_proportion=0.05586, over 10525.12 utterances.], batch size: 35, lr: 2.15e-02, grad_scale: 8.0 2023-03-07 17:46:29,302 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-03-07 17:46:45,378 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.659e+02 3.410e+02 4.345e+02 5.323e+02 1.056e+03, threshold=8.690e+02, percent-clipped=4.0 2023-03-07 17:47:11,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9722, 5.4287, 4.8457, 5.4622, 4.8573, 5.0801, 5.5855, 5.3338], device='cuda:0'), covar=tensor([0.0303, 0.0210, 0.0594, 0.0129, 0.0383, 0.0126, 0.0201, 0.0111], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0159, 0.0212, 0.0133, 0.0183, 0.0129, 0.0154, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0005, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-07 17:47:31,222 INFO [train2.py:809] (0/4) Epoch 5, batch 1200, loss[ctc_loss=0.1771, att_loss=0.295, loss=0.2714, over 17122.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01403, over 56.00 utterances.], tot_loss[ctc_loss=0.1661, att_loss=0.2843, loss=0.2606, over 3274264.57 frames. utt_duration=1218 frames, utt_pad_proportion=0.0598, over 10769.64 utterances.], batch size: 56, lr: 2.15e-02, grad_scale: 8.0 2023-03-07 17:47:50,414 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0123, 6.1866, 5.5989, 6.1209, 5.9050, 5.6244, 5.7696, 5.4538], device='cuda:0'), covar=tensor([0.1006, 0.0623, 0.0754, 0.0588, 0.0561, 0.1149, 0.1678, 0.1932], device='cuda:0'), in_proj_covar=tensor([0.0329, 0.0356, 0.0295, 0.0287, 0.0269, 0.0352, 0.0390, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 17:48:19,576 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0013, 4.6088, 4.5846, 4.7606, 5.3262, 5.0870, 4.8131, 2.2388], device='cuda:0'), covar=tensor([0.0216, 0.0456, 0.0321, 0.0262, 0.0867, 0.0195, 0.0260, 0.2929], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0122, 0.0116, 0.0124, 0.0276, 0.0132, 0.0112, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 17:48:40,389 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7313, 2.2500, 5.0501, 3.7706, 3.0095, 4.3213, 4.5259, 4.6953], device='cuda:0'), covar=tensor([0.0142, 0.2077, 0.0098, 0.1271, 0.2387, 0.0276, 0.0159, 0.0234], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0241, 0.0121, 0.0300, 0.0319, 0.0183, 0.0108, 0.0132], device='cuda:0'), out_proj_covar=tensor([1.1135e-04, 1.9041e-04, 1.0184e-04, 2.3825e-04, 2.5425e-04, 1.5521e-04, 9.2940e-05, 1.1480e-04], device='cuda:0') 2023-03-07 17:48:52,439 INFO [train2.py:809] (0/4) Epoch 5, batch 1250, loss[ctc_loss=0.1875, att_loss=0.2851, loss=0.2656, over 16295.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006366, over 43.00 utterances.], tot_loss[ctc_loss=0.1662, att_loss=0.2835, loss=0.26, over 3271729.45 frames. utt_duration=1244 frames, utt_pad_proportion=0.05319, over 10535.82 utterances.], batch size: 43, lr: 2.14e-02, grad_scale: 8.0 2023-03-07 17:48:57,281 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17189.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:49:29,493 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.117e+02 3.900e+02 4.861e+02 6.296e+02 1.076e+03, threshold=9.721e+02, percent-clipped=3.0 2023-03-07 17:49:36,188 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.6317, 2.5434, 3.0691, 2.7204, 3.0264, 3.1189, 2.5727, 1.3712], device='cuda:0'), covar=tensor([0.1578, 0.1833, 0.1383, 0.2804, 0.2576, 0.3201, 0.1538, 1.0728], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0066, 0.0070, 0.0085, 0.0066, 0.0082, 0.0064, 0.0109], device='cuda:0'), out_proj_covar=tensor([4.8598e-05, 4.3512e-05, 4.6216e-05, 5.9454e-05, 4.6278e-05, 6.0876e-05, 4.6287e-05, 7.9673e-05], device='cuda:0') 2023-03-07 17:49:55,862 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17224.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 17:50:14,611 INFO [train2.py:809] (0/4) Epoch 5, batch 1300, loss[ctc_loss=0.1294, att_loss=0.2429, loss=0.2202, over 14548.00 frames. utt_duration=1820 frames, utt_pad_proportion=0.03915, over 32.00 utterances.], tot_loss[ctc_loss=0.1657, att_loss=0.2837, loss=0.2601, over 3276899.51 frames. utt_duration=1235 frames, utt_pad_proportion=0.05414, over 10624.01 utterances.], batch size: 32, lr: 2.14e-02, grad_scale: 8.0 2023-03-07 17:50:36,827 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17250.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:50:56,136 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17262.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 17:51:15,825 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9601, 4.5474, 4.6222, 4.6255, 5.2412, 4.9677, 4.5759, 2.1030], device='cuda:0'), covar=tensor([0.0188, 0.0524, 0.0273, 0.0358, 0.0693, 0.0183, 0.0387, 0.3127], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0123, 0.0115, 0.0126, 0.0276, 0.0132, 0.0113, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 17:51:33,064 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2602, 4.1975, 4.0734, 4.4199, 4.4665, 4.3259, 4.0059, 2.0208], device='cuda:0'), covar=tensor([0.0292, 0.0344, 0.0288, 0.0103, 0.1170, 0.0249, 0.0385, 0.3106], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0124, 0.0115, 0.0127, 0.0280, 0.0134, 0.0114, 0.0250], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 17:51:35,795 INFO [train2.py:809] (0/4) Epoch 5, batch 1350, loss[ctc_loss=0.1474, att_loss=0.2864, loss=0.2586, over 16888.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006288, over 49.00 utterances.], tot_loss[ctc_loss=0.1641, att_loss=0.2822, loss=0.2586, over 3277318.22 frames. utt_duration=1271 frames, utt_pad_proportion=0.04674, over 10328.03 utterances.], batch size: 49, lr: 2.14e-02, grad_scale: 8.0 2023-03-07 17:51:37,611 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8138, 5.2237, 5.1403, 5.0566, 5.3163, 5.2095, 4.9553, 4.7620], device='cuda:0'), covar=tensor([0.0982, 0.0412, 0.0209, 0.0387, 0.0216, 0.0252, 0.0250, 0.0331], device='cuda:0'), in_proj_covar=tensor([0.0376, 0.0226, 0.0158, 0.0199, 0.0241, 0.0272, 0.0208, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 17:52:12,510 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 3.130e+02 4.012e+02 5.410e+02 1.134e+03, threshold=8.024e+02, percent-clipped=3.0 2023-03-07 17:52:57,028 INFO [train2.py:809] (0/4) Epoch 5, batch 1400, loss[ctc_loss=0.1409, att_loss=0.2676, loss=0.2423, over 16007.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.006959, over 40.00 utterances.], tot_loss[ctc_loss=0.1629, att_loss=0.2811, loss=0.2574, over 3272930.81 frames. utt_duration=1263 frames, utt_pad_proportion=0.05104, over 10375.80 utterances.], batch size: 40, lr: 2.14e-02, grad_scale: 8.0 2023-03-07 17:53:09,983 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1671, 4.7385, 4.3837, 4.4352, 4.7785, 4.3243, 3.9293, 4.5362], device='cuda:0'), covar=tensor([0.0128, 0.0142, 0.0115, 0.0131, 0.0072, 0.0115, 0.0390, 0.0261], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0052, 0.0056, 0.0039, 0.0039, 0.0049, 0.0071, 0.0068], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 17:53:26,831 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3863, 5.0949, 5.0826, 2.8267, 1.9624, 2.6001, 4.8874, 3.7752], device='cuda:0'), covar=tensor([0.0410, 0.0146, 0.0169, 0.2551, 0.6365, 0.2575, 0.0201, 0.1844], device='cuda:0'), in_proj_covar=tensor([0.0265, 0.0166, 0.0193, 0.0182, 0.0366, 0.0327, 0.0175, 0.0309], device='cuda:0'), out_proj_covar=tensor([1.3950e-04, 7.4962e-05, 9.0806e-05, 8.5505e-05, 1.8353e-04, 1.5494e-04, 7.9501e-05, 1.5583e-04], device='cuda:0') 2023-03-07 17:53:38,746 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17362.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:54:13,783 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17383.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 17:54:18,025 INFO [train2.py:809] (0/4) Epoch 5, batch 1450, loss[ctc_loss=0.1758, att_loss=0.2767, loss=0.2565, over 15943.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.006288, over 41.00 utterances.], tot_loss[ctc_loss=0.1632, att_loss=0.2812, loss=0.2576, over 3259055.88 frames. utt_duration=1256 frames, utt_pad_proportion=0.05563, over 10393.20 utterances.], batch size: 41, lr: 2.13e-02, grad_scale: 8.0 2023-03-07 17:54:41,452 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2023-03-07 17:54:55,332 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.883e+02 3.556e+02 4.184e+02 5.156e+02 1.348e+03, threshold=8.369e+02, percent-clipped=7.0 2023-03-07 17:55:19,596 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17423.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:55:39,528 INFO [train2.py:809] (0/4) Epoch 5, batch 1500, loss[ctc_loss=0.1688, att_loss=0.2746, loss=0.2534, over 16129.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005452, over 42.00 utterances.], tot_loss[ctc_loss=0.1627, att_loss=0.2816, loss=0.2578, over 3266857.24 frames. utt_duration=1263 frames, utt_pad_proportion=0.05192, over 10354.94 utterances.], batch size: 42, lr: 2.13e-02, grad_scale: 8.0 2023-03-07 17:55:52,493 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17444.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 17:56:59,735 INFO [train2.py:809] (0/4) Epoch 5, batch 1550, loss[ctc_loss=0.1942, att_loss=0.3024, loss=0.2808, over 16629.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.00508, over 47.00 utterances.], tot_loss[ctc_loss=0.1619, att_loss=0.2811, loss=0.2572, over 3274172.42 frames. utt_duration=1283 frames, utt_pad_proportion=0.04527, over 10221.38 utterances.], batch size: 47, lr: 2.13e-02, grad_scale: 8.0 2023-03-07 17:57:36,014 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.215e+02 3.523e+02 4.297e+02 5.219e+02 1.408e+03, threshold=8.593e+02, percent-clipped=2.0 2023-03-07 17:57:52,080 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.15 vs. limit=5.0 2023-03-07 17:58:01,423 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17524.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 17:58:20,678 INFO [train2.py:809] (0/4) Epoch 5, batch 1600, loss[ctc_loss=0.1176, att_loss=0.2371, loss=0.2132, over 15634.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009747, over 37.00 utterances.], tot_loss[ctc_loss=0.1628, att_loss=0.2814, loss=0.2577, over 3270531.45 frames. utt_duration=1259 frames, utt_pad_proportion=0.05259, over 10404.49 utterances.], batch size: 37, lr: 2.12e-02, grad_scale: 8.0 2023-03-07 17:58:34,852 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17545.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 17:59:03,257 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17562.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 17:59:20,712 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17572.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 17:59:41,753 INFO [train2.py:809] (0/4) Epoch 5, batch 1650, loss[ctc_loss=0.1829, att_loss=0.2999, loss=0.2765, over 16961.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.006944, over 50.00 utterances.], tot_loss[ctc_loss=0.1628, att_loss=0.2818, loss=0.258, over 3277540.30 frames. utt_duration=1247 frames, utt_pad_proportion=0.05403, over 10524.69 utterances.], batch size: 50, lr: 2.12e-02, grad_scale: 8.0 2023-03-07 17:59:48,365 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1163, 4.3451, 4.8100, 4.8298, 2.4138, 4.8574, 2.9956, 2.8216], device='cuda:0'), covar=tensor([0.0297, 0.0214, 0.0632, 0.0213, 0.2906, 0.0148, 0.1695, 0.1598], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0100, 0.0241, 0.0117, 0.0236, 0.0098, 0.0222, 0.0208], device='cuda:0'), out_proj_covar=tensor([1.0298e-04, 9.8576e-05, 2.1094e-04, 1.0195e-04, 2.0471e-04, 9.2352e-05, 1.9007e-04, 1.8016e-04], device='cuda:0') 2023-03-07 18:00:09,812 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8956, 4.6771, 4.6971, 4.8857, 5.1074, 4.7915, 4.8292, 2.1101], device='cuda:0'), covar=tensor([0.0217, 0.0290, 0.0218, 0.0120, 0.0999, 0.0186, 0.0197, 0.2994], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0119, 0.0113, 0.0124, 0.0275, 0.0127, 0.0110, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 18:00:19,756 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.071e+02 3.498e+02 4.244e+02 5.016e+02 1.623e+03, threshold=8.488e+02, percent-clipped=3.0 2023-03-07 18:00:22,199 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17610.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:00:46,423 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3289, 2.0202, 3.0053, 4.2122, 3.9245, 4.0473, 2.5337, 2.0806], device='cuda:0'), covar=tensor([0.0600, 0.2885, 0.1305, 0.0694, 0.0506, 0.0212, 0.2128, 0.2787], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0189, 0.0189, 0.0146, 0.0136, 0.0116, 0.0184, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 18:01:04,845 INFO [train2.py:809] (0/4) Epoch 5, batch 1700, loss[ctc_loss=0.1313, att_loss=0.2476, loss=0.2243, over 14481.00 frames. utt_duration=1812 frames, utt_pad_proportion=0.0395, over 32.00 utterances.], tot_loss[ctc_loss=0.1622, att_loss=0.2812, loss=0.2574, over 3274748.02 frames. utt_duration=1265 frames, utt_pad_proportion=0.04913, over 10369.82 utterances.], batch size: 32, lr: 2.12e-02, grad_scale: 8.0 2023-03-07 18:02:26,023 INFO [train2.py:809] (0/4) Epoch 5, batch 1750, loss[ctc_loss=0.1594, att_loss=0.2801, loss=0.256, over 17296.00 frames. utt_duration=876.9 frames, utt_pad_proportion=0.07983, over 79.00 utterances.], tot_loss[ctc_loss=0.1624, att_loss=0.2817, loss=0.2579, over 3282568.90 frames. utt_duration=1248 frames, utt_pad_proportion=0.05088, over 10533.57 utterances.], batch size: 79, lr: 2.12e-02, grad_scale: 8.0 2023-03-07 18:02:29,232 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0133, 5.2159, 5.6724, 5.6498, 5.1748, 5.9734, 5.0873, 5.9794], device='cuda:0'), covar=tensor([0.0508, 0.0577, 0.0495, 0.0612, 0.1814, 0.0695, 0.0561, 0.0486], device='cuda:0'), in_proj_covar=tensor([0.0501, 0.0323, 0.0330, 0.0395, 0.0557, 0.0329, 0.0280, 0.0350], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 18:02:47,915 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17700.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:02:55,658 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2609, 1.9879, 2.9975, 4.1662, 3.9564, 4.1301, 2.5084, 1.8853], device='cuda:0'), covar=tensor([0.0656, 0.2891, 0.1390, 0.0625, 0.0460, 0.0216, 0.2198, 0.3114], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0194, 0.0191, 0.0148, 0.0138, 0.0116, 0.0188, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 18:03:02,985 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.276e+02 3.443e+02 4.314e+02 5.475e+02 1.397e+03, threshold=8.628e+02, percent-clipped=3.0 2023-03-07 18:03:17,802 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17718.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:03:37,955 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17730.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:03:46,847 INFO [train2.py:809] (0/4) Epoch 5, batch 1800, loss[ctc_loss=0.1785, att_loss=0.2938, loss=0.2707, over 17025.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007569, over 51.00 utterances.], tot_loss[ctc_loss=0.1638, att_loss=0.2823, loss=0.2586, over 3284079.12 frames. utt_duration=1248 frames, utt_pad_proportion=0.05018, over 10537.19 utterances.], batch size: 51, lr: 2.11e-02, grad_scale: 8.0 2023-03-07 18:03:51,588 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=17739.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 18:04:17,040 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.6425, 3.1004, 3.5276, 2.2638, 3.1033, 3.1401, 2.7473, 1.4503], device='cuda:0'), covar=tensor([0.1745, 0.0914, 0.0759, 0.3809, 0.0934, 0.3576, 0.1628, 0.8319], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0067, 0.0068, 0.0087, 0.0064, 0.0086, 0.0064, 0.0107], device='cuda:0'), out_proj_covar=tensor([4.7807e-05, 4.3956e-05, 4.5078e-05, 6.1322e-05, 4.4786e-05, 6.3523e-05, 4.5616e-05, 7.9612e-05], device='cuda:0') 2023-03-07 18:04:27,134 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17761.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:04:34,570 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-03-07 18:05:07,329 INFO [train2.py:809] (0/4) Epoch 5, batch 1850, loss[ctc_loss=0.1354, att_loss=0.2419, loss=0.2206, over 15351.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.01116, over 35.00 utterances.], tot_loss[ctc_loss=0.1633, att_loss=0.2821, loss=0.2584, over 3289433.38 frames. utt_duration=1274 frames, utt_pad_proportion=0.04309, over 10339.21 utterances.], batch size: 35, lr: 2.11e-02, grad_scale: 8.0 2023-03-07 18:05:09,270 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.5225, 2.3232, 2.4517, 1.7399, 2.0695, 2.0991, 2.3878, 1.3402], device='cuda:0'), covar=tensor([0.1174, 0.1141, 0.1823, 0.3968, 0.1629, 0.4047, 0.1195, 0.8678], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0068, 0.0071, 0.0090, 0.0066, 0.0091, 0.0065, 0.0113], device='cuda:0'), out_proj_covar=tensor([4.9487e-05, 4.5338e-05, 4.7064e-05, 6.3757e-05, 4.6110e-05, 6.6742e-05, 4.6946e-05, 8.3312e-05], device='cuda:0') 2023-03-07 18:05:15,573 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=17791.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:05:45,056 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.153e+02 3.254e+02 3.960e+02 5.162e+02 1.019e+03, threshold=7.921e+02, percent-clipped=3.0 2023-03-07 18:06:28,251 INFO [train2.py:809] (0/4) Epoch 5, batch 1900, loss[ctc_loss=0.1743, att_loss=0.2967, loss=0.2722, over 17430.00 frames. utt_duration=1012 frames, utt_pad_proportion=0.04624, over 69.00 utterances.], tot_loss[ctc_loss=0.1651, att_loss=0.2834, loss=0.2597, over 3291210.05 frames. utt_duration=1237 frames, utt_pad_proportion=0.05251, over 10654.04 utterances.], batch size: 69, lr: 2.11e-02, grad_scale: 8.0 2023-03-07 18:06:42,537 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=17845.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:07:49,573 INFO [train2.py:809] (0/4) Epoch 5, batch 1950, loss[ctc_loss=0.1374, att_loss=0.2533, loss=0.2301, over 15628.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009827, over 37.00 utterances.], tot_loss[ctc_loss=0.1647, att_loss=0.283, loss=0.2593, over 3278196.38 frames. utt_duration=1219 frames, utt_pad_proportion=0.061, over 10765.98 utterances.], batch size: 37, lr: 2.11e-02, grad_scale: 8.0 2023-03-07 18:08:00,641 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=17893.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:08:16,299 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-07 18:08:27,510 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.984e+02 3.619e+02 4.276e+02 5.445e+02 1.197e+03, threshold=8.551e+02, percent-clipped=4.0 2023-03-07 18:09:10,816 INFO [train2.py:809] (0/4) Epoch 5, batch 2000, loss[ctc_loss=0.1478, att_loss=0.2866, loss=0.2589, over 17023.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007584, over 51.00 utterances.], tot_loss[ctc_loss=0.1634, att_loss=0.2817, loss=0.258, over 3275883.96 frames. utt_duration=1201 frames, utt_pad_proportion=0.0653, over 10924.25 utterances.], batch size: 51, lr: 2.10e-02, grad_scale: 8.0 2023-03-07 18:09:35,645 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3487, 5.0540, 4.9444, 2.7618, 1.9916, 2.5772, 4.8322, 3.7106], device='cuda:0'), covar=tensor([0.0388, 0.0144, 0.0180, 0.2432, 0.5906, 0.2627, 0.0185, 0.1866], device='cuda:0'), in_proj_covar=tensor([0.0265, 0.0166, 0.0194, 0.0181, 0.0360, 0.0322, 0.0173, 0.0309], device='cuda:0'), out_proj_covar=tensor([1.4026e-04, 7.6309e-05, 8.9872e-05, 8.3851e-05, 1.7963e-04, 1.5176e-04, 7.9070e-05, 1.5401e-04], device='cuda:0') 2023-03-07 18:09:50,813 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=17960.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:10:32,022 INFO [train2.py:809] (0/4) Epoch 5, batch 2050, loss[ctc_loss=0.1263, att_loss=0.2679, loss=0.2396, over 16181.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.005603, over 41.00 utterances.], tot_loss[ctc_loss=0.1626, att_loss=0.2815, loss=0.2577, over 3278777.31 frames. utt_duration=1211 frames, utt_pad_proportion=0.06151, over 10847.45 utterances.], batch size: 41, lr: 2.10e-02, grad_scale: 8.0 2023-03-07 18:10:54,536 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-18000.pt 2023-03-07 18:11:11,920 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-03-07 18:11:14,251 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.268e+02 3.678e+02 4.390e+02 5.847e+02 1.003e+03, threshold=8.779e+02, percent-clipped=6.0 2023-03-07 18:11:29,715 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18018.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:11:34,438 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18021.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:11:44,834 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-07 18:11:57,544 INFO [train2.py:809] (0/4) Epoch 5, batch 2100, loss[ctc_loss=0.1624, att_loss=0.2983, loss=0.2711, over 17016.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008021, over 51.00 utterances.], tot_loss[ctc_loss=0.1624, att_loss=0.282, loss=0.2581, over 3287433.40 frames. utt_duration=1231 frames, utt_pad_proportion=0.05563, over 10691.18 utterances.], batch size: 51, lr: 2.10e-02, grad_scale: 8.0 2023-03-07 18:12:02,548 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18039.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 18:12:21,574 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.11 vs. limit=5.0 2023-03-07 18:12:30,774 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18056.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:12:40,735 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.39 vs. limit=5.0 2023-03-07 18:12:42,589 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7039, 2.4697, 5.0684, 3.7313, 3.1479, 4.3326, 4.5998, 4.7200], device='cuda:0'), covar=tensor([0.0124, 0.1846, 0.0087, 0.1266, 0.2223, 0.0248, 0.0179, 0.0215], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0240, 0.0120, 0.0294, 0.0316, 0.0181, 0.0107, 0.0133], device='cuda:0'), out_proj_covar=tensor([1.1082e-04, 1.9051e-04, 1.0245e-04, 2.3549e-04, 2.5339e-04, 1.5204e-04, 9.3496e-05, 1.1626e-04], device='cuda:0') 2023-03-07 18:12:46,838 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18066.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:13:00,898 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7712, 5.1793, 4.5252, 5.2374, 4.5568, 4.9901, 5.3556, 5.1633], device='cuda:0'), covar=tensor([0.0359, 0.0176, 0.0673, 0.0140, 0.0392, 0.0144, 0.0162, 0.0122], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0169, 0.0216, 0.0136, 0.0183, 0.0134, 0.0158, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-07 18:13:00,932 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5774, 4.9344, 4.7441, 4.9103, 5.0323, 4.7368, 4.0334, 4.8514], device='cuda:0'), covar=tensor([0.0076, 0.0114, 0.0072, 0.0063, 0.0056, 0.0085, 0.0361, 0.0151], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0052, 0.0055, 0.0038, 0.0037, 0.0048, 0.0071, 0.0066], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 18:13:17,736 INFO [train2.py:809] (0/4) Epoch 5, batch 2150, loss[ctc_loss=0.1259, att_loss=0.2679, loss=0.2395, over 16708.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005398, over 46.00 utterances.], tot_loss[ctc_loss=0.1623, att_loss=0.2813, loss=0.2575, over 3278901.93 frames. utt_duration=1230 frames, utt_pad_proportion=0.05834, over 10675.31 utterances.], batch size: 46, lr: 2.09e-02, grad_scale: 8.0 2023-03-07 18:13:17,961 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18086.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:13:19,447 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18087.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 18:13:54,649 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.249e+02 3.704e+02 4.287e+02 5.437e+02 1.158e+03, threshold=8.574e+02, percent-clipped=4.0 2023-03-07 18:14:38,385 INFO [train2.py:809] (0/4) Epoch 5, batch 2200, loss[ctc_loss=0.1469, att_loss=0.2519, loss=0.2309, over 15392.00 frames. utt_duration=1761 frames, utt_pad_proportion=0.009245, over 35.00 utterances.], tot_loss[ctc_loss=0.1609, att_loss=0.2807, loss=0.2567, over 3279676.53 frames. utt_duration=1227 frames, utt_pad_proportion=0.05861, over 10701.25 utterances.], batch size: 35, lr: 2.09e-02, grad_scale: 8.0 2023-03-07 18:15:32,712 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.69 vs. limit=5.0 2023-03-07 18:15:41,249 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18174.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:15:42,922 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18175.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:15:48,075 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-07 18:16:00,116 INFO [train2.py:809] (0/4) Epoch 5, batch 2250, loss[ctc_loss=0.1702, att_loss=0.287, loss=0.2636, over 17115.00 frames. utt_duration=693 frames, utt_pad_proportion=0.1305, over 99.00 utterances.], tot_loss[ctc_loss=0.1615, att_loss=0.2812, loss=0.2573, over 3276652.09 frames. utt_duration=1198 frames, utt_pad_proportion=0.06681, over 10952.79 utterances.], batch size: 99, lr: 2.09e-02, grad_scale: 8.0 2023-03-07 18:16:24,172 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.62 vs. limit=5.0 2023-03-07 18:16:38,770 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.354e+02 3.428e+02 4.207e+02 5.393e+02 1.191e+03, threshold=8.413e+02, percent-clipped=8.0 2023-03-07 18:16:48,424 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9175, 5.1010, 5.6230, 5.4453, 4.6634, 5.6654, 5.0205, 5.7879], device='cuda:0'), covar=tensor([0.1019, 0.1322, 0.0757, 0.1412, 0.3442, 0.1239, 0.0830, 0.0909], device='cuda:0'), in_proj_covar=tensor([0.0501, 0.0318, 0.0331, 0.0392, 0.0544, 0.0331, 0.0285, 0.0348], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 18:16:53,842 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-07 18:17:18,343 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4184, 4.5467, 4.2611, 4.7362, 2.2335, 4.6460, 2.3816, 1.6547], device='cuda:0'), covar=tensor([0.0180, 0.0180, 0.0904, 0.0171, 0.2799, 0.0162, 0.1884, 0.2123], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0099, 0.0243, 0.0110, 0.0230, 0.0097, 0.0217, 0.0204], device='cuda:0'), out_proj_covar=tensor([1.0175e-04, 9.7401e-05, 2.1246e-04, 9.5831e-05, 2.0157e-04, 9.0991e-05, 1.8673e-04, 1.7772e-04], device='cuda:0') 2023-03-07 18:17:20,586 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18235.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:17:21,716 INFO [train2.py:809] (0/4) Epoch 5, batch 2300, loss[ctc_loss=0.1473, att_loss=0.2574, loss=0.2354, over 15883.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009253, over 39.00 utterances.], tot_loss[ctc_loss=0.1611, att_loss=0.2813, loss=0.2573, over 3271155.05 frames. utt_duration=1186 frames, utt_pad_proportion=0.07087, over 11047.45 utterances.], batch size: 39, lr: 2.09e-02, grad_scale: 8.0 2023-03-07 18:17:22,105 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18236.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:17:54,383 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9719, 5.2114, 5.4960, 5.5479, 5.2615, 5.7958, 5.1608, 6.0016], device='cuda:0'), covar=tensor([0.0473, 0.0593, 0.0473, 0.0581, 0.1545, 0.0734, 0.0413, 0.0409], device='cuda:0'), in_proj_covar=tensor([0.0496, 0.0316, 0.0330, 0.0390, 0.0541, 0.0335, 0.0283, 0.0348], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 18:18:43,294 INFO [train2.py:809] (0/4) Epoch 5, batch 2350, loss[ctc_loss=0.1689, att_loss=0.26, loss=0.2418, over 15357.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01151, over 35.00 utterances.], tot_loss[ctc_loss=0.1626, att_loss=0.2828, loss=0.2587, over 3285798.34 frames. utt_duration=1185 frames, utt_pad_proportion=0.06622, over 11103.48 utterances.], batch size: 35, lr: 2.08e-02, grad_scale: 8.0 2023-03-07 18:19:21,027 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.258e+02 3.468e+02 4.143e+02 4.951e+02 1.211e+03, threshold=8.286e+02, percent-clipped=3.0 2023-03-07 18:19:32,546 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18316.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:20:04,252 INFO [train2.py:809] (0/4) Epoch 5, batch 2400, loss[ctc_loss=0.1359, att_loss=0.26, loss=0.2352, over 15882.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009518, over 39.00 utterances.], tot_loss[ctc_loss=0.1608, att_loss=0.2806, loss=0.2566, over 3271897.52 frames. utt_duration=1205 frames, utt_pad_proportion=0.06524, over 10871.66 utterances.], batch size: 39, lr: 2.08e-02, grad_scale: 16.0 2023-03-07 18:20:38,586 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18356.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:21:26,061 INFO [train2.py:809] (0/4) Epoch 5, batch 2450, loss[ctc_loss=0.1499, att_loss=0.2892, loss=0.2613, over 17038.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009151, over 52.00 utterances.], tot_loss[ctc_loss=0.1618, att_loss=0.2812, loss=0.2573, over 3273045.05 frames. utt_duration=1195 frames, utt_pad_proportion=0.06864, over 10972.66 utterances.], batch size: 52, lr: 2.08e-02, grad_scale: 16.0 2023-03-07 18:21:26,382 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18386.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:21:56,486 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18404.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:22:04,374 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 3.376e+02 4.245e+02 5.566e+02 1.055e+03, threshold=8.490e+02, percent-clipped=7.0 2023-03-07 18:22:14,031 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9161, 5.4008, 5.1815, 5.2900, 5.4471, 5.3587, 5.0382, 4.8848], device='cuda:0'), covar=tensor([0.1045, 0.0431, 0.0225, 0.0378, 0.0218, 0.0228, 0.0247, 0.0289], device='cuda:0'), in_proj_covar=tensor([0.0378, 0.0229, 0.0159, 0.0201, 0.0246, 0.0268, 0.0205, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 18:22:44,606 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18434.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:22:47,529 INFO [train2.py:809] (0/4) Epoch 5, batch 2500, loss[ctc_loss=0.1154, att_loss=0.258, loss=0.2295, over 16759.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006323, over 48.00 utterances.], tot_loss[ctc_loss=0.1611, att_loss=0.281, loss=0.257, over 3276364.74 frames. utt_duration=1194 frames, utt_pad_proportion=0.06761, over 10990.92 utterances.], batch size: 48, lr: 2.08e-02, grad_scale: 16.0 2023-03-07 18:23:12,701 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-07 18:24:09,079 INFO [train2.py:809] (0/4) Epoch 5, batch 2550, loss[ctc_loss=0.1409, att_loss=0.2836, loss=0.2551, over 17274.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01349, over 55.00 utterances.], tot_loss[ctc_loss=0.1619, att_loss=0.282, loss=0.258, over 3269215.24 frames. utt_duration=1159 frames, utt_pad_proportion=0.07715, over 11293.56 utterances.], batch size: 55, lr: 2.07e-02, grad_scale: 16.0 2023-03-07 18:24:15,590 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18490.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:24:32,070 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9105, 4.0785, 3.2713, 3.7590, 4.1327, 3.8414, 2.4221, 4.7351], device='cuda:0'), covar=tensor([0.1147, 0.0392, 0.1220, 0.0569, 0.0439, 0.0581, 0.1162, 0.0285], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0128, 0.0177, 0.0143, 0.0156, 0.0168, 0.0150, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 18:24:44,701 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18508.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 18:24:45,767 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.865e+02 3.434e+02 4.162e+02 5.436e+02 1.075e+03, threshold=8.324e+02, percent-clipped=2.0 2023-03-07 18:25:20,139 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18530.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:25:21,710 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18531.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:25:29,463 INFO [train2.py:809] (0/4) Epoch 5, batch 2600, loss[ctc_loss=0.1857, att_loss=0.2968, loss=0.2746, over 17396.00 frames. utt_duration=882.1 frames, utt_pad_proportion=0.0744, over 79.00 utterances.], tot_loss[ctc_loss=0.1607, att_loss=0.2807, loss=0.2567, over 3262189.14 frames. utt_duration=1192 frames, utt_pad_proportion=0.07206, over 10959.87 utterances.], batch size: 79, lr: 2.07e-02, grad_scale: 16.0 2023-03-07 18:25:54,344 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18551.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:26:22,999 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18569.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 18:26:50,086 INFO [train2.py:809] (0/4) Epoch 5, batch 2650, loss[ctc_loss=0.1318, att_loss=0.2551, loss=0.2304, over 16166.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007148, over 41.00 utterances.], tot_loss[ctc_loss=0.1618, att_loss=0.2814, loss=0.2575, over 3271730.61 frames. utt_duration=1211 frames, utt_pad_proportion=0.06526, over 10817.88 utterances.], batch size: 41, lr: 2.07e-02, grad_scale: 16.0 2023-03-07 18:27:27,353 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.996e+02 3.747e+02 4.549e+02 5.816e+02 1.798e+03, threshold=9.097e+02, percent-clipped=9.0 2023-03-07 18:27:33,776 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9140, 6.0308, 5.5016, 5.9554, 5.6905, 5.4750, 5.4500, 5.3543], device='cuda:0'), covar=tensor([0.0874, 0.0853, 0.0576, 0.0565, 0.0577, 0.1052, 0.1894, 0.1935], device='cuda:0'), in_proj_covar=tensor([0.0330, 0.0374, 0.0288, 0.0291, 0.0274, 0.0357, 0.0397, 0.0380], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 18:27:38,437 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18616.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:27:48,382 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-03-07 18:28:10,693 INFO [train2.py:809] (0/4) Epoch 5, batch 2700, loss[ctc_loss=0.13, att_loss=0.2714, loss=0.2431, over 17052.00 frames. utt_duration=1339 frames, utt_pad_proportion=0.006793, over 51.00 utterances.], tot_loss[ctc_loss=0.1607, att_loss=0.2803, loss=0.2564, over 3272941.84 frames. utt_duration=1240 frames, utt_pad_proportion=0.05836, over 10568.53 utterances.], batch size: 51, lr: 2.07e-02, grad_scale: 16.0 2023-03-07 18:28:56,340 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18664.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:29:13,363 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18674.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:29:14,894 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5497, 2.3734, 3.0629, 4.1726, 3.9772, 4.1427, 2.5598, 1.7952], device='cuda:0'), covar=tensor([0.0512, 0.2785, 0.1362, 0.0649, 0.0603, 0.0223, 0.2285, 0.3163], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0199, 0.0195, 0.0151, 0.0137, 0.0121, 0.0191, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 18:29:31,909 INFO [train2.py:809] (0/4) Epoch 5, batch 2750, loss[ctc_loss=0.1362, att_loss=0.2677, loss=0.2414, over 16015.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007632, over 40.00 utterances.], tot_loss[ctc_loss=0.1609, att_loss=0.2801, loss=0.2563, over 3274274.56 frames. utt_duration=1257 frames, utt_pad_proportion=0.05402, over 10428.76 utterances.], batch size: 40, lr: 2.06e-02, grad_scale: 16.0 2023-03-07 18:30:10,727 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.386e+02 3.574e+02 4.434e+02 5.832e+02 1.352e+03, threshold=8.867e+02, percent-clipped=7.0 2023-03-07 18:30:28,098 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4839, 3.8934, 2.9100, 3.5549, 3.8271, 3.6597, 2.6163, 4.4321], device='cuda:0'), covar=tensor([0.1269, 0.0287, 0.1255, 0.0534, 0.0505, 0.0646, 0.1014, 0.0340], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0130, 0.0180, 0.0146, 0.0158, 0.0169, 0.0154, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 18:30:51,292 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18735.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:30:52,510 INFO [train2.py:809] (0/4) Epoch 5, batch 2800, loss[ctc_loss=0.1439, att_loss=0.2773, loss=0.2506, over 16687.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006438, over 46.00 utterances.], tot_loss[ctc_loss=0.16, att_loss=0.2793, loss=0.2554, over 3263126.18 frames. utt_duration=1262 frames, utt_pad_proportion=0.05482, over 10352.80 utterances.], batch size: 46, lr: 2.06e-02, grad_scale: 8.0 2023-03-07 18:32:14,581 INFO [train2.py:809] (0/4) Epoch 5, batch 2850, loss[ctc_loss=0.1454, att_loss=0.2634, loss=0.2398, over 16109.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006622, over 42.00 utterances.], tot_loss[ctc_loss=0.1607, att_loss=0.2802, loss=0.2563, over 3257612.05 frames. utt_duration=1223 frames, utt_pad_proportion=0.06594, over 10671.09 utterances.], batch size: 42, lr: 2.06e-02, grad_scale: 8.0 2023-03-07 18:32:54,815 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.041e+02 3.458e+02 4.600e+02 5.611e+02 1.235e+03, threshold=9.200e+02, percent-clipped=3.0 2023-03-07 18:33:27,983 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18830.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:33:29,497 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=18831.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:33:37,334 INFO [train2.py:809] (0/4) Epoch 5, batch 2900, loss[ctc_loss=0.1763, att_loss=0.3041, loss=0.2786, over 17349.00 frames. utt_duration=1178 frames, utt_pad_proportion=0.02109, over 59.00 utterances.], tot_loss[ctc_loss=0.1598, att_loss=0.2804, loss=0.2563, over 3266322.07 frames. utt_duration=1225 frames, utt_pad_proportion=0.06325, over 10682.70 utterances.], batch size: 59, lr: 2.06e-02, grad_scale: 8.0 2023-03-07 18:33:53,710 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18846.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:34:21,717 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=18864.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 18:34:44,633 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18878.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:34:46,046 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=18879.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:34:56,889 INFO [train2.py:809] (0/4) Epoch 5, batch 2950, loss[ctc_loss=0.1379, att_loss=0.2597, loss=0.2354, over 16112.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.00633, over 42.00 utterances.], tot_loss[ctc_loss=0.1589, att_loss=0.2798, loss=0.2556, over 3274940.45 frames. utt_duration=1235 frames, utt_pad_proportion=0.05749, over 10619.05 utterances.], batch size: 42, lr: 2.05e-02, grad_scale: 8.0 2023-03-07 18:35:34,922 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.226e+02 3.285e+02 4.050e+02 4.764e+02 8.688e+02, threshold=8.101e+02, percent-clipped=0.0 2023-03-07 18:36:07,995 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=18930.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:36:17,501 INFO [train2.py:809] (0/4) Epoch 5, batch 3000, loss[ctc_loss=0.1072, att_loss=0.2455, loss=0.2178, over 15768.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.007561, over 38.00 utterances.], tot_loss[ctc_loss=0.1596, att_loss=0.2798, loss=0.2558, over 3266813.22 frames. utt_duration=1202 frames, utt_pad_proportion=0.06704, over 10889.13 utterances.], batch size: 38, lr: 2.05e-02, grad_scale: 8.0 2023-03-07 18:36:17,504 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 18:36:31,699 INFO [train2.py:843] (0/4) Epoch 5, validation: ctc_loss=0.07531, att_loss=0.2513, loss=0.2161, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 18:36:31,700 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 18:37:52,832 INFO [train2.py:809] (0/4) Epoch 5, batch 3050, loss[ctc_loss=0.1668, att_loss=0.2986, loss=0.2722, over 17045.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009318, over 53.00 utterances.], tot_loss[ctc_loss=0.16, att_loss=0.2802, loss=0.2562, over 3274986.85 frames. utt_duration=1220 frames, utt_pad_proportion=0.05946, over 10748.47 utterances.], batch size: 53, lr: 2.05e-02, grad_scale: 8.0 2023-03-07 18:38:01,057 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=18991.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:38:17,398 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-03-07 18:38:31,927 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.565e+02 3.659e+02 4.471e+02 5.591e+02 1.197e+03, threshold=8.941e+02, percent-clipped=8.0 2023-03-07 18:39:04,296 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19030.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:39:13,978 INFO [train2.py:809] (0/4) Epoch 5, batch 3100, loss[ctc_loss=0.2458, att_loss=0.3155, loss=0.3016, over 16862.00 frames. utt_duration=682.8 frames, utt_pad_proportion=0.1411, over 99.00 utterances.], tot_loss[ctc_loss=0.1604, att_loss=0.2808, loss=0.2567, over 3279383.57 frames. utt_duration=1208 frames, utt_pad_proportion=0.06264, over 10869.41 utterances.], batch size: 99, lr: 2.05e-02, grad_scale: 8.0 2023-03-07 18:39:56,108 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2711, 5.0921, 4.9924, 3.0050, 1.9418, 2.3910, 4.9710, 3.7844], device='cuda:0'), covar=tensor([0.0421, 0.0167, 0.0242, 0.2506, 0.6681, 0.3160, 0.0192, 0.1938], device='cuda:0'), in_proj_covar=tensor([0.0269, 0.0173, 0.0200, 0.0184, 0.0360, 0.0326, 0.0182, 0.0314], device='cuda:0'), out_proj_covar=tensor([1.3942e-04, 7.6729e-05, 9.0078e-05, 8.6282e-05, 1.7757e-04, 1.4999e-04, 8.0455e-05, 1.5367e-04], device='cuda:0') 2023-03-07 18:40:36,264 INFO [train2.py:809] (0/4) Epoch 5, batch 3150, loss[ctc_loss=0.145, att_loss=0.2772, loss=0.2508, over 16331.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006186, over 45.00 utterances.], tot_loss[ctc_loss=0.1603, att_loss=0.2808, loss=0.2567, over 3274643.31 frames. utt_duration=1205 frames, utt_pad_proportion=0.06619, over 10882.14 utterances.], batch size: 45, lr: 2.04e-02, grad_scale: 8.0 2023-03-07 18:41:12,120 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7684, 2.2936, 5.0332, 3.9943, 3.0615, 4.5274, 4.4985, 4.6453], device='cuda:0'), covar=tensor([0.0136, 0.1932, 0.0099, 0.1094, 0.2125, 0.0205, 0.0158, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0240, 0.0120, 0.0297, 0.0308, 0.0186, 0.0108, 0.0136], device='cuda:0'), out_proj_covar=tensor([1.1574e-04, 1.9226e-04, 1.0210e-04, 2.3781e-04, 2.5183e-04, 1.5767e-04, 9.6238e-05, 1.1995e-04], device='cuda:0') 2023-03-07 18:41:13,276 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-07 18:41:14,955 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.126e+02 3.661e+02 4.363e+02 5.729e+02 1.353e+03, threshold=8.726e+02, percent-clipped=4.0 2023-03-07 18:41:46,203 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19129.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 18:41:57,278 INFO [train2.py:809] (0/4) Epoch 5, batch 3200, loss[ctc_loss=0.1794, att_loss=0.3095, loss=0.2835, over 17326.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02256, over 59.00 utterances.], tot_loss[ctc_loss=0.159, att_loss=0.28, loss=0.2558, over 3278106.25 frames. utt_duration=1244 frames, utt_pad_proportion=0.05621, over 10555.33 utterances.], batch size: 59, lr: 2.04e-02, grad_scale: 8.0 2023-03-07 18:42:14,036 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19146.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:42:22,174 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4837, 1.5383, 2.3304, 1.8354, 3.9164, 2.1090, 1.0958, 1.9042], device='cuda:0'), covar=tensor([0.0409, 0.2442, 0.1836, 0.1265, 0.0260, 0.1596, 0.3145, 0.1578], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0104, 0.0098, 0.0090, 0.0083, 0.0088, 0.0106, 0.0090], device='cuda:0'), out_proj_covar=tensor([4.1794e-05, 5.5985e-05, 5.3429e-05, 4.4836e-05, 3.9704e-05, 4.8610e-05, 5.3652e-05, 4.9109e-05], device='cuda:0') 2023-03-07 18:42:43,156 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19164.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 18:43:10,298 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4310, 2.4548, 3.1664, 4.2627, 3.9586, 4.1141, 2.5457, 1.7962], device='cuda:0'), covar=tensor([0.0589, 0.2296, 0.1100, 0.0582, 0.0465, 0.0248, 0.2098, 0.2824], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0192, 0.0187, 0.0149, 0.0139, 0.0120, 0.0190, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 18:43:19,476 INFO [train2.py:809] (0/4) Epoch 5, batch 3250, loss[ctc_loss=0.1371, att_loss=0.2761, loss=0.2483, over 17040.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01035, over 53.00 utterances.], tot_loss[ctc_loss=0.1596, att_loss=0.2805, loss=0.2563, over 3277386.30 frames. utt_duration=1228 frames, utt_pad_proportion=0.0601, over 10690.20 utterances.], batch size: 53, lr: 2.04e-02, grad_scale: 8.0 2023-03-07 18:43:26,730 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19190.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 18:43:32,626 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19194.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:43:58,427 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.144e+02 3.305e+02 4.293e+02 5.572e+02 8.816e+02, threshold=8.587e+02, percent-clipped=1.0 2023-03-07 18:44:01,713 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19212.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 18:44:19,555 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3349, 2.7657, 3.5110, 2.3737, 3.1877, 4.4462, 4.2507, 3.1194], device='cuda:0'), covar=tensor([0.0341, 0.1698, 0.1111, 0.1517, 0.1089, 0.0514, 0.0501, 0.1294], device='cuda:0'), in_proj_covar=tensor([0.0200, 0.0200, 0.0200, 0.0188, 0.0205, 0.0199, 0.0167, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 18:44:40,691 INFO [train2.py:809] (0/4) Epoch 5, batch 3300, loss[ctc_loss=0.1242, att_loss=0.2446, loss=0.2205, over 15510.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.008274, over 36.00 utterances.], tot_loss[ctc_loss=0.1594, att_loss=0.2803, loss=0.2561, over 3278769.76 frames. utt_duration=1227 frames, utt_pad_proportion=0.0597, over 10705.54 utterances.], batch size: 36, lr: 2.04e-02, grad_scale: 8.0 2023-03-07 18:44:42,787 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6706, 2.3335, 5.1316, 3.9684, 3.2268, 4.3714, 4.5499, 4.8400], device='cuda:0'), covar=tensor([0.0180, 0.1962, 0.0129, 0.1163, 0.2034, 0.0280, 0.0196, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0243, 0.0121, 0.0300, 0.0310, 0.0190, 0.0111, 0.0137], device='cuda:0'), out_proj_covar=tensor([1.1670e-04, 1.9442e-04, 1.0340e-04, 2.4104e-04, 2.5378e-04, 1.6153e-04, 9.8278e-05, 1.2070e-04], device='cuda:0') 2023-03-07 18:46:01,524 INFO [train2.py:809] (0/4) Epoch 5, batch 3350, loss[ctc_loss=0.1485, att_loss=0.2888, loss=0.2608, over 17320.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.01091, over 55.00 utterances.], tot_loss[ctc_loss=0.1586, att_loss=0.2794, loss=0.2552, over 3270962.90 frames. utt_duration=1235 frames, utt_pad_proportion=0.05761, over 10606.50 utterances.], batch size: 55, lr: 2.03e-02, grad_scale: 8.0 2023-03-07 18:46:01,702 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19286.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:46:39,663 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.279e+02 3.326e+02 4.027e+02 5.244e+02 1.066e+03, threshold=8.053e+02, percent-clipped=2.0 2023-03-07 18:47:13,622 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19330.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:47:22,652 INFO [train2.py:809] (0/4) Epoch 5, batch 3400, loss[ctc_loss=0.1863, att_loss=0.3095, loss=0.2849, over 17331.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02285, over 59.00 utterances.], tot_loss[ctc_loss=0.1587, att_loss=0.2795, loss=0.2553, over 3264753.73 frames. utt_duration=1200 frames, utt_pad_proportion=0.06705, over 10897.71 utterances.], batch size: 59, lr: 2.03e-02, grad_scale: 8.0 2023-03-07 18:47:39,187 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.2329, 2.9857, 3.1180, 2.1799, 2.7719, 2.9825, 2.7927, 1.5217], device='cuda:0'), covar=tensor([0.3282, 0.0804, 0.3283, 0.6459, 0.2618, 0.3627, 0.0961, 1.0227], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0068, 0.0072, 0.0093, 0.0069, 0.0089, 0.0065, 0.0109], device='cuda:0'), out_proj_covar=tensor([5.1564e-05, 4.6690e-05, 5.2070e-05, 6.7814e-05, 4.9428e-05, 6.7893e-05, 4.7814e-05, 8.2406e-05], device='cuda:0') 2023-03-07 18:47:50,850 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2023-03-07 18:48:30,583 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19378.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:48:42,963 INFO [train2.py:809] (0/4) Epoch 5, batch 3450, loss[ctc_loss=0.1662, att_loss=0.269, loss=0.2484, over 16002.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.00759, over 40.00 utterances.], tot_loss[ctc_loss=0.1573, att_loss=0.2791, loss=0.2547, over 3276852.00 frames. utt_duration=1233 frames, utt_pad_proportion=0.0571, over 10646.71 utterances.], batch size: 40, lr: 2.03e-02, grad_scale: 8.0 2023-03-07 18:48:56,557 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19394.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:49:21,717 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.889e+02 3.490e+02 4.079e+02 5.311e+02 1.453e+03, threshold=8.158e+02, percent-clipped=3.0 2023-03-07 18:49:25,045 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6338, 4.9918, 4.5275, 5.0453, 4.5000, 4.7953, 5.3175, 4.8926], device='cuda:0'), covar=tensor([0.0422, 0.0297, 0.0777, 0.0237, 0.0400, 0.0203, 0.0154, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0176, 0.0231, 0.0149, 0.0192, 0.0146, 0.0164, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-07 18:50:03,983 INFO [train2.py:809] (0/4) Epoch 5, batch 3500, loss[ctc_loss=0.1635, att_loss=0.2858, loss=0.2613, over 17045.00 frames. utt_duration=690.1 frames, utt_pad_proportion=0.1319, over 99.00 utterances.], tot_loss[ctc_loss=0.1566, att_loss=0.2785, loss=0.2542, over 3280002.68 frames. utt_duration=1234 frames, utt_pad_proportion=0.05686, over 10645.18 utterances.], batch size: 99, lr: 2.03e-02, grad_scale: 8.0 2023-03-07 18:50:17,677 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2650, 4.5204, 4.5399, 4.8828, 2.1177, 4.5522, 2.3039, 2.0426], device='cuda:0'), covar=tensor([0.0200, 0.0121, 0.0747, 0.0125, 0.3007, 0.0222, 0.2049, 0.2051], device='cuda:0'), in_proj_covar=tensor([0.0112, 0.0102, 0.0252, 0.0112, 0.0235, 0.0102, 0.0229, 0.0213], device='cuda:0'), out_proj_covar=tensor([1.0830e-04, 1.0137e-04, 2.2214e-04, 1.0089e-04, 2.0835e-04, 9.8689e-05, 1.9955e-04, 1.8700e-04], device='cuda:0') 2023-03-07 18:50:34,971 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19455.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:51:24,366 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19485.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 18:51:25,715 INFO [train2.py:809] (0/4) Epoch 5, batch 3550, loss[ctc_loss=0.1826, att_loss=0.2951, loss=0.2726, over 17198.00 frames. utt_duration=872.4 frames, utt_pad_proportion=0.08742, over 79.00 utterances.], tot_loss[ctc_loss=0.1568, att_loss=0.2787, loss=0.2543, over 3278306.45 frames. utt_duration=1214 frames, utt_pad_proportion=0.06303, over 10814.24 utterances.], batch size: 79, lr: 2.02e-02, grad_scale: 8.0 2023-03-07 18:52:02,972 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.286e+02 3.371e+02 4.237e+02 4.987e+02 1.005e+03, threshold=8.474e+02, percent-clipped=3.0 2023-03-07 18:52:46,421 INFO [train2.py:809] (0/4) Epoch 5, batch 3600, loss[ctc_loss=0.1603, att_loss=0.2857, loss=0.2606, over 16412.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007042, over 44.00 utterances.], tot_loss[ctc_loss=0.1573, att_loss=0.2786, loss=0.2543, over 3278285.35 frames. utt_duration=1219 frames, utt_pad_proportion=0.06175, over 10771.69 utterances.], batch size: 44, lr: 2.02e-02, grad_scale: 8.0 2023-03-07 18:53:28,838 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8103, 1.8794, 1.9129, 2.4158, 3.4044, 1.5900, 2.1272, 1.1151], device='cuda:0'), covar=tensor([0.0402, 0.1398, 0.1364, 0.0748, 0.0326, 0.1345, 0.1580, 0.1964], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0103, 0.0098, 0.0088, 0.0085, 0.0088, 0.0106, 0.0092], device='cuda:0'), out_proj_covar=tensor([4.2239e-05, 5.5753e-05, 5.3787e-05, 4.4281e-05, 4.0526e-05, 4.9494e-05, 5.4607e-05, 5.0162e-05], device='cuda:0') 2023-03-07 18:54:07,426 INFO [train2.py:809] (0/4) Epoch 5, batch 3650, loss[ctc_loss=0.1513, att_loss=0.2903, loss=0.2625, over 16756.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.007235, over 48.00 utterances.], tot_loss[ctc_loss=0.1589, att_loss=0.2791, loss=0.2551, over 3278561.18 frames. utt_duration=1193 frames, utt_pad_proportion=0.0662, over 11007.29 utterances.], batch size: 48, lr: 2.02e-02, grad_scale: 8.0 2023-03-07 18:54:07,769 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19586.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:54:13,020 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-03-07 18:54:44,974 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.264e+02 3.543e+02 4.665e+02 6.112e+02 1.723e+03, threshold=9.329e+02, percent-clipped=10.0 2023-03-07 18:54:59,195 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=19618.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:55:25,673 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19634.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:55:28,660 INFO [train2.py:809] (0/4) Epoch 5, batch 3700, loss[ctc_loss=0.1557, att_loss=0.2842, loss=0.2585, over 17401.00 frames. utt_duration=1181 frames, utt_pad_proportion=0.01806, over 59.00 utterances.], tot_loss[ctc_loss=0.1582, att_loss=0.2792, loss=0.255, over 3274658.87 frames. utt_duration=1189 frames, utt_pad_proportion=0.06953, over 11026.71 utterances.], batch size: 59, lr: 2.02e-02, grad_scale: 8.0 2023-03-07 18:56:39,186 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=19679.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:56:49,596 INFO [train2.py:809] (0/4) Epoch 5, batch 3750, loss[ctc_loss=0.1657, att_loss=0.2623, loss=0.2429, over 16280.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006945, over 43.00 utterances.], tot_loss[ctc_loss=0.158, att_loss=0.2784, loss=0.2543, over 3273715.86 frames. utt_duration=1200 frames, utt_pad_proportion=0.0665, over 10924.17 utterances.], batch size: 43, lr: 2.01e-02, grad_scale: 8.0 2023-03-07 18:56:59,438 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3995, 2.7425, 3.4451, 2.5134, 3.4259, 4.4050, 4.2285, 3.2729], device='cuda:0'), covar=tensor([0.0335, 0.1804, 0.1041, 0.1458, 0.0897, 0.0687, 0.0459, 0.1228], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0205, 0.0199, 0.0186, 0.0206, 0.0205, 0.0165, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 18:57:27,470 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.299e+02 3.451e+02 3.922e+02 5.029e+02 1.290e+03, threshold=7.843e+02, percent-clipped=3.0 2023-03-07 18:57:42,041 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5497, 5.1417, 5.0317, 3.2678, 1.9720, 2.7038, 5.1371, 3.7927], device='cuda:0'), covar=tensor([0.0346, 0.0211, 0.0246, 0.1998, 0.6553, 0.2591, 0.0164, 0.1960], device='cuda:0'), in_proj_covar=tensor([0.0276, 0.0174, 0.0200, 0.0188, 0.0365, 0.0330, 0.0186, 0.0326], device='cuda:0'), out_proj_covar=tensor([1.4171e-04, 7.6809e-05, 9.0248e-05, 8.6461e-05, 1.7862e-04, 1.5098e-04, 8.1205e-05, 1.5732e-04], device='cuda:0') 2023-03-07 18:58:11,651 INFO [train2.py:809] (0/4) Epoch 5, batch 3800, loss[ctc_loss=0.1612, att_loss=0.2926, loss=0.2663, over 16877.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007038, over 49.00 utterances.], tot_loss[ctc_loss=0.1572, att_loss=0.278, loss=0.2538, over 3273901.37 frames. utt_duration=1203 frames, utt_pad_proportion=0.06695, over 10900.04 utterances.], batch size: 49, lr: 2.01e-02, grad_scale: 8.0 2023-03-07 18:58:24,579 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0967, 4.6071, 4.2560, 4.6442, 4.5849, 4.4540, 3.8460, 4.4677], device='cuda:0'), covar=tensor([0.0100, 0.0104, 0.0122, 0.0082, 0.0089, 0.0092, 0.0402, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0055, 0.0061, 0.0041, 0.0042, 0.0052, 0.0076, 0.0072], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0001, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-03-07 18:58:33,859 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19750.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 18:59:30,510 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=19785.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 18:59:31,759 INFO [train2.py:809] (0/4) Epoch 5, batch 3850, loss[ctc_loss=0.1569, att_loss=0.2775, loss=0.2534, over 16179.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.007006, over 41.00 utterances.], tot_loss[ctc_loss=0.1567, att_loss=0.2778, loss=0.2536, over 3273578.11 frames. utt_duration=1226 frames, utt_pad_proportion=0.06096, over 10692.76 utterances.], batch size: 41, lr: 2.01e-02, grad_scale: 8.0 2023-03-07 19:00:09,267 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.273e+02 3.504e+02 4.097e+02 4.952e+02 1.319e+03, threshold=8.195e+02, percent-clipped=3.0 2023-03-07 19:00:13,854 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1719, 4.9295, 5.1001, 5.2137, 5.0334, 5.1297, 5.0187, 4.8131], device='cuda:0'), covar=tensor([0.2631, 0.0953, 0.0304, 0.0458, 0.0784, 0.0472, 0.0355, 0.0398], device='cuda:0'), in_proj_covar=tensor([0.0387, 0.0225, 0.0165, 0.0204, 0.0257, 0.0283, 0.0216, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 19:00:44,677 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=19833.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 19:00:48,871 INFO [train2.py:809] (0/4) Epoch 5, batch 3900, loss[ctc_loss=0.1627, att_loss=0.294, loss=0.2678, over 17050.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008897, over 52.00 utterances.], tot_loss[ctc_loss=0.1578, att_loss=0.2786, loss=0.2544, over 3272222.19 frames. utt_duration=1225 frames, utt_pad_proportion=0.06246, over 10699.76 utterances.], batch size: 52, lr: 2.01e-02, grad_scale: 8.0 2023-03-07 19:01:22,012 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-07 19:02:05,452 INFO [train2.py:809] (0/4) Epoch 5, batch 3950, loss[ctc_loss=0.1264, att_loss=0.2465, loss=0.2225, over 15512.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.008114, over 36.00 utterances.], tot_loss[ctc_loss=0.1579, att_loss=0.2788, loss=0.2546, over 3276193.45 frames. utt_duration=1232 frames, utt_pad_proportion=0.05851, over 10651.80 utterances.], batch size: 36, lr: 2.00e-02, grad_scale: 8.0 2023-03-07 19:02:43,149 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.585e+02 3.417e+02 4.181e+02 5.217e+02 1.635e+03, threshold=8.363e+02, percent-clipped=3.0 2023-03-07 19:02:58,733 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-5.pt 2023-03-07 19:03:26,307 INFO [train2.py:809] (0/4) Epoch 6, batch 0, loss[ctc_loss=0.1313, att_loss=0.2603, loss=0.2345, over 16269.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007904, over 43.00 utterances.], tot_loss[ctc_loss=0.1313, att_loss=0.2603, loss=0.2345, over 16269.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007904, over 43.00 utterances.], batch size: 43, lr: 1.87e-02, grad_scale: 8.0 2023-03-07 19:03:26,309 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 19:03:38,952 INFO [train2.py:843] (0/4) Epoch 6, validation: ctc_loss=0.07525, att_loss=0.2502, loss=0.2152, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 19:03:38,953 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 19:04:58,944 INFO [train2.py:809] (0/4) Epoch 6, batch 50, loss[ctc_loss=0.1698, att_loss=0.2893, loss=0.2654, over 17025.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.008349, over 51.00 utterances.], tot_loss[ctc_loss=0.1523, att_loss=0.2784, loss=0.2532, over 745051.85 frames. utt_duration=1296 frames, utt_pad_proportion=0.0355, over 2302.33 utterances.], batch size: 51, lr: 1.87e-02, grad_scale: 8.0 2023-03-07 19:05:05,308 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=19974.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:05:45,993 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-20000.pt 2023-03-07 19:05:58,069 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7182, 2.1680, 3.3113, 4.1190, 4.0723, 4.1644, 2.8702, 2.0564], device='cuda:0'), covar=tensor([0.0461, 0.2751, 0.1052, 0.0595, 0.0515, 0.0277, 0.1607, 0.2589], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0194, 0.0185, 0.0154, 0.0141, 0.0120, 0.0181, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:06:05,489 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.789e+02 3.240e+02 3.986e+02 5.094e+02 1.358e+03, threshold=7.972e+02, percent-clipped=5.0 2023-03-07 19:06:22,311 INFO [train2.py:809] (0/4) Epoch 6, batch 100, loss[ctc_loss=0.1229, att_loss=0.2383, loss=0.2152, over 15497.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.007996, over 36.00 utterances.], tot_loss[ctc_loss=0.153, att_loss=0.2763, loss=0.2516, over 1311176.27 frames. utt_duration=1314 frames, utt_pad_proportion=0.03104, over 3995.06 utterances.], batch size: 36, lr: 1.86e-02, grad_scale: 8.0 2023-03-07 19:06:22,655 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20020.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:06:46,248 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5097, 1.4839, 1.8123, 1.6529, 2.8549, 1.5424, 1.7034, 1.7751], device='cuda:0'), covar=tensor([0.0665, 0.3393, 0.2543, 0.2317, 0.1017, 0.3507, 0.4045, 0.2518], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0104, 0.0100, 0.0094, 0.0088, 0.0089, 0.0106, 0.0087], device='cuda:0'), out_proj_covar=tensor([4.3180e-05, 5.5771e-05, 5.4652e-05, 4.6118e-05, 4.1530e-05, 5.0412e-05, 5.5028e-05, 4.8540e-05], device='cuda:0') 2023-03-07 19:07:08,194 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2402, 2.4920, 3.4162, 2.7418, 3.4948, 4.3423, 4.2280, 3.2117], device='cuda:0'), covar=tensor([0.0444, 0.2185, 0.1119, 0.1452, 0.1005, 0.0500, 0.0423, 0.1431], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0209, 0.0203, 0.0189, 0.0213, 0.0206, 0.0168, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:07:09,773 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20050.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:07:41,940 INFO [train2.py:809] (0/4) Epoch 6, batch 150, loss[ctc_loss=0.1154, att_loss=0.2442, loss=0.2185, over 10935.00 frames. utt_duration=1824 frames, utt_pad_proportion=0.2038, over 24.00 utterances.], tot_loss[ctc_loss=0.1518, att_loss=0.2753, loss=0.2506, over 1737716.33 frames. utt_duration=1280 frames, utt_pad_proportion=0.04757, over 5438.07 utterances.], batch size: 24, lr: 1.86e-02, grad_scale: 8.0 2023-03-07 19:08:00,418 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20081.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:08:12,655 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20089.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:08:27,242 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20098.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:08:45,958 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.802e+02 3.260e+02 3.980e+02 4.904e+02 1.174e+03, threshold=7.960e+02, percent-clipped=3.0 2023-03-07 19:08:49,516 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20112.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:09:02,827 INFO [train2.py:809] (0/4) Epoch 6, batch 200, loss[ctc_loss=0.09561, att_loss=0.2205, loss=0.1956, over 14114.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.04766, over 31.00 utterances.], tot_loss[ctc_loss=0.1505, att_loss=0.2751, loss=0.2502, over 2083173.00 frames. utt_duration=1256 frames, utt_pad_proportion=0.0499, over 6642.40 utterances.], batch size: 31, lr: 1.86e-02, grad_scale: 8.0 2023-03-07 19:09:41,812 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.95 vs. limit=2.0 2023-03-07 19:09:51,000 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20150.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:10:23,263 INFO [train2.py:809] (0/4) Epoch 6, batch 250, loss[ctc_loss=0.1354, att_loss=0.255, loss=0.2311, over 15496.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.008977, over 36.00 utterances.], tot_loss[ctc_loss=0.1479, att_loss=0.2732, loss=0.2481, over 2344160.88 frames. utt_duration=1266 frames, utt_pad_proportion=0.04844, over 7413.94 utterances.], batch size: 36, lr: 1.86e-02, grad_scale: 8.0 2023-03-07 19:10:25,038 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0230, 4.9666, 4.8864, 3.2596, 4.8619, 4.4269, 4.4435, 2.5040], device='cuda:0'), covar=tensor([0.0137, 0.0083, 0.0242, 0.0861, 0.0074, 0.0163, 0.0226, 0.1409], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0064, 0.0054, 0.0099, 0.0059, 0.0072, 0.0080, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 19:10:28,284 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20173.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:11:17,534 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20204.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:11:27,955 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.870e+02 2.978e+02 3.813e+02 4.681e+02 1.334e+03, threshold=7.626e+02, percent-clipped=7.0 2023-03-07 19:11:43,058 INFO [train2.py:809] (0/4) Epoch 6, batch 300, loss[ctc_loss=0.1203, att_loss=0.2444, loss=0.2195, over 15993.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.00761, over 40.00 utterances.], tot_loss[ctc_loss=0.1488, att_loss=0.2744, loss=0.2492, over 2550178.66 frames. utt_duration=1233 frames, utt_pad_proportion=0.05834, over 8284.29 utterances.], batch size: 40, lr: 1.86e-02, grad_scale: 4.0 2023-03-07 19:12:54,599 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20265.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:13:01,870 INFO [train2.py:809] (0/4) Epoch 6, batch 350, loss[ctc_loss=0.1604, att_loss=0.2894, loss=0.2636, over 17310.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01139, over 55.00 utterances.], tot_loss[ctc_loss=0.1493, att_loss=0.275, loss=0.2499, over 2716845.07 frames. utt_duration=1234 frames, utt_pad_proportion=0.05663, over 8819.25 utterances.], batch size: 55, lr: 1.85e-02, grad_scale: 4.0 2023-03-07 19:13:08,318 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20274.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:13:11,980 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9229, 5.4050, 4.8156, 5.3620, 4.7782, 5.0873, 5.5516, 5.2506], device='cuda:0'), covar=tensor([0.0373, 0.0198, 0.0633, 0.0169, 0.0421, 0.0144, 0.0173, 0.0143], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0185, 0.0238, 0.0157, 0.0199, 0.0150, 0.0173, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-07 19:14:07,909 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.145e+02 3.150e+02 3.613e+02 4.726e+02 1.759e+03, threshold=7.227e+02, percent-clipped=5.0 2023-03-07 19:14:22,591 INFO [train2.py:809] (0/4) Epoch 6, batch 400, loss[ctc_loss=0.2158, att_loss=0.3266, loss=0.3044, over 17095.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01643, over 56.00 utterances.], tot_loss[ctc_loss=0.1499, att_loss=0.2758, loss=0.2506, over 2844174.64 frames. utt_duration=1205 frames, utt_pad_proportion=0.06269, over 9453.68 utterances.], batch size: 56, lr: 1.85e-02, grad_scale: 8.0 2023-03-07 19:14:25,714 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20322.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:15:42,859 INFO [train2.py:809] (0/4) Epoch 6, batch 450, loss[ctc_loss=0.2283, att_loss=0.3061, loss=0.2905, over 16857.00 frames. utt_duration=689.4 frames, utt_pad_proportion=0.1361, over 98.00 utterances.], tot_loss[ctc_loss=0.1502, att_loss=0.2753, loss=0.2503, over 2933163.58 frames. utt_duration=1219 frames, utt_pad_proportion=0.06167, over 9634.84 utterances.], batch size: 98, lr: 1.85e-02, grad_scale: 8.0 2023-03-07 19:15:53,012 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20376.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:16:12,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3672, 2.4442, 3.7374, 2.6525, 3.5544, 4.4674, 4.2259, 3.0914], device='cuda:0'), covar=tensor([0.0391, 0.2091, 0.0748, 0.1451, 0.0932, 0.0741, 0.0464, 0.1432], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0212, 0.0204, 0.0190, 0.0216, 0.0214, 0.0174, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:16:12,235 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20388.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:16:48,767 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.120e+02 3.385e+02 4.079e+02 4.609e+02 1.123e+03, threshold=8.157e+02, percent-clipped=3.0 2023-03-07 19:17:00,068 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20418.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:17:02,924 INFO [train2.py:809] (0/4) Epoch 6, batch 500, loss[ctc_loss=0.1559, att_loss=0.29, loss=0.2632, over 16690.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.00623, over 46.00 utterances.], tot_loss[ctc_loss=0.15, att_loss=0.2753, loss=0.2502, over 3008420.90 frames. utt_duration=1223 frames, utt_pad_proportion=0.06041, over 9851.74 utterances.], batch size: 46, lr: 1.85e-02, grad_scale: 8.0 2023-03-07 19:17:42,539 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20445.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:17:46,362 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-03-07 19:17:46,399 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.02 vs. limit=5.0 2023-03-07 19:17:48,933 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20449.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:17:51,710 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5948, 4.9225, 4.8668, 4.8891, 5.0139, 4.9923, 4.7025, 4.4880], device='cuda:0'), covar=tensor([0.1147, 0.0415, 0.0236, 0.0366, 0.0263, 0.0289, 0.0278, 0.0350], device='cuda:0'), in_proj_covar=tensor([0.0388, 0.0225, 0.0169, 0.0203, 0.0261, 0.0288, 0.0219, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 19:18:19,442 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20468.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:18:22,350 INFO [train2.py:809] (0/4) Epoch 6, batch 550, loss[ctc_loss=0.1362, att_loss=0.259, loss=0.2344, over 15960.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006731, over 41.00 utterances.], tot_loss[ctc_loss=0.1505, att_loss=0.2753, loss=0.2504, over 3067126.60 frames. utt_duration=1225 frames, utt_pad_proportion=0.05977, over 10030.67 utterances.], batch size: 41, lr: 1.84e-02, grad_scale: 8.0 2023-03-07 19:18:37,154 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20479.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:19:19,365 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9518, 4.0993, 3.2799, 3.8504, 4.0430, 3.7282, 2.4603, 4.6832], device='cuda:0'), covar=tensor([0.1093, 0.0403, 0.1030, 0.0553, 0.0579, 0.0668, 0.1173, 0.0277], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0137, 0.0184, 0.0152, 0.0169, 0.0177, 0.0155, 0.0162], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 19:19:28,294 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.949e+02 3.263e+02 3.939e+02 5.001e+02 1.057e+03, threshold=7.878e+02, percent-clipped=3.0 2023-03-07 19:19:43,048 INFO [train2.py:809] (0/4) Epoch 6, batch 600, loss[ctc_loss=0.155, att_loss=0.2842, loss=0.2584, over 17045.00 frames. utt_duration=1338 frames, utt_pad_proportion=0.006419, over 51.00 utterances.], tot_loss[ctc_loss=0.149, att_loss=0.2745, loss=0.2494, over 3118284.27 frames. utt_duration=1254 frames, utt_pad_proportion=0.05137, over 9960.70 utterances.], batch size: 51, lr: 1.84e-02, grad_scale: 8.0 2023-03-07 19:20:28,083 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-03-07 19:20:47,296 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20560.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:21:03,190 INFO [train2.py:809] (0/4) Epoch 6, batch 650, loss[ctc_loss=0.1182, att_loss=0.2412, loss=0.2166, over 15868.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01036, over 39.00 utterances.], tot_loss[ctc_loss=0.1495, att_loss=0.2755, loss=0.2503, over 3156953.64 frames. utt_duration=1225 frames, utt_pad_proportion=0.05818, over 10324.33 utterances.], batch size: 39, lr: 1.84e-02, grad_scale: 8.0 2023-03-07 19:21:32,111 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=20588.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:22:07,921 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-03-07 19:22:08,669 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.746e+02 2.894e+02 3.503e+02 4.491e+02 1.231e+03, threshold=7.006e+02, percent-clipped=2.0 2023-03-07 19:22:16,514 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2023-03-07 19:22:23,250 INFO [train2.py:809] (0/4) Epoch 6, batch 700, loss[ctc_loss=0.1377, att_loss=0.2748, loss=0.2474, over 16692.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005699, over 46.00 utterances.], tot_loss[ctc_loss=0.149, att_loss=0.2748, loss=0.2496, over 3179027.87 frames. utt_duration=1224 frames, utt_pad_proportion=0.06031, over 10402.50 utterances.], batch size: 46, lr: 1.84e-02, grad_scale: 8.0 2023-03-07 19:23:09,515 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=20649.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:23:42,151 INFO [train2.py:809] (0/4) Epoch 6, batch 750, loss[ctc_loss=0.1751, att_loss=0.2965, loss=0.2722, over 17480.00 frames. utt_duration=1015 frames, utt_pad_proportion=0.0436, over 69.00 utterances.], tot_loss[ctc_loss=0.15, att_loss=0.2751, loss=0.2501, over 3203979.09 frames. utt_duration=1233 frames, utt_pad_proportion=0.05685, over 10404.90 utterances.], batch size: 69, lr: 1.84e-02, grad_scale: 8.0 2023-03-07 19:23:48,424 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.2840, 5.4689, 5.8827, 5.9088, 5.6428, 6.1827, 5.2670, 6.2435], device='cuda:0'), covar=tensor([0.0448, 0.0606, 0.0416, 0.0593, 0.1350, 0.0600, 0.0399, 0.0486], device='cuda:0'), in_proj_covar=tensor([0.0525, 0.0345, 0.0363, 0.0426, 0.0588, 0.0368, 0.0303, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 19:23:51,711 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20676.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:24:47,042 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.294e+02 3.454e+02 4.265e+02 5.138e+02 1.250e+03, threshold=8.530e+02, percent-clipped=7.0 2023-03-07 19:25:01,488 INFO [train2.py:809] (0/4) Epoch 6, batch 800, loss[ctc_loss=0.2078, att_loss=0.3144, loss=0.2931, over 17295.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02402, over 59.00 utterances.], tot_loss[ctc_loss=0.151, att_loss=0.2755, loss=0.2506, over 3219506.66 frames. utt_duration=1245 frames, utt_pad_proportion=0.05391, over 10356.56 utterances.], batch size: 59, lr: 1.83e-02, grad_scale: 8.0 2023-03-07 19:25:07,725 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20724.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:25:39,392 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20744.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:25:41,035 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20745.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:26:18,421 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20768.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:26:21,151 INFO [train2.py:809] (0/4) Epoch 6, batch 850, loss[ctc_loss=0.1358, att_loss=0.2426, loss=0.2213, over 15378.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01, over 35.00 utterances.], tot_loss[ctc_loss=0.1507, att_loss=0.2744, loss=0.2497, over 3224435.88 frames. utt_duration=1252 frames, utt_pad_proportion=0.05459, over 10314.47 utterances.], batch size: 35, lr: 1.83e-02, grad_scale: 8.0 2023-03-07 19:26:27,673 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20774.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:26:57,841 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20793.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:27:09,038 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8942, 5.1029, 5.4635, 5.4652, 5.1047, 5.8131, 5.0298, 5.9199], device='cuda:0'), covar=tensor([0.0582, 0.0638, 0.0495, 0.0699, 0.2111, 0.0753, 0.0516, 0.0517], device='cuda:0'), in_proj_covar=tensor([0.0524, 0.0343, 0.0361, 0.0426, 0.0591, 0.0367, 0.0302, 0.0368], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 19:27:26,710 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.013e+02 3.098e+02 3.888e+02 4.985e+02 1.005e+03, threshold=7.777e+02, percent-clipped=2.0 2023-03-07 19:27:35,349 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20816.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:27:41,275 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.94 vs. limit=5.0 2023-03-07 19:27:41,724 INFO [train2.py:809] (0/4) Epoch 6, batch 900, loss[ctc_loss=0.144, att_loss=0.2756, loss=0.2493, over 17005.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.008792, over 51.00 utterances.], tot_loss[ctc_loss=0.1488, att_loss=0.2732, loss=0.2483, over 3225974.10 frames. utt_duration=1262 frames, utt_pad_proportion=0.05457, over 10239.86 utterances.], batch size: 51, lr: 1.83e-02, grad_scale: 8.0 2023-03-07 19:28:34,004 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-03-07 19:28:45,603 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=20860.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:29:01,655 INFO [train2.py:809] (0/4) Epoch 6, batch 950, loss[ctc_loss=0.1719, att_loss=0.2905, loss=0.2668, over 17364.00 frames. utt_duration=1179 frames, utt_pad_proportion=0.01954, over 59.00 utterances.], tot_loss[ctc_loss=0.1496, att_loss=0.2741, loss=0.2492, over 3244223.80 frames. utt_duration=1261 frames, utt_pad_proportion=0.05158, over 10300.90 utterances.], batch size: 59, lr: 1.83e-02, grad_scale: 8.0 2023-03-07 19:30:00,989 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=20908.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:30:06,285 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.745e+02 3.096e+02 4.099e+02 5.928e+02 9.906e+02, threshold=8.198e+02, percent-clipped=5.0 2023-03-07 19:30:21,095 INFO [train2.py:809] (0/4) Epoch 6, batch 1000, loss[ctc_loss=0.1248, att_loss=0.2407, loss=0.2175, over 15366.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01162, over 35.00 utterances.], tot_loss[ctc_loss=0.15, att_loss=0.2749, loss=0.2499, over 3255995.65 frames. utt_duration=1264 frames, utt_pad_proportion=0.0485, over 10318.36 utterances.], batch size: 35, lr: 1.83e-02, grad_scale: 8.0 2023-03-07 19:30:38,652 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5300, 5.2275, 4.8512, 5.1031, 5.0404, 4.8839, 3.9672, 4.9935], device='cuda:0'), covar=tensor([0.0093, 0.0114, 0.0094, 0.0084, 0.0076, 0.0081, 0.0454, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0055, 0.0062, 0.0044, 0.0042, 0.0052, 0.0075, 0.0073], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-07 19:30:59,398 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=20944.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:31:07,118 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4812, 2.6859, 3.7779, 2.7778, 3.5302, 4.7369, 4.4851, 3.1006], device='cuda:0'), covar=tensor([0.0452, 0.1996, 0.0977, 0.1623, 0.0968, 0.0501, 0.0389, 0.1606], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0209, 0.0207, 0.0196, 0.0209, 0.0221, 0.0176, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:31:40,996 INFO [train2.py:809] (0/4) Epoch 6, batch 1050, loss[ctc_loss=0.1427, att_loss=0.2832, loss=0.2551, over 16967.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007553, over 50.00 utterances.], tot_loss[ctc_loss=0.149, att_loss=0.2742, loss=0.2492, over 3260452.19 frames. utt_duration=1266 frames, utt_pad_proportion=0.04744, over 10316.49 utterances.], batch size: 50, lr: 1.82e-02, grad_scale: 8.0 2023-03-07 19:32:21,041 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4832, 2.0506, 3.3268, 4.3500, 4.0239, 4.4619, 3.1952, 2.4767], device='cuda:0'), covar=tensor([0.0570, 0.3070, 0.1096, 0.0516, 0.0561, 0.0168, 0.1325, 0.2256], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0195, 0.0189, 0.0154, 0.0148, 0.0120, 0.0188, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:32:22,689 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6544, 3.0293, 3.8633, 3.1567, 3.5687, 4.7484, 4.3824, 3.1933], device='cuda:0'), covar=tensor([0.0380, 0.1646, 0.0904, 0.1265, 0.0953, 0.0771, 0.0471, 0.1421], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0209, 0.0207, 0.0193, 0.0209, 0.0220, 0.0176, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:32:47,828 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.831e+02 3.142e+02 3.611e+02 4.263e+02 9.074e+02, threshold=7.223e+02, percent-clipped=1.0 2023-03-07 19:33:02,164 INFO [train2.py:809] (0/4) Epoch 6, batch 1100, loss[ctc_loss=0.2328, att_loss=0.3321, loss=0.3122, over 14053.00 frames. utt_duration=389.2 frames, utt_pad_proportion=0.3231, over 145.00 utterances.], tot_loss[ctc_loss=0.1492, att_loss=0.2745, loss=0.2494, over 3257292.60 frames. utt_duration=1245 frames, utt_pad_proportion=0.05577, over 10476.20 utterances.], batch size: 145, lr: 1.82e-02, grad_scale: 8.0 2023-03-07 19:33:20,030 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9908, 4.0329, 4.2082, 3.8904, 2.1219, 4.2065, 2.8159, 2.1915], device='cuda:0'), covar=tensor([0.0270, 0.0202, 0.0732, 0.0398, 0.2666, 0.0307, 0.1641, 0.1838], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0101, 0.0257, 0.0114, 0.0233, 0.0110, 0.0230, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:33:40,398 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21044.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:33:52,751 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21052.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:34:22,325 INFO [train2.py:809] (0/4) Epoch 6, batch 1150, loss[ctc_loss=0.1592, att_loss=0.2722, loss=0.2496, over 16552.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005612, over 45.00 utterances.], tot_loss[ctc_loss=0.1493, att_loss=0.2737, loss=0.2488, over 3252341.10 frames. utt_duration=1250 frames, utt_pad_proportion=0.05504, over 10417.06 utterances.], batch size: 45, lr: 1.82e-02, grad_scale: 8.0 2023-03-07 19:34:28,563 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21074.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:34:40,471 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.94 vs. limit=5.0 2023-03-07 19:34:57,176 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21092.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:35:27,185 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.084e+02 3.236e+02 4.040e+02 5.073e+02 1.159e+03, threshold=8.079e+02, percent-clipped=5.0 2023-03-07 19:35:29,771 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2023-03-07 19:35:30,908 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21113.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:35:39,367 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21118.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:35:41,977 INFO [train2.py:809] (0/4) Epoch 6, batch 1200, loss[ctc_loss=0.153, att_loss=0.2639, loss=0.2417, over 16180.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006991, over 41.00 utterances.], tot_loss[ctc_loss=0.1491, att_loss=0.2736, loss=0.2487, over 3261245.29 frames. utt_duration=1238 frames, utt_pad_proportion=0.05657, over 10553.66 utterances.], batch size: 41, lr: 1.82e-02, grad_scale: 8.0 2023-03-07 19:35:45,289 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21122.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:36:12,625 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8933, 3.8156, 3.0819, 3.3521, 3.9901, 3.4211, 2.0804, 4.4771], device='cuda:0'), covar=tensor([0.0990, 0.0385, 0.1080, 0.0705, 0.0513, 0.0714, 0.1341, 0.0317], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0137, 0.0182, 0.0148, 0.0170, 0.0176, 0.0156, 0.0166], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 19:37:02,262 INFO [train2.py:809] (0/4) Epoch 6, batch 1250, loss[ctc_loss=0.1332, att_loss=0.2763, loss=0.2476, over 16758.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006308, over 48.00 utterances.], tot_loss[ctc_loss=0.149, att_loss=0.2738, loss=0.2488, over 3260464.86 frames. utt_duration=1225 frames, utt_pad_proportion=0.06097, over 10658.11 utterances.], batch size: 48, lr: 1.82e-02, grad_scale: 8.0 2023-03-07 19:37:17,154 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21179.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:38:08,270 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.094e+02 3.230e+02 3.853e+02 4.858e+02 1.560e+03, threshold=7.707e+02, percent-clipped=4.0 2023-03-07 19:38:22,118 INFO [train2.py:809] (0/4) Epoch 6, batch 1300, loss[ctc_loss=0.1506, att_loss=0.2859, loss=0.2589, over 17233.00 frames. utt_duration=1170 frames, utt_pad_proportion=0.02835, over 59.00 utterances.], tot_loss[ctc_loss=0.1486, att_loss=0.2736, loss=0.2486, over 3254403.17 frames. utt_duration=1229 frames, utt_pad_proportion=0.06137, over 10602.25 utterances.], batch size: 59, lr: 1.81e-02, grad_scale: 8.0 2023-03-07 19:38:59,951 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21244.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:39:42,388 INFO [train2.py:809] (0/4) Epoch 6, batch 1350, loss[ctc_loss=0.1706, att_loss=0.2957, loss=0.2707, over 17131.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01438, over 56.00 utterances.], tot_loss[ctc_loss=0.1476, att_loss=0.2727, loss=0.2476, over 3259340.82 frames. utt_duration=1251 frames, utt_pad_proportion=0.05637, over 10434.73 utterances.], batch size: 56, lr: 1.81e-02, grad_scale: 8.0 2023-03-07 19:39:49,140 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2304, 5.1394, 5.0319, 2.8249, 1.9479, 2.8262, 4.9181, 3.8512], device='cuda:0'), covar=tensor([0.0490, 0.0139, 0.0171, 0.2540, 0.6040, 0.2427, 0.0202, 0.1701], device='cuda:0'), in_proj_covar=tensor([0.0284, 0.0174, 0.0203, 0.0178, 0.0354, 0.0334, 0.0187, 0.0324], device='cuda:0'), out_proj_covar=tensor([1.4415e-04, 7.5677e-05, 9.2457e-05, 8.0913e-05, 1.7116e-04, 1.5011e-04, 8.0020e-05, 1.5402e-04], device='cuda:0') 2023-03-07 19:39:49,598 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-03-07 19:40:17,770 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21292.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:40:48,695 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.207e+02 3.220e+02 3.890e+02 4.663e+02 1.481e+03, threshold=7.781e+02, percent-clipped=4.0 2023-03-07 19:40:50,584 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21312.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:41:02,794 INFO [train2.py:809] (0/4) Epoch 6, batch 1400, loss[ctc_loss=0.1264, att_loss=0.2683, loss=0.24, over 17365.00 frames. utt_duration=880.6 frames, utt_pad_proportion=0.07889, over 79.00 utterances.], tot_loss[ctc_loss=0.1462, att_loss=0.2721, loss=0.247, over 3270929.53 frames. utt_duration=1263 frames, utt_pad_proportion=0.05152, over 10373.68 utterances.], batch size: 79, lr: 1.81e-02, grad_scale: 8.0 2023-03-07 19:41:15,990 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21328.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:41:59,864 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2021, 4.9754, 5.0490, 3.2130, 4.9084, 4.3944, 4.4011, 2.8116], device='cuda:0'), covar=tensor([0.0071, 0.0075, 0.0145, 0.0820, 0.0072, 0.0168, 0.0241, 0.1232], device='cuda:0'), in_proj_covar=tensor([0.0052, 0.0066, 0.0053, 0.0098, 0.0059, 0.0074, 0.0081, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 19:42:23,224 INFO [train2.py:809] (0/4) Epoch 6, batch 1450, loss[ctc_loss=0.1435, att_loss=0.2611, loss=0.2375, over 15388.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.01004, over 35.00 utterances.], tot_loss[ctc_loss=0.1469, att_loss=0.2726, loss=0.2474, over 3276631.92 frames. utt_duration=1247 frames, utt_pad_proportion=0.05365, over 10522.00 utterances.], batch size: 35, lr: 1.81e-02, grad_scale: 8.0 2023-03-07 19:42:28,307 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21373.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:42:54,028 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21389.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:43:24,879 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21408.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:43:29,215 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.143e+02 3.096e+02 3.926e+02 4.783e+02 1.162e+03, threshold=7.851e+02, percent-clipped=3.0 2023-03-07 19:43:32,050 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.01 vs. limit=5.0 2023-03-07 19:43:43,053 INFO [train2.py:809] (0/4) Epoch 6, batch 1500, loss[ctc_loss=0.1356, att_loss=0.2419, loss=0.2206, over 15337.00 frames. utt_duration=1754 frames, utt_pad_proportion=0.0127, over 35.00 utterances.], tot_loss[ctc_loss=0.1461, att_loss=0.2717, loss=0.2466, over 3264322.78 frames. utt_duration=1249 frames, utt_pad_proportion=0.05606, over 10466.62 utterances.], batch size: 35, lr: 1.81e-02, grad_scale: 8.0 2023-03-07 19:45:01,845 INFO [train2.py:809] (0/4) Epoch 6, batch 1550, loss[ctc_loss=0.1542, att_loss=0.2846, loss=0.2585, over 17330.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02304, over 59.00 utterances.], tot_loss[ctc_loss=0.1463, att_loss=0.272, loss=0.2468, over 3270295.06 frames. utt_duration=1249 frames, utt_pad_proportion=0.0537, over 10483.19 utterances.], batch size: 59, lr: 1.80e-02, grad_scale: 8.0 2023-03-07 19:45:04,783 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-07 19:45:08,940 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21474.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:45:40,217 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1126, 5.1002, 4.9860, 2.4560, 1.7061, 2.5247, 4.6997, 3.6375], device='cuda:0'), covar=tensor([0.0556, 0.0126, 0.0210, 0.3166, 0.6631, 0.2837, 0.0279, 0.2003], device='cuda:0'), in_proj_covar=tensor([0.0289, 0.0176, 0.0207, 0.0180, 0.0356, 0.0330, 0.0188, 0.0326], device='cuda:0'), out_proj_covar=tensor([1.4523e-04, 7.5359e-05, 9.3886e-05, 8.1891e-05, 1.7107e-04, 1.4801e-04, 8.0335e-05, 1.5413e-04], device='cuda:0') 2023-03-07 19:45:54,435 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6240, 4.2260, 4.3338, 4.6601, 4.9142, 4.3037, 4.3207, 1.8951], device='cuda:0'), covar=tensor([0.0305, 0.0643, 0.0454, 0.0171, 0.1144, 0.0372, 0.0429, 0.2944], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0124, 0.0121, 0.0124, 0.0300, 0.0129, 0.0116, 0.0237], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 19:46:07,491 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.008e+02 3.352e+02 3.791e+02 4.457e+02 9.611e+02, threshold=7.583e+02, percent-clipped=1.0 2023-03-07 19:46:21,675 INFO [train2.py:809] (0/4) Epoch 6, batch 1600, loss[ctc_loss=0.1358, att_loss=0.273, loss=0.2456, over 17030.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007322, over 51.00 utterances.], tot_loss[ctc_loss=0.1456, att_loss=0.2712, loss=0.2461, over 3269681.64 frames. utt_duration=1274 frames, utt_pad_proportion=0.04941, over 10280.76 utterances.], batch size: 51, lr: 1.80e-02, grad_scale: 8.0 2023-03-07 19:47:41,020 INFO [train2.py:809] (0/4) Epoch 6, batch 1650, loss[ctc_loss=0.181, att_loss=0.3117, loss=0.2855, over 17104.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01582, over 56.00 utterances.], tot_loss[ctc_loss=0.1453, att_loss=0.2715, loss=0.2463, over 3270359.01 frames. utt_duration=1262 frames, utt_pad_proportion=0.05342, over 10374.67 utterances.], batch size: 56, lr: 1.80e-02, grad_scale: 8.0 2023-03-07 19:47:50,968 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7841, 2.5496, 3.5879, 4.6259, 4.2312, 4.5304, 3.0882, 2.1239], device='cuda:0'), covar=tensor([0.0536, 0.2529, 0.1043, 0.0399, 0.0554, 0.0211, 0.1673, 0.2497], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0200, 0.0192, 0.0156, 0.0149, 0.0124, 0.0194, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 19:48:20,312 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21595.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:48:47,370 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.750e+02 3.087e+02 4.072e+02 5.275e+02 1.721e+03, threshold=8.145e+02, percent-clipped=7.0 2023-03-07 19:48:50,752 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21613.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:48:55,523 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7127, 1.5525, 2.1023, 1.7902, 2.0360, 1.6211, 2.1612, 2.0759], device='cuda:0'), covar=tensor([0.0435, 0.4228, 0.2832, 0.2106, 0.1671, 0.3030, 0.1864, 0.2685], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0102, 0.0098, 0.0086, 0.0085, 0.0086, 0.0094, 0.0076], device='cuda:0'), out_proj_covar=tensor([4.0343e-05, 5.5024e-05, 5.3168e-05, 4.4320e-05, 4.0913e-05, 4.8541e-05, 5.1059e-05, 4.4085e-05], device='cuda:0') 2023-03-07 19:49:01,645 INFO [train2.py:809] (0/4) Epoch 6, batch 1700, loss[ctc_loss=0.0905, att_loss=0.2268, loss=0.1995, over 15877.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.01002, over 39.00 utterances.], tot_loss[ctc_loss=0.1454, att_loss=0.2716, loss=0.2464, over 3271405.41 frames. utt_duration=1276 frames, utt_pad_proportion=0.0496, over 10269.33 utterances.], batch size: 39, lr: 1.80e-02, grad_scale: 8.0 2023-03-07 19:50:00,458 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21656.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:50:14,519 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5841, 5.0746, 4.7489, 5.1205, 5.1096, 4.6840, 4.2151, 5.0823], device='cuda:0'), covar=tensor([0.0078, 0.0092, 0.0096, 0.0060, 0.0075, 0.0088, 0.0328, 0.0118], device='cuda:0'), in_proj_covar=tensor([0.0057, 0.0056, 0.0063, 0.0044, 0.0044, 0.0053, 0.0076, 0.0073], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-07 19:50:19,144 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21668.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:50:22,101 INFO [train2.py:809] (0/4) Epoch 6, batch 1750, loss[ctc_loss=0.1496, att_loss=0.2717, loss=0.2473, over 17444.00 frames. utt_duration=706.6 frames, utt_pad_proportion=0.1135, over 99.00 utterances.], tot_loss[ctc_loss=0.1453, att_loss=0.2717, loss=0.2464, over 3273029.96 frames. utt_duration=1278 frames, utt_pad_proportion=0.04832, over 10258.17 utterances.], batch size: 99, lr: 1.80e-02, grad_scale: 8.0 2023-03-07 19:50:29,320 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21674.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:50:44,629 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21684.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:50:51,632 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-07 19:51:23,942 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21708.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:51:28,243 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.826e+02 3.408e+02 4.076e+02 5.243e+02 8.515e+02, threshold=8.152e+02, percent-clipped=2.0 2023-03-07 19:51:36,023 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8295, 4.4356, 4.3803, 4.3159, 4.4859, 4.4698, 4.3389, 4.1197], device='cuda:0'), covar=tensor([0.1466, 0.0580, 0.0272, 0.0527, 0.0401, 0.0409, 0.0321, 0.0370], device='cuda:0'), in_proj_covar=tensor([0.0402, 0.0236, 0.0175, 0.0215, 0.0274, 0.0304, 0.0230, 0.0258], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 19:51:42,602 INFO [train2.py:809] (0/4) Epoch 6, batch 1800, loss[ctc_loss=0.135, att_loss=0.2656, loss=0.2394, over 16533.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006842, over 45.00 utterances.], tot_loss[ctc_loss=0.146, att_loss=0.2718, loss=0.2466, over 3269483.17 frames. utt_duration=1268 frames, utt_pad_proportion=0.04992, over 10322.51 utterances.], batch size: 45, lr: 1.79e-02, grad_scale: 8.0 2023-03-07 19:52:39,936 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21756.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:52:51,299 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21763.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:53:02,476 INFO [train2.py:809] (0/4) Epoch 6, batch 1850, loss[ctc_loss=0.1237, att_loss=0.2459, loss=0.2215, over 15883.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009534, over 39.00 utterances.], tot_loss[ctc_loss=0.1458, att_loss=0.2719, loss=0.2467, over 3273163.50 frames. utt_duration=1284 frames, utt_pad_proportion=0.04656, over 10208.32 utterances.], batch size: 39, lr: 1.79e-02, grad_scale: 8.0 2023-03-07 19:53:09,196 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21774.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:53:26,288 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7098, 5.1467, 4.8523, 5.0950, 5.1520, 4.8693, 4.2298, 5.0206], device='cuda:0'), covar=tensor([0.0065, 0.0091, 0.0069, 0.0066, 0.0070, 0.0068, 0.0380, 0.0148], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0055, 0.0063, 0.0044, 0.0044, 0.0052, 0.0076, 0.0073], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-07 19:54:09,692 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.389e+02 3.296e+02 3.946e+02 5.007e+02 1.050e+03, threshold=7.891e+02, percent-clipped=3.0 2023-03-07 19:54:24,813 INFO [train2.py:809] (0/4) Epoch 6, batch 1900, loss[ctc_loss=0.1331, att_loss=0.264, loss=0.2378, over 17432.00 frames. utt_duration=884.1 frames, utt_pad_proportion=0.07517, over 79.00 utterances.], tot_loss[ctc_loss=0.1455, att_loss=0.2718, loss=0.2465, over 3275651.13 frames. utt_duration=1270 frames, utt_pad_proportion=0.04904, over 10331.81 utterances.], batch size: 79, lr: 1.79e-02, grad_scale: 8.0 2023-03-07 19:54:28,002 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=21822.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:54:31,357 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21824.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:54:54,372 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-07 19:55:14,604 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21850.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:55:47,727 INFO [train2.py:809] (0/4) Epoch 6, batch 1950, loss[ctc_loss=0.1276, att_loss=0.246, loss=0.2223, over 15509.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.007987, over 36.00 utterances.], tot_loss[ctc_loss=0.1459, att_loss=0.2719, loss=0.2467, over 3265745.67 frames. utt_duration=1265 frames, utt_pad_proportion=0.05164, over 10335.86 utterances.], batch size: 36, lr: 1.79e-02, grad_scale: 8.0 2023-03-07 19:56:54,340 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.830e+02 3.406e+02 4.124e+02 5.166e+02 8.214e+02, threshold=8.249e+02, percent-clipped=4.0 2023-03-07 19:56:54,761 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=21911.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:57:09,081 INFO [train2.py:809] (0/4) Epoch 6, batch 2000, loss[ctc_loss=0.1161, att_loss=0.2685, loss=0.238, over 16615.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006126, over 47.00 utterances.], tot_loss[ctc_loss=0.1467, att_loss=0.273, loss=0.2478, over 3268537.78 frames. utt_duration=1251 frames, utt_pad_proportion=0.05363, over 10461.50 utterances.], batch size: 47, lr: 1.79e-02, grad_scale: 8.0 2023-03-07 19:57:35,516 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-07 19:57:59,923 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21951.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:58:27,690 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21968.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:58:29,190 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=21969.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:58:30,579 INFO [train2.py:809] (0/4) Epoch 6, batch 2050, loss[ctc_loss=0.1246, att_loss=0.264, loss=0.2361, over 16534.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.005841, over 45.00 utterances.], tot_loss[ctc_loss=0.1461, att_loss=0.2724, loss=0.2472, over 3272210.50 frames. utt_duration=1259 frames, utt_pad_proportion=0.05102, over 10404.81 utterances.], batch size: 45, lr: 1.78e-02, grad_scale: 8.0 2023-03-07 19:58:30,974 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=21970.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:58:53,143 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=21984.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:59:20,918 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-22000.pt 2023-03-07 19:59:42,650 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.988e+02 3.148e+02 3.962e+02 4.824e+02 9.666e+02, threshold=7.923e+02, percent-clipped=1.0 2023-03-07 19:59:51,261 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22016.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 19:59:57,611 INFO [train2.py:809] (0/4) Epoch 6, batch 2100, loss[ctc_loss=0.1538, att_loss=0.283, loss=0.2571, over 17047.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01011, over 53.00 utterances.], tot_loss[ctc_loss=0.145, att_loss=0.2711, loss=0.2459, over 3264731.55 frames. utt_duration=1281 frames, utt_pad_proportion=0.04846, over 10203.60 utterances.], batch size: 53, lr: 1.78e-02, grad_scale: 8.0 2023-03-07 20:00:07,232 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8714, 4.9080, 4.7198, 4.8108, 5.2619, 4.8746, 4.7125, 2.1471], device='cuda:0'), covar=tensor([0.0169, 0.0153, 0.0151, 0.0123, 0.0856, 0.0174, 0.0180, 0.2663], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0121, 0.0122, 0.0125, 0.0295, 0.0128, 0.0114, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 20:00:15,230 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22031.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:00:16,529 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22032.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:00:21,470 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4146, 2.4699, 3.8618, 2.5481, 3.4844, 4.7404, 4.6061, 3.1258], device='cuda:0'), covar=tensor([0.0456, 0.2124, 0.0878, 0.1628, 0.1018, 0.0550, 0.0367, 0.1529], device='cuda:0'), in_proj_covar=tensor([0.0213, 0.0214, 0.0213, 0.0195, 0.0217, 0.0224, 0.0172, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 20:01:20,714 INFO [train2.py:809] (0/4) Epoch 6, batch 2150, loss[ctc_loss=0.2016, att_loss=0.3161, loss=0.2932, over 17001.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.008792, over 51.00 utterances.], tot_loss[ctc_loss=0.1457, att_loss=0.2715, loss=0.2463, over 3255123.05 frames. utt_duration=1283 frames, utt_pad_proportion=0.05009, over 10163.21 utterances.], batch size: 51, lr: 1.78e-02, grad_scale: 8.0 2023-03-07 20:02:28,349 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.159e+02 3.198e+02 3.917e+02 4.536e+02 8.271e+02, threshold=7.834e+02, percent-clipped=2.0 2023-03-07 20:02:42,266 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22119.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:02:43,572 INFO [train2.py:809] (0/4) Epoch 6, batch 2200, loss[ctc_loss=0.1773, att_loss=0.2958, loss=0.2721, over 17332.00 frames. utt_duration=879.2 frames, utt_pad_proportion=0.07934, over 79.00 utterances.], tot_loss[ctc_loss=0.1452, att_loss=0.2715, loss=0.2462, over 3254619.91 frames. utt_duration=1263 frames, utt_pad_proportion=0.05393, over 10323.04 utterances.], batch size: 79, lr: 1.78e-02, grad_scale: 8.0 2023-03-07 20:02:59,610 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22130.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:04:05,340 INFO [train2.py:809] (0/4) Epoch 6, batch 2250, loss[ctc_loss=0.1162, att_loss=0.2432, loss=0.2178, over 16029.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006656, over 40.00 utterances.], tot_loss[ctc_loss=0.1447, att_loss=0.2712, loss=0.2459, over 3261963.89 frames. utt_duration=1262 frames, utt_pad_proportion=0.05253, over 10350.24 utterances.], batch size: 40, lr: 1.78e-02, grad_scale: 8.0 2023-03-07 20:04:14,633 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22176.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:04:38,843 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22191.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:05:03,078 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22206.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:05:11,249 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.756e+02 2.999e+02 3.625e+02 4.758e+02 1.652e+03, threshold=7.250e+02, percent-clipped=4.0 2023-03-07 20:05:16,214 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9922, 4.9312, 4.9221, 2.6073, 4.7052, 4.1383, 3.8381, 2.3523], device='cuda:0'), covar=tensor([0.0096, 0.0094, 0.0142, 0.1188, 0.0096, 0.0233, 0.0386, 0.1618], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0067, 0.0054, 0.0100, 0.0060, 0.0077, 0.0082, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 20:05:17,898 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22215.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:05:25,447 INFO [train2.py:809] (0/4) Epoch 6, batch 2300, loss[ctc_loss=0.1397, att_loss=0.2654, loss=0.2403, over 15986.00 frames. utt_duration=1600 frames, utt_pad_proportion=0.00806, over 40.00 utterances.], tot_loss[ctc_loss=0.1455, att_loss=0.2717, loss=0.2465, over 3262514.96 frames. utt_duration=1254 frames, utt_pad_proportion=0.05581, over 10417.19 utterances.], batch size: 40, lr: 1.77e-02, grad_scale: 16.0 2023-03-07 20:05:52,507 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:05:52,534 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22237.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:06:15,978 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22251.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:06:36,212 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-03-07 20:06:44,737 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22269.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:06:46,016 INFO [train2.py:809] (0/4) Epoch 6, batch 2350, loss[ctc_loss=0.1319, att_loss=0.2683, loss=0.241, over 16313.00 frames. utt_duration=1451 frames, utt_pad_proportion=0.00722, over 45.00 utterances.], tot_loss[ctc_loss=0.144, att_loss=0.2711, loss=0.2457, over 3266035.55 frames. utt_duration=1259 frames, utt_pad_proportion=0.05439, over 10389.83 utterances.], batch size: 45, lr: 1.77e-02, grad_scale: 16.0 2023-03-07 20:06:56,059 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22276.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:07:30,844 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8402, 4.9264, 4.8897, 3.1224, 4.7346, 4.2056, 4.0526, 2.3981], device='cuda:0'), covar=tensor([0.0195, 0.0075, 0.0146, 0.0870, 0.0073, 0.0204, 0.0315, 0.1491], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0068, 0.0055, 0.0101, 0.0060, 0.0078, 0.0082, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 20:07:30,941 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22298.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:07:32,203 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22299.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:07:45,758 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5633, 4.5286, 4.5341, 4.5878, 4.8686, 4.6402, 4.5050, 2.0778], device='cuda:0'), covar=tensor([0.0270, 0.0380, 0.0272, 0.0169, 0.1489, 0.0241, 0.0315, 0.3064], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0126, 0.0125, 0.0125, 0.0303, 0.0130, 0.0117, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 20:07:52,172 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.236e+02 3.128e+02 3.825e+02 4.605e+02 9.373e+02, threshold=7.650e+02, percent-clipped=4.0 2023-03-07 20:08:01,631 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22317.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:08:06,062 INFO [train2.py:809] (0/4) Epoch 6, batch 2400, loss[ctc_loss=0.1204, att_loss=0.2605, loss=0.2324, over 16490.00 frames. utt_duration=1436 frames, utt_pad_proportion=0.004972, over 46.00 utterances.], tot_loss[ctc_loss=0.1448, att_loss=0.2723, loss=0.2468, over 3279201.17 frames. utt_duration=1262 frames, utt_pad_proportion=0.05061, over 10407.50 utterances.], batch size: 46, lr: 1.77e-02, grad_scale: 16.0 2023-03-07 20:08:15,547 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22326.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:09:26,763 INFO [train2.py:809] (0/4) Epoch 6, batch 2450, loss[ctc_loss=0.1046, att_loss=0.2526, loss=0.223, over 16539.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006347, over 45.00 utterances.], tot_loss[ctc_loss=0.1468, att_loss=0.2735, loss=0.2481, over 3279831.45 frames. utt_duration=1245 frames, utt_pad_proportion=0.05508, over 10549.11 utterances.], batch size: 45, lr: 1.77e-02, grad_scale: 16.0 2023-03-07 20:09:39,365 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5301, 4.9093, 4.8134, 4.7006, 4.9550, 4.8923, 4.5382, 4.4210], device='cuda:0'), covar=tensor([0.1186, 0.0440, 0.0260, 0.0678, 0.0310, 0.0337, 0.0372, 0.0369], device='cuda:0'), in_proj_covar=tensor([0.0402, 0.0231, 0.0171, 0.0219, 0.0274, 0.0298, 0.0230, 0.0258], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 20:09:51,353 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22385.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:10:21,615 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-07 20:10:34,026 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.011e+02 3.249e+02 3.944e+02 4.920e+02 7.510e+02, threshold=7.889e+02, percent-clipped=0.0 2023-03-07 20:10:47,092 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22419.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:10:48,445 INFO [train2.py:809] (0/4) Epoch 6, batch 2500, loss[ctc_loss=0.1274, att_loss=0.2474, loss=0.2234, over 14547.00 frames. utt_duration=1820 frames, utt_pad_proportion=0.0462, over 32.00 utterances.], tot_loss[ctc_loss=0.1462, att_loss=0.2726, loss=0.2473, over 3274045.28 frames. utt_duration=1237 frames, utt_pad_proportion=0.0586, over 10601.59 utterances.], batch size: 32, lr: 1.77e-02, grad_scale: 16.0 2023-03-07 20:11:07,001 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-07 20:11:20,601 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0964, 5.0709, 4.9828, 2.7313, 1.8092, 2.5773, 4.7693, 3.6949], device='cuda:0'), covar=tensor([0.0596, 0.0179, 0.0221, 0.3013, 0.6224, 0.2824, 0.0294, 0.2081], device='cuda:0'), in_proj_covar=tensor([0.0296, 0.0181, 0.0214, 0.0182, 0.0362, 0.0338, 0.0197, 0.0333], device='cuda:0'), out_proj_covar=tensor([1.4849e-04, 7.6467e-05, 9.5606e-05, 8.2879e-05, 1.7245e-04, 1.5047e-04, 8.4146e-05, 1.5658e-04], device='cuda:0') 2023-03-07 20:11:30,397 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22446.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:12:04,119 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22467.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:12:08,535 INFO [train2.py:809] (0/4) Epoch 6, batch 2550, loss[ctc_loss=0.148, att_loss=0.2852, loss=0.2577, over 16970.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.006725, over 50.00 utterances.], tot_loss[ctc_loss=0.1464, att_loss=0.2731, loss=0.2478, over 3284519.48 frames. utt_duration=1239 frames, utt_pad_proportion=0.05495, over 10617.06 utterances.], batch size: 50, lr: 1.76e-02, grad_scale: 16.0 2023-03-07 20:12:34,610 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22486.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:12:43,100 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3377, 4.5774, 4.3228, 4.4484, 2.2505, 4.5379, 2.1994, 2.0232], device='cuda:0'), covar=tensor([0.0218, 0.0158, 0.0899, 0.0212, 0.2540, 0.0214, 0.2135, 0.1883], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0099, 0.0252, 0.0109, 0.0225, 0.0100, 0.0227, 0.0206], device='cuda:0'), out_proj_covar=tensor([1.0887e-04, 9.9263e-05, 2.2323e-04, 9.9936e-05, 2.0607e-04, 9.7233e-05, 2.0083e-04, 1.8436e-04], device='cuda:0') 2023-03-07 20:13:07,433 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22506.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:13:14,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.039e+02 3.216e+02 3.975e+02 4.594e+02 1.058e+03, threshold=7.951e+02, percent-clipped=4.0 2023-03-07 20:13:28,844 INFO [train2.py:809] (0/4) Epoch 6, batch 2600, loss[ctc_loss=0.14, att_loss=0.2777, loss=0.2502, over 16629.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.00523, over 47.00 utterances.], tot_loss[ctc_loss=0.1467, att_loss=0.2737, loss=0.2483, over 3283678.26 frames. utt_duration=1216 frames, utt_pad_proportion=0.06161, over 10817.63 utterances.], batch size: 47, lr: 1.76e-02, grad_scale: 16.0 2023-03-07 20:13:47,881 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22532.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:14:00,638 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.62 vs. limit=5.0 2023-03-07 20:14:06,674 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7333, 4.9549, 4.4211, 5.1239, 4.3933, 4.8297, 5.1506, 4.9776], device='cuda:0'), covar=tensor([0.0366, 0.0260, 0.0810, 0.0184, 0.0519, 0.0228, 0.0298, 0.0158], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0190, 0.0251, 0.0169, 0.0208, 0.0161, 0.0185, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 20:14:24,993 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22554.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:14:51,054 INFO [train2.py:809] (0/4) Epoch 6, batch 2650, loss[ctc_loss=0.1127, att_loss=0.2558, loss=0.2272, over 16768.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006466, over 48.00 utterances.], tot_loss[ctc_loss=0.1458, att_loss=0.2736, loss=0.248, over 3289201.94 frames. utt_duration=1215 frames, utt_pad_proportion=0.05928, over 10844.89 utterances.], batch size: 48, lr: 1.76e-02, grad_scale: 16.0 2023-03-07 20:14:52,700 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22571.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:15:28,363 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22593.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:15:34,307 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.50 vs. limit=5.0 2023-03-07 20:15:59,449 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.951e+02 3.146e+02 3.715e+02 4.789e+02 1.567e+03, threshold=7.430e+02, percent-clipped=3.0 2023-03-07 20:16:09,925 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-03-07 20:16:11,932 INFO [train2.py:809] (0/4) Epoch 6, batch 2700, loss[ctc_loss=0.1424, att_loss=0.2889, loss=0.2596, over 16622.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005528, over 47.00 utterances.], tot_loss[ctc_loss=0.1458, att_loss=0.2731, loss=0.2476, over 3288079.72 frames. utt_duration=1199 frames, utt_pad_proportion=0.06317, over 10979.48 utterances.], batch size: 47, lr: 1.76e-02, grad_scale: 8.0 2023-03-07 20:16:21,349 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22626.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:16:41,838 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22639.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:17:32,380 INFO [train2.py:809] (0/4) Epoch 6, batch 2750, loss[ctc_loss=0.1439, att_loss=0.2792, loss=0.2521, over 16633.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.00499, over 47.00 utterances.], tot_loss[ctc_loss=0.147, att_loss=0.2734, loss=0.2481, over 3288831.44 frames. utt_duration=1198 frames, utt_pad_proportion=0.0631, over 10998.92 utterances.], batch size: 47, lr: 1.76e-02, grad_scale: 8.0 2023-03-07 20:17:38,674 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22674.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:18:21,029 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22700.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 20:18:40,004 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.820e+02 3.125e+02 3.816e+02 4.658e+02 9.758e+02, threshold=7.632e+02, percent-clipped=3.0 2023-03-07 20:18:52,609 INFO [train2.py:809] (0/4) Epoch 6, batch 2800, loss[ctc_loss=0.181, att_loss=0.2983, loss=0.2748, over 17055.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008897, over 52.00 utterances.], tot_loss[ctc_loss=0.1462, att_loss=0.2733, loss=0.2478, over 3289171.93 frames. utt_duration=1201 frames, utt_pad_proportion=0.06262, over 10968.66 utterances.], batch size: 52, lr: 1.76e-02, grad_scale: 8.0 2023-03-07 20:19:13,020 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-03-07 20:19:26,404 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22741.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:20:11,899 INFO [train2.py:809] (0/4) Epoch 6, batch 2850, loss[ctc_loss=0.1488, att_loss=0.2717, loss=0.2472, over 15954.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006331, over 41.00 utterances.], tot_loss[ctc_loss=0.1462, att_loss=0.2729, loss=0.2475, over 3284560.50 frames. utt_duration=1201 frames, utt_pad_proportion=0.06383, over 10957.39 utterances.], batch size: 41, lr: 1.75e-02, grad_scale: 8.0 2023-03-07 20:20:22,070 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4587, 2.4097, 4.9495, 3.5405, 2.7688, 4.2289, 4.6058, 4.6186], device='cuda:0'), covar=tensor([0.0226, 0.1889, 0.0161, 0.1601, 0.2537, 0.0318, 0.0132, 0.0238], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0241, 0.0124, 0.0300, 0.0294, 0.0182, 0.0105, 0.0139], device='cuda:0'), out_proj_covar=tensor([1.2450e-04, 1.9637e-04, 1.0978e-04, 2.4270e-04, 2.4806e-04, 1.5825e-04, 9.4408e-05, 1.2492e-04], device='cuda:0') 2023-03-07 20:20:28,225 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9736, 3.8596, 2.9583, 3.4065, 3.7872, 3.5837, 2.6967, 4.4468], device='cuda:0'), covar=tensor([0.0983, 0.0393, 0.1209, 0.0653, 0.0557, 0.0637, 0.0944, 0.0380], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0143, 0.0189, 0.0155, 0.0176, 0.0184, 0.0159, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 20:20:38,173 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22786.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:20:51,813 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22794.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:21:10,121 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1240, 4.9740, 5.0185, 3.1632, 4.7714, 4.4399, 4.0922, 2.3915], device='cuda:0'), covar=tensor([0.0118, 0.0092, 0.0180, 0.0907, 0.0091, 0.0168, 0.0312, 0.1605], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0067, 0.0057, 0.0099, 0.0060, 0.0077, 0.0082, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 20:21:20,768 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.881e+02 3.039e+02 3.742e+02 4.818e+02 9.718e+02, threshold=7.483e+02, percent-clipped=5.0 2023-03-07 20:21:33,090 INFO [train2.py:809] (0/4) Epoch 6, batch 2900, loss[ctc_loss=0.1413, att_loss=0.2733, loss=0.2469, over 17210.00 frames. utt_duration=872.8 frames, utt_pad_proportion=0.08512, over 79.00 utterances.], tot_loss[ctc_loss=0.1461, att_loss=0.2725, loss=0.2473, over 3275177.67 frames. utt_duration=1181 frames, utt_pad_proportion=0.07155, over 11105.60 utterances.], batch size: 79, lr: 1.75e-02, grad_scale: 8.0 2023-03-07 20:21:41,134 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=22825.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:21:52,475 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22832.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:21:55,389 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22834.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:22:30,792 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22855.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:22:45,472 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.45 vs. limit=5.0 2023-03-07 20:22:53,777 INFO [train2.py:809] (0/4) Epoch 6, batch 2950, loss[ctc_loss=0.1068, att_loss=0.2462, loss=0.2183, over 16171.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.00655, over 41.00 utterances.], tot_loss[ctc_loss=0.144, att_loss=0.2711, loss=0.2456, over 3273885.47 frames. utt_duration=1192 frames, utt_pad_proportion=0.0689, over 10997.87 utterances.], batch size: 41, lr: 1.75e-02, grad_scale: 8.0 2023-03-07 20:22:55,580 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22871.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:23:10,175 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22880.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:23:20,710 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=22886.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:23:32,143 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=22893.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:24:02,046 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.974e+02 3.285e+02 3.915e+02 4.917e+02 1.190e+03, threshold=7.831e+02, percent-clipped=3.0 2023-03-07 20:24:12,916 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22919.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:24:14,352 INFO [train2.py:809] (0/4) Epoch 6, batch 3000, loss[ctc_loss=0.2415, att_loss=0.3252, loss=0.3085, over 13719.00 frames. utt_duration=379.9 frames, utt_pad_proportion=0.3393, over 145.00 utterances.], tot_loss[ctc_loss=0.1457, att_loss=0.2726, loss=0.2472, over 3276598.26 frames. utt_duration=1169 frames, utt_pad_proportion=0.07381, over 11230.01 utterances.], batch size: 145, lr: 1.75e-02, grad_scale: 8.0 2023-03-07 20:24:14,355 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 20:24:28,229 INFO [train2.py:843] (0/4) Epoch 6, validation: ctc_loss=0.06806, att_loss=0.2473, loss=0.2115, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 20:24:28,230 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16035MB 2023-03-07 20:25:01,615 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=22941.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:25:01,783 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7924, 5.1821, 4.5738, 5.2582, 4.6346, 4.8603, 5.3576, 5.1713], device='cuda:0'), covar=tensor([0.0529, 0.0269, 0.0813, 0.0146, 0.0448, 0.0247, 0.0210, 0.0142], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0195, 0.0255, 0.0175, 0.0214, 0.0165, 0.0187, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 20:25:32,139 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9669, 4.0454, 3.2696, 3.6485, 4.1490, 3.9051, 2.7610, 4.6814], device='cuda:0'), covar=tensor([0.0962, 0.0381, 0.0993, 0.0568, 0.0465, 0.0543, 0.0924, 0.0332], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0142, 0.0189, 0.0155, 0.0178, 0.0185, 0.0159, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 20:25:47,896 INFO [train2.py:809] (0/4) Epoch 6, batch 3050, loss[ctc_loss=0.1124, att_loss=0.2523, loss=0.2244, over 16124.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.005868, over 42.00 utterances.], tot_loss[ctc_loss=0.1452, att_loss=0.2724, loss=0.2469, over 3273306.64 frames. utt_duration=1202 frames, utt_pad_proportion=0.06623, over 10906.00 utterances.], batch size: 42, lr: 1.75e-02, grad_scale: 8.0 2023-03-07 20:26:27,879 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=22995.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 20:26:55,429 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.122e+02 3.380e+02 4.169e+02 5.152e+02 1.227e+03, threshold=8.338e+02, percent-clipped=8.0 2023-03-07 20:27:07,727 INFO [train2.py:809] (0/4) Epoch 6, batch 3100, loss[ctc_loss=0.1238, att_loss=0.2666, loss=0.2381, over 16957.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008079, over 50.00 utterances.], tot_loss[ctc_loss=0.1464, att_loss=0.2727, loss=0.2475, over 3273167.22 frames. utt_duration=1202 frames, utt_pad_proportion=0.06709, over 10905.94 utterances.], batch size: 50, lr: 1.74e-02, grad_scale: 8.0 2023-03-07 20:27:42,212 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23041.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:28:28,019 INFO [train2.py:809] (0/4) Epoch 6, batch 3150, loss[ctc_loss=0.1499, att_loss=0.2801, loss=0.2541, over 17047.00 frames. utt_duration=864.6 frames, utt_pad_proportion=0.09465, over 79.00 utterances.], tot_loss[ctc_loss=0.147, att_loss=0.2731, loss=0.2479, over 3271571.11 frames. utt_duration=1197 frames, utt_pad_proportion=0.0686, over 10947.59 utterances.], batch size: 79, lr: 1.74e-02, grad_scale: 8.0 2023-03-07 20:28:59,010 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23089.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:29:29,571 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23108.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:29:36,089 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.237e+02 3.296e+02 3.877e+02 4.841e+02 1.237e+03, threshold=7.753e+02, percent-clipped=8.0 2023-03-07 20:29:48,526 INFO [train2.py:809] (0/4) Epoch 6, batch 3200, loss[ctc_loss=0.127, att_loss=0.2402, loss=0.2176, over 15620.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.01024, over 37.00 utterances.], tot_loss[ctc_loss=0.1464, att_loss=0.2731, loss=0.2477, over 3274297.54 frames. utt_duration=1198 frames, utt_pad_proportion=0.06842, over 10945.50 utterances.], batch size: 37, lr: 1.74e-02, grad_scale: 8.0 2023-03-07 20:30:36,817 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23150.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:30:41,660 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8669, 1.4922, 2.1724, 1.6845, 3.6876, 2.9611, 1.5565, 1.4199], device='cuda:0'), covar=tensor([0.0674, 0.3391, 0.2845, 0.1964, 0.0376, 0.0937, 0.3051, 0.2575], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0091, 0.0093, 0.0082, 0.0077, 0.0077, 0.0091, 0.0075], device='cuda:0'), out_proj_covar=tensor([3.9128e-05, 5.1613e-05, 5.2241e-05, 4.3874e-05, 3.8435e-05, 4.5031e-05, 5.0644e-05, 4.4327e-05], device='cuda:0') 2023-03-07 20:30:55,104 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5746, 2.4770, 4.9684, 3.7317, 3.1470, 4.4784, 4.5639, 4.7004], device='cuda:0'), covar=tensor([0.0164, 0.1839, 0.0112, 0.1184, 0.1977, 0.0235, 0.0124, 0.0188], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0236, 0.0122, 0.0296, 0.0290, 0.0182, 0.0103, 0.0137], device='cuda:0'), out_proj_covar=tensor([1.2289e-04, 1.9302e-04, 1.0788e-04, 2.4018e-04, 2.4489e-04, 1.5885e-04, 9.3180e-05, 1.2308e-04], device='cuda:0') 2023-03-07 20:31:07,686 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23169.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:31:08,931 INFO [train2.py:809] (0/4) Epoch 6, batch 3250, loss[ctc_loss=0.1252, att_loss=0.2438, loss=0.2201, over 14461.00 frames. utt_duration=1809 frames, utt_pad_proportion=0.03174, over 32.00 utterances.], tot_loss[ctc_loss=0.1468, att_loss=0.2736, loss=0.2482, over 3269273.09 frames. utt_duration=1173 frames, utt_pad_proportion=0.07612, over 11158.65 utterances.], batch size: 32, lr: 1.74e-02, grad_scale: 8.0 2023-03-07 20:31:26,728 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23181.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:32:00,201 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8527, 5.1879, 4.6801, 5.3127, 4.6448, 5.0240, 5.3754, 5.1799], device='cuda:0'), covar=tensor([0.0422, 0.0239, 0.0692, 0.0149, 0.0415, 0.0187, 0.0192, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0193, 0.0250, 0.0170, 0.0211, 0.0163, 0.0183, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 20:32:18,012 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.115e+02 3.273e+02 3.918e+02 4.726e+02 1.280e+03, threshold=7.837e+02, percent-clipped=1.0 2023-03-07 20:32:26,169 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23217.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:32:27,834 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1512, 4.4979, 4.4486, 4.5230, 2.4827, 4.7208, 2.5756, 1.6504], device='cuda:0'), covar=tensor([0.0223, 0.0130, 0.0749, 0.0235, 0.2256, 0.0183, 0.1725, 0.2234], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0104, 0.0256, 0.0112, 0.0231, 0.0102, 0.0227, 0.0211], device='cuda:0'), out_proj_covar=tensor([1.1347e-04, 1.0373e-04, 2.2818e-04, 1.0258e-04, 2.1178e-04, 9.8202e-05, 2.0167e-04, 1.8828e-04], device='cuda:0') 2023-03-07 20:32:30,541 INFO [train2.py:809] (0/4) Epoch 6, batch 3300, loss[ctc_loss=0.1459, att_loss=0.2446, loss=0.2249, over 15372.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01043, over 35.00 utterances.], tot_loss[ctc_loss=0.1457, att_loss=0.273, loss=0.2476, over 3265302.38 frames. utt_duration=1183 frames, utt_pad_proportion=0.07347, over 11051.01 utterances.], batch size: 35, lr: 1.74e-02, grad_scale: 8.0 2023-03-07 20:32:45,786 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23229.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:33:47,178 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5969, 4.9227, 4.4445, 5.1121, 4.3250, 4.7523, 5.1778, 4.9244], device='cuda:0'), covar=tensor([0.0491, 0.0268, 0.0708, 0.0161, 0.0549, 0.0261, 0.0213, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0191, 0.0246, 0.0167, 0.0211, 0.0162, 0.0180, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-07 20:33:51,764 INFO [train2.py:809] (0/4) Epoch 6, batch 3350, loss[ctc_loss=0.1338, att_loss=0.2739, loss=0.2458, over 16885.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.007209, over 49.00 utterances.], tot_loss[ctc_loss=0.1456, att_loss=0.2732, loss=0.2477, over 3276516.30 frames. utt_duration=1194 frames, utt_pad_proportion=0.06709, over 10988.94 utterances.], batch size: 49, lr: 1.74e-02, grad_scale: 8.0 2023-03-07 20:34:05,044 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23278.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:34:24,947 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23290.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:34:32,773 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23295.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:34:59,021 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.052e+02 3.226e+02 3.955e+02 4.943e+02 9.794e+02, threshold=7.910e+02, percent-clipped=2.0 2023-03-07 20:35:11,504 INFO [train2.py:809] (0/4) Epoch 6, batch 3400, loss[ctc_loss=0.1439, att_loss=0.2742, loss=0.2482, over 17293.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02409, over 59.00 utterances.], tot_loss[ctc_loss=0.1464, att_loss=0.2734, loss=0.248, over 3279642.90 frames. utt_duration=1196 frames, utt_pad_proportion=0.06567, over 10979.76 utterances.], batch size: 59, lr: 1.73e-02, grad_scale: 8.0 2023-03-07 20:35:49,181 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23343.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:35:49,283 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7749, 4.1766, 4.4629, 4.4242, 4.2487, 4.6926, 4.3422, 4.7957], device='cuda:0'), covar=tensor([0.0699, 0.0643, 0.0574, 0.0761, 0.1710, 0.0719, 0.1336, 0.0564], device='cuda:0'), in_proj_covar=tensor([0.0560, 0.0348, 0.0375, 0.0447, 0.0596, 0.0372, 0.0309, 0.0379], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 20:36:31,950 INFO [train2.py:809] (0/4) Epoch 6, batch 3450, loss[ctc_loss=0.1262, att_loss=0.2495, loss=0.2248, over 16180.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006883, over 41.00 utterances.], tot_loss[ctc_loss=0.1467, att_loss=0.2734, loss=0.2481, over 3270408.19 frames. utt_duration=1178 frames, utt_pad_proportion=0.07182, over 11115.15 utterances.], batch size: 41, lr: 1.73e-02, grad_scale: 8.0 2023-03-07 20:37:23,746 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8214, 4.6187, 4.6540, 4.6406, 5.0656, 4.7004, 4.5146, 2.0379], device='cuda:0'), covar=tensor([0.0271, 0.0339, 0.0212, 0.0170, 0.1202, 0.0257, 0.0340, 0.2847], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0125, 0.0120, 0.0121, 0.0296, 0.0129, 0.0116, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 20:37:39,556 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.288e+02 3.391e+02 4.040e+02 5.091e+02 1.147e+03, threshold=8.079e+02, percent-clipped=5.0 2023-03-07 20:37:41,370 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9639, 5.2417, 5.5169, 5.4689, 5.2612, 5.8862, 5.0186, 5.9964], device='cuda:0'), covar=tensor([0.0571, 0.0540, 0.0543, 0.0816, 0.1799, 0.0716, 0.0611, 0.0463], device='cuda:0'), in_proj_covar=tensor([0.0569, 0.0348, 0.0381, 0.0448, 0.0606, 0.0379, 0.0319, 0.0383], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 20:37:52,185 INFO [train2.py:809] (0/4) Epoch 6, batch 3500, loss[ctc_loss=0.1387, att_loss=0.2772, loss=0.2495, over 16409.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006427, over 44.00 utterances.], tot_loss[ctc_loss=0.1454, att_loss=0.2729, loss=0.2474, over 3278398.69 frames. utt_duration=1205 frames, utt_pad_proportion=0.06268, over 10896.99 utterances.], batch size: 44, lr: 1.73e-02, grad_scale: 8.0 2023-03-07 20:38:40,690 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23450.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:39:02,967 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23464.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:39:12,997 INFO [train2.py:809] (0/4) Epoch 6, batch 3550, loss[ctc_loss=0.1192, att_loss=0.2574, loss=0.2297, over 16032.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006517, over 40.00 utterances.], tot_loss[ctc_loss=0.1456, att_loss=0.2725, loss=0.2471, over 3263932.39 frames. utt_duration=1188 frames, utt_pad_proportion=0.07137, over 11002.78 utterances.], batch size: 40, lr: 1.73e-02, grad_scale: 8.0 2023-03-07 20:39:30,644 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23481.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:39:42,264 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=23488.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:39:57,382 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23498.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:40:19,336 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.856e+02 3.330e+02 3.992e+02 4.756e+02 1.054e+03, threshold=7.983e+02, percent-clipped=2.0 2023-03-07 20:40:31,489 INFO [train2.py:809] (0/4) Epoch 6, batch 3600, loss[ctc_loss=0.1113, att_loss=0.2401, loss=0.2143, over 15770.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.007981, over 38.00 utterances.], tot_loss[ctc_loss=0.146, att_loss=0.2723, loss=0.247, over 3255749.33 frames. utt_duration=1195 frames, utt_pad_proportion=0.07089, over 10911.51 utterances.], batch size: 38, lr: 1.73e-02, grad_scale: 8.0 2023-03-07 20:40:46,276 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23529.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:41:18,562 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=23549.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 20:41:51,579 INFO [train2.py:809] (0/4) Epoch 6, batch 3650, loss[ctc_loss=0.1499, att_loss=0.2917, loss=0.2634, over 17313.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02375, over 59.00 utterances.], tot_loss[ctc_loss=0.1443, att_loss=0.2712, loss=0.2458, over 3247735.65 frames. utt_duration=1211 frames, utt_pad_proportion=0.06803, over 10736.58 utterances.], batch size: 59, lr: 1.72e-02, grad_scale: 8.0 2023-03-07 20:41:57,498 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23573.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:42:16,984 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23585.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:42:59,613 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.210e+02 3.073e+02 3.816e+02 5.187e+02 1.095e+03, threshold=7.631e+02, percent-clipped=5.0 2023-03-07 20:43:12,936 INFO [train2.py:809] (0/4) Epoch 6, batch 3700, loss[ctc_loss=0.118, att_loss=0.2721, loss=0.2413, over 17106.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01592, over 56.00 utterances.], tot_loss[ctc_loss=0.1441, att_loss=0.2708, loss=0.2455, over 3243794.45 frames. utt_duration=1216 frames, utt_pad_proportion=0.06861, over 10684.32 utterances.], batch size: 56, lr: 1.72e-02, grad_scale: 8.0 2023-03-07 20:43:30,383 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3622, 1.6890, 1.9601, 1.1267, 2.4355, 1.4275, 1.7858, 2.5868], device='cuda:0'), covar=tensor([0.0353, 0.2489, 0.2012, 0.1917, 0.0806, 0.1817, 0.2058, 0.0916], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0089, 0.0090, 0.0080, 0.0079, 0.0077, 0.0089, 0.0070], device='cuda:0'), out_proj_covar=tensor([3.8274e-05, 5.1509e-05, 5.0967e-05, 4.3239e-05, 3.8817e-05, 4.4984e-05, 5.0164e-05, 4.1995e-05], device='cuda:0') 2023-03-07 20:43:55,820 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7111, 4.6464, 4.5032, 4.6796, 4.9235, 4.8479, 4.4637, 2.3174], device='cuda:0'), covar=tensor([0.0249, 0.0254, 0.0247, 0.0125, 0.1145, 0.0198, 0.0294, 0.2587], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0124, 0.0121, 0.0124, 0.0293, 0.0127, 0.0115, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 20:44:32,073 INFO [train2.py:809] (0/4) Epoch 6, batch 3750, loss[ctc_loss=0.1212, att_loss=0.2383, loss=0.2149, over 15645.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008782, over 37.00 utterances.], tot_loss[ctc_loss=0.1462, att_loss=0.2727, loss=0.2474, over 3255518.75 frames. utt_duration=1209 frames, utt_pad_proportion=0.06638, over 10782.13 utterances.], batch size: 37, lr: 1.72e-02, grad_scale: 8.0 2023-03-07 20:45:38,747 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.845e+02 3.181e+02 3.944e+02 5.174e+02 1.956e+03, threshold=7.887e+02, percent-clipped=9.0 2023-03-07 20:45:52,112 INFO [train2.py:809] (0/4) Epoch 6, batch 3800, loss[ctc_loss=0.1505, att_loss=0.2704, loss=0.2464, over 16114.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006787, over 42.00 utterances.], tot_loss[ctc_loss=0.1458, att_loss=0.2722, loss=0.2469, over 3254312.37 frames. utt_duration=1218 frames, utt_pad_proportion=0.06467, over 10700.34 utterances.], batch size: 42, lr: 1.72e-02, grad_scale: 8.0 2023-03-07 20:47:02,182 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23764.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:47:11,174 INFO [train2.py:809] (0/4) Epoch 6, batch 3850, loss[ctc_loss=0.1243, att_loss=0.2452, loss=0.221, over 14572.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.0277, over 32.00 utterances.], tot_loss[ctc_loss=0.1454, att_loss=0.2725, loss=0.2471, over 3262662.93 frames. utt_duration=1219 frames, utt_pad_proportion=0.06177, over 10719.06 utterances.], batch size: 32, lr: 1.72e-02, grad_scale: 8.0 2023-03-07 20:48:18,382 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.854e+02 3.226e+02 4.048e+02 5.028e+02 1.314e+03, threshold=8.096e+02, percent-clipped=6.0 2023-03-07 20:48:18,489 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23812.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:48:30,432 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-03-07 20:48:31,079 INFO [train2.py:809] (0/4) Epoch 6, batch 3900, loss[ctc_loss=0.1267, att_loss=0.2555, loss=0.2298, over 15655.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.008213, over 37.00 utterances.], tot_loss[ctc_loss=0.1437, att_loss=0.2715, loss=0.2459, over 3264243.36 frames. utt_duration=1221 frames, utt_pad_proportion=0.06033, over 10703.63 utterances.], batch size: 37, lr: 1.72e-02, grad_scale: 8.0 2023-03-07 20:49:08,494 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=23844.0, num_to_drop=1, layers_to_drop={3} 2023-03-07 20:49:48,972 INFO [train2.py:809] (0/4) Epoch 6, batch 3950, loss[ctc_loss=0.1338, att_loss=0.2745, loss=0.2463, over 17074.00 frames. utt_duration=691.2 frames, utt_pad_proportion=0.1294, over 99.00 utterances.], tot_loss[ctc_loss=0.1441, att_loss=0.2721, loss=0.2465, over 3273157.82 frames. utt_duration=1220 frames, utt_pad_proportion=0.05756, over 10743.29 utterances.], batch size: 99, lr: 1.71e-02, grad_scale: 8.0 2023-03-07 20:49:54,008 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23873.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:50:03,010 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2673, 4.8441, 4.6037, 4.8494, 4.9757, 4.4955, 3.6755, 4.6318], device='cuda:0'), covar=tensor([0.0125, 0.0120, 0.0106, 0.0077, 0.0086, 0.0104, 0.0535, 0.0191], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0057, 0.0065, 0.0044, 0.0045, 0.0055, 0.0078, 0.0075], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-07 20:50:12,159 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=23885.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:50:38,459 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-6.pt 2023-03-07 20:51:06,488 INFO [train2.py:809] (0/4) Epoch 7, batch 0, loss[ctc_loss=0.123, att_loss=0.2516, loss=0.2259, over 15761.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009358, over 38.00 utterances.], tot_loss[ctc_loss=0.123, att_loss=0.2516, loss=0.2259, over 15761.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009358, over 38.00 utterances.], batch size: 38, lr: 1.61e-02, grad_scale: 8.0 2023-03-07 20:51:06,490 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 20:51:19,282 INFO [train2.py:843] (0/4) Epoch 7, validation: ctc_loss=0.06772, att_loss=0.2471, loss=0.2112, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 20:51:19,283 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16076MB 2023-03-07 20:51:34,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.137e+02 3.227e+02 4.296e+02 5.385e+02 1.552e+03, threshold=8.591e+02, percent-clipped=5.0 2023-03-07 20:51:48,669 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23921.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:52:07,301 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=23933.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:52:38,297 INFO [train2.py:809] (0/4) Epoch 7, batch 50, loss[ctc_loss=0.1341, att_loss=0.2606, loss=0.2353, over 16332.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005791, over 45.00 utterances.], tot_loss[ctc_loss=0.1401, att_loss=0.2697, loss=0.2438, over 737279.83 frames. utt_duration=1289 frames, utt_pad_proportion=0.04955, over 2291.38 utterances.], batch size: 45, lr: 1.60e-02, grad_scale: 8.0 2023-03-07 20:53:10,549 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-07 20:53:53,947 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-24000.pt 2023-03-07 20:54:02,450 INFO [train2.py:809] (0/4) Epoch 7, batch 100, loss[ctc_loss=0.1475, att_loss=0.2641, loss=0.2407, over 16172.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007282, over 41.00 utterances.], tot_loss[ctc_loss=0.1414, att_loss=0.2717, loss=0.2457, over 1311087.14 frames. utt_duration=1272 frames, utt_pad_proportion=0.04438, over 4129.24 utterances.], batch size: 41, lr: 1.60e-02, grad_scale: 8.0 2023-03-07 20:54:18,137 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.071e+02 3.045e+02 3.704e+02 4.782e+02 7.393e+02, threshold=7.408e+02, percent-clipped=0.0 2023-03-07 20:54:32,451 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.5898, 5.8315, 5.1570, 5.7454, 5.3394, 5.1824, 5.3120, 5.1088], device='cuda:0'), covar=tensor([0.1388, 0.1044, 0.0936, 0.0689, 0.0817, 0.1405, 0.2436, 0.2212], device='cuda:0'), in_proj_covar=tensor([0.0367, 0.0410, 0.0314, 0.0333, 0.0301, 0.0373, 0.0441, 0.0397], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 20:54:34,673 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.48 vs. limit=5.0 2023-03-07 20:54:57,369 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24037.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:55:23,558 INFO [train2.py:809] (0/4) Epoch 7, batch 150, loss[ctc_loss=0.1399, att_loss=0.2519, loss=0.2295, over 15865.00 frames. utt_duration=1628 frames, utt_pad_proportion=0.009483, over 39.00 utterances.], tot_loss[ctc_loss=0.1412, att_loss=0.2709, loss=0.245, over 1744910.54 frames. utt_duration=1263 frames, utt_pad_proportion=0.05035, over 5531.65 utterances.], batch size: 39, lr: 1.60e-02, grad_scale: 8.0 2023-03-07 20:55:24,536 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6909, 2.8539, 3.6095, 3.0317, 3.5949, 4.7687, 4.4492, 3.3694], device='cuda:0'), covar=tensor([0.0338, 0.1577, 0.1062, 0.1224, 0.0911, 0.0535, 0.0423, 0.1118], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0211, 0.0211, 0.0193, 0.0216, 0.0232, 0.0179, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 20:56:11,596 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.36 vs. limit=5.0 2023-03-07 20:56:34,148 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24098.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:56:41,473 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3466, 4.6034, 4.5860, 4.6299, 2.2535, 4.5549, 2.9675, 1.9259], device='cuda:0'), covar=tensor([0.0229, 0.0133, 0.0643, 0.0217, 0.2376, 0.0178, 0.1439, 0.1868], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0101, 0.0251, 0.0108, 0.0223, 0.0099, 0.0222, 0.0207], device='cuda:0'), out_proj_covar=tensor([1.1505e-04, 1.0076e-04, 2.2485e-04, 9.8338e-05, 2.0604e-04, 9.6720e-05, 1.9853e-04, 1.8573e-04], device='cuda:0') 2023-03-07 20:56:42,638 INFO [train2.py:809] (0/4) Epoch 7, batch 200, loss[ctc_loss=0.1087, att_loss=0.2406, loss=0.2142, over 15944.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007617, over 41.00 utterances.], tot_loss[ctc_loss=0.1407, att_loss=0.2706, loss=0.2446, over 2087703.92 frames. utt_duration=1262 frames, utt_pad_proportion=0.04761, over 6622.81 utterances.], batch size: 41, lr: 1.60e-02, grad_scale: 8.0 2023-03-07 20:56:56,754 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.983e+02 2.953e+02 3.642e+02 4.338e+02 9.354e+02, threshold=7.285e+02, percent-clipped=1.0 2023-03-07 20:57:23,048 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-03-07 20:57:47,312 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24144.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 20:58:02,068 INFO [train2.py:809] (0/4) Epoch 7, batch 250, loss[ctc_loss=0.1249, att_loss=0.2655, loss=0.2374, over 16771.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006436, over 48.00 utterances.], tot_loss[ctc_loss=0.1406, att_loss=0.2704, loss=0.2444, over 2354779.73 frames. utt_duration=1268 frames, utt_pad_proportion=0.04541, over 7438.54 utterances.], batch size: 48, lr: 1.60e-02, grad_scale: 8.0 2023-03-07 20:58:04,556 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5849, 2.3682, 3.3579, 4.5821, 4.1092, 4.0768, 2.9498, 2.0109], device='cuda:0'), covar=tensor([0.0578, 0.2603, 0.1086, 0.0395, 0.0473, 0.0306, 0.1674, 0.2687], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0195, 0.0191, 0.0165, 0.0144, 0.0128, 0.0188, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-07 20:58:14,464 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-07 20:59:02,755 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24192.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 20:59:22,692 INFO [train2.py:809] (0/4) Epoch 7, batch 300, loss[ctc_loss=0.1035, att_loss=0.2182, loss=0.1952, over 15784.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.007725, over 38.00 utterances.], tot_loss[ctc_loss=0.1378, att_loss=0.2677, loss=0.2417, over 2554503.06 frames. utt_duration=1286 frames, utt_pad_proportion=0.04353, over 7952.73 utterances.], batch size: 38, lr: 1.60e-02, grad_scale: 8.0 2023-03-07 20:59:25,893 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5463, 4.8096, 4.2317, 4.7518, 4.3312, 4.2377, 4.3158, 4.0385], device='cuda:0'), covar=tensor([0.1402, 0.1057, 0.0976, 0.0854, 0.1082, 0.1544, 0.2344, 0.2624], device='cuda:0'), in_proj_covar=tensor([0.0374, 0.0418, 0.0317, 0.0339, 0.0308, 0.0374, 0.0445, 0.0406], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 20:59:25,993 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8232, 3.9604, 4.0516, 3.9560, 4.0785, 4.1058, 3.9181, 3.7720], device='cuda:0'), covar=tensor([0.0950, 0.0557, 0.0220, 0.0419, 0.0318, 0.0335, 0.0281, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0398, 0.0236, 0.0180, 0.0219, 0.0280, 0.0308, 0.0229, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 20:59:36,416 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.964e+02 2.956e+02 3.932e+02 4.844e+02 1.089e+03, threshold=7.865e+02, percent-clipped=8.0 2023-03-07 20:59:43,041 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6684, 4.9796, 4.9404, 4.8748, 4.9820, 4.9975, 4.7114, 4.4264], device='cuda:0'), covar=tensor([0.0932, 0.0416, 0.0216, 0.0401, 0.0292, 0.0297, 0.0277, 0.0367], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0237, 0.0181, 0.0220, 0.0281, 0.0310, 0.0230, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 21:00:42,738 INFO [train2.py:809] (0/4) Epoch 7, batch 350, loss[ctc_loss=0.1159, att_loss=0.2435, loss=0.218, over 15364.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01147, over 35.00 utterances.], tot_loss[ctc_loss=0.1382, att_loss=0.2684, loss=0.2424, over 2720174.95 frames. utt_duration=1264 frames, utt_pad_proportion=0.04722, over 8617.71 utterances.], batch size: 35, lr: 1.59e-02, grad_scale: 8.0 2023-03-07 21:01:46,800 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-07 21:02:03,066 INFO [train2.py:809] (0/4) Epoch 7, batch 400, loss[ctc_loss=0.1194, att_loss=0.261, loss=0.2327, over 16122.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006618, over 42.00 utterances.], tot_loss[ctc_loss=0.1382, att_loss=0.2682, loss=0.2422, over 2827795.88 frames. utt_duration=1245 frames, utt_pad_proportion=0.05718, over 9093.87 utterances.], batch size: 42, lr: 1.59e-02, grad_scale: 8.0 2023-03-07 21:02:03,391 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0405, 3.8299, 3.2489, 3.5405, 3.9468, 3.6713, 2.7747, 4.5625], device='cuda:0'), covar=tensor([0.0904, 0.0414, 0.0976, 0.0582, 0.0544, 0.0610, 0.0905, 0.0346], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0152, 0.0191, 0.0157, 0.0183, 0.0186, 0.0162, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 21:02:16,567 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.699e+02 2.848e+02 3.496e+02 4.281e+02 6.983e+02, threshold=6.993e+02, percent-clipped=0.0 2023-03-07 21:03:22,771 INFO [train2.py:809] (0/4) Epoch 7, batch 450, loss[ctc_loss=0.1216, att_loss=0.2412, loss=0.2173, over 14526.00 frames. utt_duration=1817 frames, utt_pad_proportion=0.02916, over 32.00 utterances.], tot_loss[ctc_loss=0.1386, att_loss=0.2688, loss=0.2427, over 2933728.75 frames. utt_duration=1228 frames, utt_pad_proportion=0.05823, over 9571.64 utterances.], batch size: 32, lr: 1.59e-02, grad_scale: 8.0 2023-03-07 21:04:23,005 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6061, 5.8463, 5.1590, 5.6612, 5.3506, 5.0288, 5.2483, 5.1158], device='cuda:0'), covar=tensor([0.1280, 0.0904, 0.0820, 0.0810, 0.0861, 0.1514, 0.2297, 0.2458], device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0426, 0.0326, 0.0351, 0.0316, 0.0389, 0.0452, 0.0416], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 21:04:25,993 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24393.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:04:42,800 INFO [train2.py:809] (0/4) Epoch 7, batch 500, loss[ctc_loss=0.1454, att_loss=0.2864, loss=0.2582, over 17043.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01021, over 53.00 utterances.], tot_loss[ctc_loss=0.1379, att_loss=0.2683, loss=0.2422, over 3008308.72 frames. utt_duration=1246 frames, utt_pad_proportion=0.05376, over 9666.14 utterances.], batch size: 53, lr: 1.59e-02, grad_scale: 8.0 2023-03-07 21:04:56,598 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 3.293e+02 3.980e+02 5.188e+02 9.157e+02, threshold=7.960e+02, percent-clipped=6.0 2023-03-07 21:05:29,720 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-07 21:06:02,363 INFO [train2.py:809] (0/4) Epoch 7, batch 550, loss[ctc_loss=0.1818, att_loss=0.2878, loss=0.2666, over 16941.00 frames. utt_duration=686.2 frames, utt_pad_proportion=0.138, over 99.00 utterances.], tot_loss[ctc_loss=0.1383, att_loss=0.2681, loss=0.2421, over 3065269.97 frames. utt_duration=1256 frames, utt_pad_proportion=0.0515, over 9773.67 utterances.], batch size: 99, lr: 1.59e-02, grad_scale: 8.0 2023-03-07 21:06:40,060 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3771, 5.2072, 5.1520, 3.2161, 5.0996, 4.3471, 4.7166, 3.0793], device='cuda:0'), covar=tensor([0.0103, 0.0062, 0.0181, 0.0847, 0.0072, 0.0171, 0.0209, 0.1103], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0067, 0.0058, 0.0099, 0.0062, 0.0078, 0.0082, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 21:07:23,380 INFO [train2.py:809] (0/4) Epoch 7, batch 600, loss[ctc_loss=0.1706, att_loss=0.3022, loss=0.2759, over 17443.00 frames. utt_duration=1013 frames, utt_pad_proportion=0.04463, over 69.00 utterances.], tot_loss[ctc_loss=0.1376, att_loss=0.2683, loss=0.2422, over 3114863.22 frames. utt_duration=1255 frames, utt_pad_proportion=0.05143, over 9941.50 utterances.], batch size: 69, lr: 1.59e-02, grad_scale: 8.0 2023-03-07 21:07:23,763 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1255, 1.5637, 1.7464, 1.5944, 3.0529, 1.8414, 1.8421, 1.9218], device='cuda:0'), covar=tensor([0.0640, 0.3311, 0.3201, 0.1628, 0.0824, 0.2089, 0.2298, 0.2085], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0086, 0.0089, 0.0079, 0.0078, 0.0075, 0.0087, 0.0071], device='cuda:0'), out_proj_covar=tensor([3.8004e-05, 5.0743e-05, 5.0961e-05, 4.3007e-05, 3.8700e-05, 4.3976e-05, 5.0137e-05, 4.2994e-05], device='cuda:0') 2023-03-07 21:07:37,150 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.166e+02 3.070e+02 3.787e+02 4.448e+02 1.341e+03, threshold=7.574e+02, percent-clipped=2.0 2023-03-07 21:08:28,237 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9786, 5.0381, 4.9954, 2.3201, 4.7267, 4.3884, 4.1646, 2.0070], device='cuda:0'), covar=tensor([0.0167, 0.0078, 0.0184, 0.1527, 0.0100, 0.0186, 0.0399, 0.2375], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0068, 0.0057, 0.0099, 0.0062, 0.0078, 0.0082, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 21:08:33,802 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24547.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:08:43,731 INFO [train2.py:809] (0/4) Epoch 7, batch 650, loss[ctc_loss=0.1378, att_loss=0.2741, loss=0.2469, over 16538.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006542, over 45.00 utterances.], tot_loss[ctc_loss=0.1384, att_loss=0.2692, loss=0.243, over 3154111.15 frames. utt_duration=1222 frames, utt_pad_proportion=0.05884, over 10339.39 utterances.], batch size: 45, lr: 1.59e-02, grad_scale: 8.0 2023-03-07 21:09:08,907 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24569.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:10:04,225 INFO [train2.py:809] (0/4) Epoch 7, batch 700, loss[ctc_loss=0.1318, att_loss=0.2749, loss=0.2463, over 16887.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006538, over 49.00 utterances.], tot_loss[ctc_loss=0.1362, att_loss=0.2675, loss=0.2413, over 3176149.31 frames. utt_duration=1247 frames, utt_pad_proportion=0.05432, over 10200.02 utterances.], batch size: 49, lr: 1.58e-02, grad_scale: 8.0 2023-03-07 21:10:12,517 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24608.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:10:18,242 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.862e+02 2.863e+02 3.503e+02 4.304e+02 9.533e+02, threshold=7.005e+02, percent-clipped=2.0 2023-03-07 21:10:46,378 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24630.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:11:23,258 INFO [train2.py:809] (0/4) Epoch 7, batch 750, loss[ctc_loss=0.1556, att_loss=0.2557, loss=0.2356, over 15483.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.00976, over 36.00 utterances.], tot_loss[ctc_loss=0.1378, att_loss=0.2681, loss=0.242, over 3200773.50 frames. utt_duration=1240 frames, utt_pad_proportion=0.05427, over 10333.93 utterances.], batch size: 36, lr: 1.58e-02, grad_scale: 16.0 2023-03-07 21:12:03,731 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3997, 3.4358, 3.3776, 2.6852, 3.3022, 3.3679, 3.4852, 2.0646], device='cuda:0'), covar=tensor([0.1143, 0.1249, 0.3347, 0.6513, 0.1574, 0.3842, 0.0816, 0.8540], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0078, 0.0080, 0.0124, 0.0072, 0.0111, 0.0070, 0.0123], device='cuda:0'), out_proj_covar=tensor([6.1333e-05, 5.8500e-05, 6.5159e-05, 9.4302e-05, 5.7865e-05, 8.7268e-05, 5.3356e-05, 9.7507e-05], device='cuda:0') 2023-03-07 21:12:26,734 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=24693.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:12:43,278 INFO [train2.py:809] (0/4) Epoch 7, batch 800, loss[ctc_loss=0.1466, att_loss=0.2881, loss=0.2598, over 17397.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03318, over 63.00 utterances.], tot_loss[ctc_loss=0.1374, att_loss=0.2684, loss=0.2422, over 3225168.37 frames. utt_duration=1253 frames, utt_pad_proportion=0.04862, over 10311.95 utterances.], batch size: 63, lr: 1.58e-02, grad_scale: 16.0 2023-03-07 21:12:57,268 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.749e+02 2.992e+02 3.996e+02 4.981e+02 1.398e+03, threshold=7.991e+02, percent-clipped=9.0 2023-03-07 21:13:43,788 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=24741.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:13:56,059 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-03-07 21:14:03,646 INFO [train2.py:809] (0/4) Epoch 7, batch 850, loss[ctc_loss=0.1153, att_loss=0.2416, loss=0.2164, over 15627.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009985, over 37.00 utterances.], tot_loss[ctc_loss=0.1366, att_loss=0.2681, loss=0.2418, over 3242134.55 frames. utt_duration=1250 frames, utt_pad_proportion=0.04788, over 10391.35 utterances.], batch size: 37, lr: 1.58e-02, grad_scale: 16.0 2023-03-07 21:14:22,636 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24765.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:14:52,577 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-07 21:15:24,350 INFO [train2.py:809] (0/4) Epoch 7, batch 900, loss[ctc_loss=0.1531, att_loss=0.2801, loss=0.2547, over 16463.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006764, over 46.00 utterances.], tot_loss[ctc_loss=0.136, att_loss=0.2672, loss=0.2409, over 3249235.44 frames. utt_duration=1247 frames, utt_pad_proportion=0.05012, over 10434.23 utterances.], batch size: 46, lr: 1.58e-02, grad_scale: 16.0 2023-03-07 21:15:38,323 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.099e+02 2.848e+02 3.280e+02 4.214e+02 1.122e+03, threshold=6.561e+02, percent-clipped=2.0 2023-03-07 21:16:00,310 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24826.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:16:36,232 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24848.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:16:44,174 INFO [train2.py:809] (0/4) Epoch 7, batch 950, loss[ctc_loss=0.1502, att_loss=0.2808, loss=0.2547, over 16466.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006809, over 46.00 utterances.], tot_loss[ctc_loss=0.1354, att_loss=0.2665, loss=0.2403, over 3250429.36 frames. utt_duration=1248 frames, utt_pad_proportion=0.05382, over 10431.89 utterances.], batch size: 46, lr: 1.58e-02, grad_scale: 16.0 2023-03-07 21:16:53,648 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6807, 4.5665, 4.6147, 4.7534, 5.0936, 4.8376, 4.4466, 2.1345], device='cuda:0'), covar=tensor([0.0314, 0.0345, 0.0236, 0.0171, 0.1097, 0.0269, 0.0335, 0.2860], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0123, 0.0123, 0.0122, 0.0304, 0.0128, 0.0117, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 21:17:27,603 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2585, 3.3339, 2.6857, 3.0068, 3.5246, 3.2047, 2.1471, 3.6434], device='cuda:0'), covar=tensor([0.1227, 0.0423, 0.1027, 0.0689, 0.0623, 0.0581, 0.1157, 0.0586], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0152, 0.0187, 0.0156, 0.0184, 0.0185, 0.0162, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 21:18:03,787 INFO [train2.py:809] (0/4) Epoch 7, batch 1000, loss[ctc_loss=0.2209, att_loss=0.3202, loss=0.3004, over 14239.00 frames. utt_duration=391.6 frames, utt_pad_proportion=0.3178, over 146.00 utterances.], tot_loss[ctc_loss=0.1362, att_loss=0.2668, loss=0.2407, over 3249513.08 frames. utt_duration=1244 frames, utt_pad_proportion=0.05512, over 10464.68 utterances.], batch size: 146, lr: 1.57e-02, grad_scale: 16.0 2023-03-07 21:18:03,980 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24903.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:18:13,407 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=24909.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:18:17,630 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.779e+02 2.930e+02 3.688e+02 4.379e+02 1.333e+03, threshold=7.376e+02, percent-clipped=8.0 2023-03-07 21:18:37,298 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-03-07 21:18:37,626 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=24925.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:19:22,726 INFO [train2.py:809] (0/4) Epoch 7, batch 1050, loss[ctc_loss=0.1178, att_loss=0.2439, loss=0.2187, over 15751.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009175, over 38.00 utterances.], tot_loss[ctc_loss=0.136, att_loss=0.2669, loss=0.2407, over 3244736.31 frames. utt_duration=1235 frames, utt_pad_proportion=0.05947, over 10524.28 utterances.], batch size: 38, lr: 1.57e-02, grad_scale: 16.0 2023-03-07 21:20:29,719 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=24995.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:20:43,052 INFO [train2.py:809] (0/4) Epoch 7, batch 1100, loss[ctc_loss=0.1841, att_loss=0.2678, loss=0.251, over 15379.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01011, over 35.00 utterances.], tot_loss[ctc_loss=0.1347, att_loss=0.2656, loss=0.2394, over 3245373.56 frames. utt_duration=1263 frames, utt_pad_proportion=0.05439, over 10291.79 utterances.], batch size: 35, lr: 1.57e-02, grad_scale: 16.0 2023-03-07 21:20:57,054 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.688e+02 2.910e+02 3.574e+02 4.609e+02 1.124e+03, threshold=7.148e+02, percent-clipped=6.0 2023-03-07 21:22:03,893 INFO [train2.py:809] (0/4) Epoch 7, batch 1150, loss[ctc_loss=0.1473, att_loss=0.2951, loss=0.2655, over 17319.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.01078, over 55.00 utterances.], tot_loss[ctc_loss=0.1343, att_loss=0.2656, loss=0.2394, over 3257935.15 frames. utt_duration=1282 frames, utt_pad_proportion=0.04757, over 10178.50 utterances.], batch size: 55, lr: 1.57e-02, grad_scale: 16.0 2023-03-07 21:22:08,920 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25056.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 21:23:24,554 INFO [train2.py:809] (0/4) Epoch 7, batch 1200, loss[ctc_loss=0.1497, att_loss=0.2771, loss=0.2516, over 16891.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.007253, over 49.00 utterances.], tot_loss[ctc_loss=0.1355, att_loss=0.2663, loss=0.2401, over 3252842.84 frames. utt_duration=1251 frames, utt_pad_proportion=0.05601, over 10411.67 utterances.], batch size: 49, lr: 1.57e-02, grad_scale: 16.0 2023-03-07 21:23:38,666 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.736e+02 2.863e+02 3.570e+02 4.681e+02 1.368e+03, threshold=7.141e+02, percent-clipped=7.0 2023-03-07 21:23:52,657 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25121.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:23:59,275 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8752, 1.7806, 2.2623, 2.0437, 3.0191, 2.0698, 1.7871, 2.4864], device='cuda:0'), covar=tensor([0.0617, 0.3560, 0.3414, 0.1314, 0.0779, 0.1570, 0.3135, 0.0979], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0090, 0.0090, 0.0079, 0.0078, 0.0077, 0.0090, 0.0071], device='cuda:0'), out_proj_covar=tensor([3.8813e-05, 5.2996e-05, 5.2231e-05, 4.3928e-05, 3.9367e-05, 4.5520e-05, 5.1662e-05, 4.3143e-05], device='cuda:0') 2023-03-07 21:24:44,723 INFO [train2.py:809] (0/4) Epoch 7, batch 1250, loss[ctc_loss=0.09775, att_loss=0.2258, loss=0.2002, over 10833.00 frames. utt_duration=1807 frames, utt_pad_proportion=0.02278, over 24.00 utterances.], tot_loss[ctc_loss=0.1347, att_loss=0.2662, loss=0.2399, over 3253301.59 frames. utt_duration=1253 frames, utt_pad_proportion=0.05575, over 10397.82 utterances.], batch size: 24, lr: 1.57e-02, grad_scale: 16.0 2023-03-07 21:26:05,283 INFO [train2.py:809] (0/4) Epoch 7, batch 1300, loss[ctc_loss=0.1465, att_loss=0.2695, loss=0.2449, over 16176.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.005756, over 41.00 utterances.], tot_loss[ctc_loss=0.1361, att_loss=0.2671, loss=0.2409, over 3260283.99 frames. utt_duration=1244 frames, utt_pad_proportion=0.05804, over 10498.45 utterances.], batch size: 41, lr: 1.57e-02, grad_scale: 16.0 2023-03-07 21:26:05,513 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25203.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:26:06,858 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25204.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:26:19,175 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.912e+02 2.854e+02 3.559e+02 4.625e+02 8.622e+02, threshold=7.119e+02, percent-clipped=5.0 2023-03-07 21:26:39,764 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25225.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:27:03,330 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2960, 4.7753, 4.6773, 4.9024, 2.5467, 4.7169, 2.9181, 2.5976], device='cuda:0'), covar=tensor([0.0198, 0.0128, 0.0689, 0.0150, 0.2250, 0.0182, 0.1628, 0.1336], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0102, 0.0252, 0.0108, 0.0223, 0.0104, 0.0226, 0.0203], device='cuda:0'), out_proj_covar=tensor([1.1588e-04, 1.0248e-04, 2.2657e-04, 9.8712e-05, 2.0689e-04, 1.0198e-04, 2.0213e-04, 1.8405e-04], device='cuda:0') 2023-03-07 21:27:22,555 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25251.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:27:25,584 INFO [train2.py:809] (0/4) Epoch 7, batch 1350, loss[ctc_loss=0.154, att_loss=0.2858, loss=0.2594, over 17430.00 frames. utt_duration=1108 frames, utt_pad_proportion=0.03037, over 63.00 utterances.], tot_loss[ctc_loss=0.1366, att_loss=0.267, loss=0.241, over 3257695.39 frames. utt_duration=1231 frames, utt_pad_proportion=0.06222, over 10597.71 utterances.], batch size: 63, lr: 1.56e-02, grad_scale: 16.0 2023-03-07 21:27:56,365 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25273.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:28:35,879 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7598, 5.9428, 5.3965, 5.8447, 5.5877, 5.3220, 5.4038, 5.2017], device='cuda:0'), covar=tensor([0.1223, 0.0786, 0.0755, 0.0658, 0.0602, 0.1261, 0.1911, 0.2197], device='cuda:0'), in_proj_covar=tensor([0.0372, 0.0419, 0.0323, 0.0338, 0.0304, 0.0382, 0.0443, 0.0417], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 21:28:36,138 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7271, 1.6464, 1.9142, 1.1193, 2.9757, 1.7615, 1.6353, 2.2814], device='cuda:0'), covar=tensor([0.0622, 0.3516, 0.2731, 0.2040, 0.0889, 0.1828, 0.3036, 0.1513], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0086, 0.0087, 0.0079, 0.0076, 0.0075, 0.0089, 0.0068], device='cuda:0'), out_proj_covar=tensor([3.7884e-05, 5.1448e-05, 5.0892e-05, 4.3642e-05, 3.8856e-05, 4.4358e-05, 5.1225e-05, 4.1888e-05], device='cuda:0') 2023-03-07 21:28:45,906 INFO [train2.py:809] (0/4) Epoch 7, batch 1400, loss[ctc_loss=0.1782, att_loss=0.2998, loss=0.2755, over 16690.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006453, over 46.00 utterances.], tot_loss[ctc_loss=0.1364, att_loss=0.2671, loss=0.241, over 3263169.65 frames. utt_duration=1230 frames, utt_pad_proportion=0.06082, over 10621.88 utterances.], batch size: 46, lr: 1.56e-02, grad_scale: 16.0 2023-03-07 21:29:01,605 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.920e+02 3.478e+02 4.487e+02 1.591e+03, threshold=6.957e+02, percent-clipped=5.0 2023-03-07 21:30:03,012 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25351.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 21:30:05,929 INFO [train2.py:809] (0/4) Epoch 7, batch 1450, loss[ctc_loss=0.1256, att_loss=0.2691, loss=0.2404, over 16887.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.007224, over 49.00 utterances.], tot_loss[ctc_loss=0.1374, att_loss=0.2675, loss=0.2415, over 3257956.39 frames. utt_duration=1213 frames, utt_pad_proportion=0.06848, over 10759.33 utterances.], batch size: 49, lr: 1.56e-02, grad_scale: 8.0 2023-03-07 21:31:25,455 INFO [train2.py:809] (0/4) Epoch 7, batch 1500, loss[ctc_loss=0.1137, att_loss=0.2479, loss=0.221, over 16405.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007435, over 44.00 utterances.], tot_loss[ctc_loss=0.136, att_loss=0.267, loss=0.2408, over 3264259.06 frames. utt_duration=1229 frames, utt_pad_proportion=0.06266, over 10635.08 utterances.], batch size: 44, lr: 1.56e-02, grad_scale: 8.0 2023-03-07 21:31:40,547 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.966e+02 3.623e+02 4.720e+02 8.485e+02, threshold=7.246e+02, percent-clipped=4.0 2023-03-07 21:31:52,953 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25421.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:32:03,880 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25428.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 21:32:45,013 INFO [train2.py:809] (0/4) Epoch 7, batch 1550, loss[ctc_loss=0.1575, att_loss=0.2735, loss=0.2503, over 16991.00 frames. utt_duration=688.2 frames, utt_pad_proportion=0.1311, over 99.00 utterances.], tot_loss[ctc_loss=0.1377, att_loss=0.2681, loss=0.242, over 3259524.71 frames. utt_duration=1179 frames, utt_pad_proportion=0.07639, over 11076.13 utterances.], batch size: 99, lr: 1.56e-02, grad_scale: 8.0 2023-03-07 21:32:58,068 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6869, 2.3448, 4.8616, 3.9203, 2.7811, 4.6113, 4.8673, 4.7029], device='cuda:0'), covar=tensor([0.0216, 0.1967, 0.0194, 0.1124, 0.2426, 0.0217, 0.0090, 0.0211], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0240, 0.0125, 0.0298, 0.0294, 0.0180, 0.0107, 0.0140], device='cuda:0'), out_proj_covar=tensor([1.3213e-04, 1.9711e-04, 1.1204e-04, 2.4417e-04, 2.5084e-04, 1.5806e-04, 9.6778e-05, 1.2743e-04], device='cuda:0') 2023-03-07 21:33:10,084 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25469.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:33:42,977 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25489.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 21:34:02,941 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9128, 4.8628, 4.8578, 2.7766, 4.7162, 4.1883, 3.9643, 2.1532], device='cuda:0'), covar=tensor([0.0137, 0.0090, 0.0180, 0.1147, 0.0098, 0.0208, 0.0396, 0.1760], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0070, 0.0058, 0.0099, 0.0064, 0.0079, 0.0084, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 21:34:04,340 INFO [train2.py:809] (0/4) Epoch 7, batch 1600, loss[ctc_loss=0.1642, att_loss=0.2922, loss=0.2666, over 17043.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01034, over 53.00 utterances.], tot_loss[ctc_loss=0.1373, att_loss=0.2682, loss=0.242, over 3267734.95 frames. utt_duration=1203 frames, utt_pad_proportion=0.06716, over 10879.94 utterances.], batch size: 53, lr: 1.56e-02, grad_scale: 8.0 2023-03-07 21:34:06,174 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25504.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:34:19,608 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.988e+02 2.911e+02 3.522e+02 4.081e+02 9.255e+02, threshold=7.044e+02, percent-clipped=3.0 2023-03-07 21:35:22,698 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25552.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:35:24,112 INFO [train2.py:809] (0/4) Epoch 7, batch 1650, loss[ctc_loss=0.1356, att_loss=0.285, loss=0.2551, over 17110.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01465, over 56.00 utterances.], tot_loss[ctc_loss=0.1355, att_loss=0.2673, loss=0.241, over 3267464.45 frames. utt_duration=1234 frames, utt_pad_proportion=0.0589, over 10607.94 utterances.], batch size: 56, lr: 1.56e-02, grad_scale: 8.0 2023-03-07 21:35:58,298 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.59 vs. limit=2.0 2023-03-07 21:36:37,933 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.0512, 2.8449, 2.9200, 2.2201, 2.7003, 2.6201, 2.6894, 1.6205], device='cuda:0'), covar=tensor([0.1635, 0.1437, 0.2425, 0.7106, 0.2255, 0.4330, 0.1446, 1.0470], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0080, 0.0082, 0.0131, 0.0076, 0.0121, 0.0073, 0.0130], device='cuda:0'), out_proj_covar=tensor([6.3212e-05, 6.1365e-05, 6.7945e-05, 9.9522e-05, 6.1739e-05, 9.4186e-05, 5.7116e-05, 1.0302e-04], device='cuda:0') 2023-03-07 21:36:44,062 INFO [train2.py:809] (0/4) Epoch 7, batch 1700, loss[ctc_loss=0.118, att_loss=0.2554, loss=0.2279, over 16545.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006047, over 45.00 utterances.], tot_loss[ctc_loss=0.1351, att_loss=0.2668, loss=0.2405, over 3271897.36 frames. utt_duration=1232 frames, utt_pad_proportion=0.05762, over 10637.16 utterances.], batch size: 45, lr: 1.55e-02, grad_scale: 8.0 2023-03-07 21:36:59,368 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.939e+02 2.859e+02 3.554e+02 4.579e+02 8.259e+02, threshold=7.107e+02, percent-clipped=6.0 2023-03-07 21:38:00,704 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=25651.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 21:38:03,528 INFO [train2.py:809] (0/4) Epoch 7, batch 1750, loss[ctc_loss=0.1112, att_loss=0.265, loss=0.2343, over 16691.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006379, over 46.00 utterances.], tot_loss[ctc_loss=0.1371, att_loss=0.2684, loss=0.2421, over 3282190.76 frames. utt_duration=1216 frames, utt_pad_proportion=0.05884, over 10808.99 utterances.], batch size: 46, lr: 1.55e-02, grad_scale: 8.0 2023-03-07 21:38:20,468 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-03-07 21:38:30,609 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7276, 3.8754, 3.9924, 3.9605, 4.0393, 4.0938, 3.8816, 3.7354], device='cuda:0'), covar=tensor([0.1119, 0.0783, 0.0298, 0.0481, 0.0315, 0.0357, 0.0320, 0.0379], device='cuda:0'), in_proj_covar=tensor([0.0404, 0.0240, 0.0187, 0.0226, 0.0279, 0.0308, 0.0232, 0.0268], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 21:38:57,784 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2580, 3.4799, 3.4726, 2.6516, 3.2527, 3.1361, 3.1876, 2.3177], device='cuda:0'), covar=tensor([0.1855, 0.1091, 0.2566, 0.6180, 0.1472, 0.4417, 0.0897, 0.7460], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0080, 0.0084, 0.0135, 0.0077, 0.0122, 0.0073, 0.0132], device='cuda:0'), out_proj_covar=tensor([6.4260e-05, 6.1402e-05, 6.9651e-05, 1.0218e-04, 6.2798e-05, 9.5053e-05, 5.7732e-05, 1.0443e-04], device='cuda:0') 2023-03-07 21:39:18,785 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=25699.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:39:25,096 INFO [train2.py:809] (0/4) Epoch 7, batch 1800, loss[ctc_loss=0.1432, att_loss=0.276, loss=0.2494, over 17352.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.0347, over 63.00 utterances.], tot_loss[ctc_loss=0.136, att_loss=0.2676, loss=0.2413, over 3281160.96 frames. utt_duration=1207 frames, utt_pad_proportion=0.06105, over 10891.51 utterances.], batch size: 63, lr: 1.55e-02, grad_scale: 8.0 2023-03-07 21:39:39,637 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0670, 4.8760, 4.9157, 2.7009, 4.7004, 4.4074, 4.1109, 2.3914], device='cuda:0'), covar=tensor([0.0133, 0.0112, 0.0193, 0.1172, 0.0115, 0.0198, 0.0379, 0.1507], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0071, 0.0058, 0.0099, 0.0063, 0.0079, 0.0084, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 21:39:40,904 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.727e+02 2.893e+02 3.531e+02 4.524e+02 7.224e+02, threshold=7.063e+02, percent-clipped=1.0 2023-03-07 21:40:27,351 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1770, 5.0431, 5.0596, 2.3505, 4.9273, 4.3617, 4.3993, 2.4530], device='cuda:0'), covar=tensor([0.0146, 0.0078, 0.0152, 0.1196, 0.0082, 0.0170, 0.0273, 0.1379], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0070, 0.0058, 0.0099, 0.0063, 0.0080, 0.0084, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 21:40:46,308 INFO [train2.py:809] (0/4) Epoch 7, batch 1850, loss[ctc_loss=0.1513, att_loss=0.2868, loss=0.2597, over 16464.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006885, over 46.00 utterances.], tot_loss[ctc_loss=0.1354, att_loss=0.2682, loss=0.2417, over 3294376.13 frames. utt_duration=1211 frames, utt_pad_proportion=0.05663, over 10894.42 utterances.], batch size: 46, lr: 1.55e-02, grad_scale: 8.0 2023-03-07 21:41:36,984 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=25784.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 21:42:06,979 INFO [train2.py:809] (0/4) Epoch 7, batch 1900, loss[ctc_loss=0.1489, att_loss=0.2606, loss=0.2383, over 16280.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007326, over 43.00 utterances.], tot_loss[ctc_loss=0.136, att_loss=0.2681, loss=0.2417, over 3284319.24 frames. utt_duration=1214 frames, utt_pad_proportion=0.05967, over 10838.80 utterances.], batch size: 43, lr: 1.55e-02, grad_scale: 8.0 2023-03-07 21:42:13,355 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8405, 5.0961, 5.3389, 5.3602, 5.1834, 5.7573, 4.9926, 5.8895], device='cuda:0'), covar=tensor([0.0600, 0.0589, 0.0588, 0.0789, 0.1663, 0.0811, 0.0583, 0.0535], device='cuda:0'), in_proj_covar=tensor([0.0581, 0.0359, 0.0399, 0.0461, 0.0631, 0.0399, 0.0330, 0.0405], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 21:42:22,402 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.824e+02 2.923e+02 3.414e+02 4.315e+02 6.616e+02, threshold=6.827e+02, percent-clipped=0.0 2023-03-07 21:42:41,763 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.77 vs. limit=5.0 2023-03-07 21:43:27,380 INFO [train2.py:809] (0/4) Epoch 7, batch 1950, loss[ctc_loss=0.1267, att_loss=0.2634, loss=0.236, over 16646.00 frames. utt_duration=1418 frames, utt_pad_proportion=0.004064, over 47.00 utterances.], tot_loss[ctc_loss=0.136, att_loss=0.268, loss=0.2416, over 3281866.69 frames. utt_duration=1198 frames, utt_pad_proportion=0.06562, over 10968.70 utterances.], batch size: 47, lr: 1.55e-02, grad_scale: 8.0 2023-03-07 21:43:35,767 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=25858.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 21:43:38,106 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-03-07 21:44:48,780 INFO [train2.py:809] (0/4) Epoch 7, batch 2000, loss[ctc_loss=0.2172, att_loss=0.3128, loss=0.2937, over 13987.00 frames. utt_duration=384.5 frames, utt_pad_proportion=0.3289, over 146.00 utterances.], tot_loss[ctc_loss=0.1353, att_loss=0.2674, loss=0.241, over 3274275.84 frames. utt_duration=1208 frames, utt_pad_proportion=0.06408, over 10851.97 utterances.], batch size: 146, lr: 1.55e-02, grad_scale: 8.0 2023-03-07 21:45:04,085 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.703e+02 2.979e+02 3.606e+02 4.355e+02 1.257e+03, threshold=7.211e+02, percent-clipped=2.0 2023-03-07 21:45:14,773 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=25919.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 21:46:09,515 INFO [train2.py:809] (0/4) Epoch 7, batch 2050, loss[ctc_loss=0.1454, att_loss=0.2771, loss=0.2507, over 17031.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007235, over 51.00 utterances.], tot_loss[ctc_loss=0.1346, att_loss=0.2666, loss=0.2402, over 3276373.76 frames. utt_duration=1230 frames, utt_pad_proportion=0.05847, over 10671.21 utterances.], batch size: 51, lr: 1.54e-02, grad_scale: 8.0 2023-03-07 21:46:48,003 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1694, 4.5684, 4.4089, 4.8143, 2.5032, 4.5323, 2.4761, 2.0883], device='cuda:0'), covar=tensor([0.0266, 0.0147, 0.0752, 0.0131, 0.2061, 0.0195, 0.1778, 0.1686], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0103, 0.0245, 0.0105, 0.0216, 0.0102, 0.0222, 0.0198], device='cuda:0'), out_proj_covar=tensor([1.1579e-04, 1.0335e-04, 2.2103e-04, 9.6934e-05, 2.0087e-04, 9.9347e-05, 1.9870e-04, 1.8037e-04], device='cuda:0') 2023-03-07 21:47:25,268 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-26000.pt 2023-03-07 21:47:33,893 INFO [train2.py:809] (0/4) Epoch 7, batch 2100, loss[ctc_loss=0.1559, att_loss=0.2821, loss=0.2569, over 17394.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03003, over 63.00 utterances.], tot_loss[ctc_loss=0.1346, att_loss=0.2674, loss=0.2408, over 3271876.19 frames. utt_duration=1205 frames, utt_pad_proportion=0.06548, over 10871.74 utterances.], batch size: 63, lr: 1.54e-02, grad_scale: 8.0 2023-03-07 21:47:49,610 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.867e+02 2.961e+02 3.539e+02 4.278e+02 1.039e+03, threshold=7.077e+02, percent-clipped=2.0 2023-03-07 21:48:29,320 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1381, 5.4322, 4.7342, 5.3232, 4.9467, 4.6256, 4.8474, 4.6760], device='cuda:0'), covar=tensor([0.1205, 0.0917, 0.0999, 0.0766, 0.0859, 0.1641, 0.2224, 0.2468], device='cuda:0'), in_proj_covar=tensor([0.0373, 0.0420, 0.0322, 0.0334, 0.0309, 0.0390, 0.0447, 0.0415], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 21:48:54,478 INFO [train2.py:809] (0/4) Epoch 7, batch 2150, loss[ctc_loss=0.1109, att_loss=0.2542, loss=0.2255, over 16330.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006004, over 45.00 utterances.], tot_loss[ctc_loss=0.1352, att_loss=0.2677, loss=0.2412, over 3275736.50 frames. utt_duration=1206 frames, utt_pad_proportion=0.06418, over 10874.28 utterances.], batch size: 45, lr: 1.54e-02, grad_scale: 8.0 2023-03-07 21:49:27,973 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.0372, 3.3011, 3.3788, 2.8194, 3.2130, 3.1667, 3.1578, 1.6205], device='cuda:0'), covar=tensor([0.1486, 0.1104, 0.2802, 0.6675, 0.2988, 0.4264, 0.1247, 1.0854], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0081, 0.0085, 0.0134, 0.0077, 0.0123, 0.0073, 0.0134], device='cuda:0'), out_proj_covar=tensor([6.3685e-05, 6.2756e-05, 7.0637e-05, 1.0193e-04, 6.3589e-05, 9.5757e-05, 5.8279e-05, 1.0548e-04], device='cuda:0') 2023-03-07 21:49:36,273 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8994, 4.9078, 4.6487, 4.6084, 5.1912, 5.0462, 4.7564, 2.3832], device='cuda:0'), covar=tensor([0.0277, 0.0244, 0.0233, 0.0289, 0.1056, 0.0169, 0.0255, 0.2427], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0124, 0.0126, 0.0122, 0.0304, 0.0126, 0.0115, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 21:49:44,678 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26084.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 21:50:05,376 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1290, 5.2488, 5.7152, 5.5642, 5.3877, 5.9999, 5.2522, 6.0687], device='cuda:0'), covar=tensor([0.0614, 0.0638, 0.0581, 0.0990, 0.2018, 0.0942, 0.0487, 0.0792], device='cuda:0'), in_proj_covar=tensor([0.0603, 0.0363, 0.0406, 0.0467, 0.0646, 0.0410, 0.0335, 0.0414], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 21:50:13,365 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26102.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:50:14,580 INFO [train2.py:809] (0/4) Epoch 7, batch 2200, loss[ctc_loss=0.2028, att_loss=0.3029, loss=0.2829, over 14197.00 frames. utt_duration=390.5 frames, utt_pad_proportion=0.3197, over 146.00 utterances.], tot_loss[ctc_loss=0.1354, att_loss=0.2677, loss=0.2413, over 3265399.84 frames. utt_duration=1171 frames, utt_pad_proportion=0.07515, over 11165.96 utterances.], batch size: 146, lr: 1.54e-02, grad_scale: 8.0 2023-03-07 21:50:30,122 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.999e+02 3.731e+02 5.039e+02 1.380e+03, threshold=7.462e+02, percent-clipped=10.0 2023-03-07 21:51:01,059 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26132.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 21:51:34,424 INFO [train2.py:809] (0/4) Epoch 7, batch 2250, loss[ctc_loss=0.1607, att_loss=0.2905, loss=0.2646, over 16882.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.005896, over 49.00 utterances.], tot_loss[ctc_loss=0.1355, att_loss=0.2674, loss=0.241, over 3254048.07 frames. utt_duration=1176 frames, utt_pad_proportion=0.07715, over 11080.55 utterances.], batch size: 49, lr: 1.54e-02, grad_scale: 8.0 2023-03-07 21:51:50,733 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26163.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:51:59,311 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26168.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:52:55,166 INFO [train2.py:809] (0/4) Epoch 7, batch 2300, loss[ctc_loss=0.08939, att_loss=0.2213, loss=0.1949, over 14547.00 frames. utt_duration=1820 frames, utt_pad_proportion=0.03915, over 32.00 utterances.], tot_loss[ctc_loss=0.1349, att_loss=0.2668, loss=0.2404, over 3251222.39 frames. utt_duration=1194 frames, utt_pad_proportion=0.07374, over 10904.24 utterances.], batch size: 32, lr: 1.54e-02, grad_scale: 8.0 2023-03-07 21:53:10,518 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.696e+02 2.838e+02 3.640e+02 4.798e+02 9.404e+02, threshold=7.281e+02, percent-clipped=6.0 2023-03-07 21:53:12,381 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26214.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 21:53:25,413 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2395, 4.5617, 4.0756, 4.5930, 4.1203, 4.2839, 4.6648, 4.5492], device='cuda:0'), covar=tensor([0.0567, 0.0251, 0.0719, 0.0222, 0.0471, 0.0384, 0.0216, 0.0163], device='cuda:0'), in_proj_covar=tensor([0.0267, 0.0208, 0.0261, 0.0186, 0.0219, 0.0166, 0.0190, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 21:53:37,050 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26229.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:53:43,256 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-07 21:54:11,171 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26250.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:54:14,308 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3028, 3.3911, 3.4070, 2.9236, 3.1901, 3.3366, 3.2010, 1.6710], device='cuda:0'), covar=tensor([0.2040, 0.1329, 0.3141, 0.5517, 0.2184, 0.7854, 0.1120, 1.1335], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0082, 0.0088, 0.0138, 0.0079, 0.0126, 0.0075, 0.0137], device='cuda:0'), out_proj_covar=tensor([6.5183e-05, 6.4328e-05, 7.3502e-05, 1.0476e-04, 6.5247e-05, 9.8242e-05, 6.0415e-05, 1.0831e-04], device='cuda:0') 2023-03-07 21:54:15,533 INFO [train2.py:809] (0/4) Epoch 7, batch 2350, loss[ctc_loss=0.1176, att_loss=0.2514, loss=0.2247, over 15958.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.00684, over 41.00 utterances.], tot_loss[ctc_loss=0.1357, att_loss=0.2671, loss=0.2408, over 3250468.61 frames. utt_duration=1190 frames, utt_pad_proportion=0.07379, over 10942.03 utterances.], batch size: 41, lr: 1.54e-02, grad_scale: 8.0 2023-03-07 21:54:43,233 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8893, 4.7919, 4.6108, 4.7650, 5.0782, 5.0211, 4.7132, 2.2990], device='cuda:0'), covar=tensor([0.0214, 0.0211, 0.0248, 0.0175, 0.1262, 0.0152, 0.0221, 0.2616], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0124, 0.0127, 0.0123, 0.0309, 0.0127, 0.0115, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 21:55:19,484 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8199, 5.2280, 5.0629, 5.0574, 5.2724, 5.2617, 5.0306, 4.6881], device='cuda:0'), covar=tensor([0.0938, 0.0364, 0.0260, 0.0427, 0.0235, 0.0265, 0.0193, 0.0293], device='cuda:0'), in_proj_covar=tensor([0.0406, 0.0240, 0.0187, 0.0224, 0.0277, 0.0304, 0.0227, 0.0262], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 21:55:36,002 INFO [train2.py:809] (0/4) Epoch 7, batch 2400, loss[ctc_loss=0.1034, att_loss=0.2471, loss=0.2183, over 16427.00 frames. utt_duration=1495 frames, utt_pad_proportion=0.005459, over 44.00 utterances.], tot_loss[ctc_loss=0.1358, att_loss=0.2674, loss=0.2411, over 3262412.67 frames. utt_duration=1194 frames, utt_pad_proportion=0.07017, over 10943.16 utterances.], batch size: 44, lr: 1.53e-02, grad_scale: 8.0 2023-03-07 21:55:48,834 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9009, 5.1610, 5.4413, 5.3457, 5.2450, 5.7779, 5.0798, 5.9782], device='cuda:0'), covar=tensor([0.0663, 0.0685, 0.0674, 0.0867, 0.1852, 0.0923, 0.0627, 0.0527], device='cuda:0'), in_proj_covar=tensor([0.0607, 0.0364, 0.0411, 0.0471, 0.0649, 0.0415, 0.0338, 0.0421], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 21:55:49,042 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26311.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:55:52,370 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.814e+02 3.013e+02 3.656e+02 4.621e+02 8.184e+02, threshold=7.313e+02, percent-clipped=1.0 2023-03-07 21:56:21,433 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.27 vs. limit=5.0 2023-03-07 21:56:40,147 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0923, 5.1026, 5.1039, 2.8967, 4.9189, 4.3633, 4.3685, 2.1733], device='cuda:0'), covar=tensor([0.0136, 0.0078, 0.0155, 0.1081, 0.0073, 0.0225, 0.0273, 0.1612], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0071, 0.0058, 0.0099, 0.0062, 0.0081, 0.0084, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 21:56:56,977 INFO [train2.py:809] (0/4) Epoch 7, batch 2450, loss[ctc_loss=0.224, att_loss=0.3106, loss=0.2933, over 14335.00 frames. utt_duration=394.3 frames, utt_pad_proportion=0.3131, over 146.00 utterances.], tot_loss[ctc_loss=0.1354, att_loss=0.2672, loss=0.2408, over 3252298.65 frames. utt_duration=1190 frames, utt_pad_proportion=0.07379, over 10949.23 utterances.], batch size: 146, lr: 1.53e-02, grad_scale: 8.0 2023-03-07 21:58:18,161 INFO [train2.py:809] (0/4) Epoch 7, batch 2500, loss[ctc_loss=0.1151, att_loss=0.2362, loss=0.212, over 14580.00 frames. utt_duration=1824 frames, utt_pad_proportion=0.03117, over 32.00 utterances.], tot_loss[ctc_loss=0.135, att_loss=0.2664, loss=0.2401, over 3244829.48 frames. utt_duration=1204 frames, utt_pad_proportion=0.07316, over 10793.17 utterances.], batch size: 32, lr: 1.53e-02, grad_scale: 8.0 2023-03-07 21:58:33,370 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.952e+02 2.816e+02 3.513e+02 4.106e+02 7.996e+02, threshold=7.027e+02, percent-clipped=1.0 2023-03-07 21:58:36,104 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26414.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 21:59:38,002 INFO [train2.py:809] (0/4) Epoch 7, batch 2550, loss[ctc_loss=0.1212, att_loss=0.2659, loss=0.237, over 16320.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006901, over 45.00 utterances.], tot_loss[ctc_loss=0.1343, att_loss=0.2665, loss=0.24, over 3251303.89 frames. utt_duration=1212 frames, utt_pad_proportion=0.0709, over 10747.33 utterances.], batch size: 45, lr: 1.53e-02, grad_scale: 8.0 2023-03-07 21:59:45,999 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26458.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:00:13,372 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26475.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:00:24,425 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-07 22:00:58,230 INFO [train2.py:809] (0/4) Epoch 7, batch 2600, loss[ctc_loss=0.08901, att_loss=0.2256, loss=0.1983, over 12666.00 frames. utt_duration=1811 frames, utt_pad_proportion=0.03105, over 28.00 utterances.], tot_loss[ctc_loss=0.1344, att_loss=0.2665, loss=0.2401, over 3256306.43 frames. utt_duration=1219 frames, utt_pad_proportion=0.06709, over 10698.19 utterances.], batch size: 28, lr: 1.53e-02, grad_scale: 8.0 2023-03-07 22:01:06,474 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7344, 3.8156, 3.0083, 3.5148, 3.9029, 3.5934, 2.6823, 4.3628], device='cuda:0'), covar=tensor([0.1046, 0.0349, 0.1205, 0.0646, 0.0582, 0.0583, 0.0969, 0.0485], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0154, 0.0190, 0.0158, 0.0191, 0.0187, 0.0162, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 22:01:14,579 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.671e+02 3.263e+02 3.835e+02 5.237e+02 1.388e+03, threshold=7.671e+02, percent-clipped=11.0 2023-03-07 22:01:16,499 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26514.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 22:01:31,416 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26524.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:02:18,264 INFO [train2.py:809] (0/4) Epoch 7, batch 2650, loss[ctc_loss=0.09619, att_loss=0.2292, loss=0.2026, over 15640.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009241, over 37.00 utterances.], tot_loss[ctc_loss=0.134, att_loss=0.2663, loss=0.2398, over 3260568.10 frames. utt_duration=1231 frames, utt_pad_proportion=0.06231, over 10605.95 utterances.], batch size: 37, lr: 1.53e-02, grad_scale: 8.0 2023-03-07 22:02:32,520 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26562.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 22:03:37,579 INFO [train2.py:809] (0/4) Epoch 7, batch 2700, loss[ctc_loss=0.1371, att_loss=0.268, loss=0.2418, over 16468.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006463, over 46.00 utterances.], tot_loss[ctc_loss=0.1342, att_loss=0.2665, loss=0.2401, over 3268554.75 frames. utt_duration=1239 frames, utt_pad_proportion=0.0581, over 10561.53 utterances.], batch size: 46, lr: 1.53e-02, grad_scale: 8.0 2023-03-07 22:03:42,445 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26606.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:03:45,697 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2548, 3.2145, 3.2521, 2.8382, 3.1845, 3.0792, 3.1775, 1.7338], device='cuda:0'), covar=tensor([0.1655, 0.1920, 0.2682, 0.5128, 0.1962, 0.5058, 0.1069, 1.0170], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0084, 0.0088, 0.0140, 0.0080, 0.0129, 0.0075, 0.0136], device='cuda:0'), out_proj_covar=tensor([6.5044e-05, 6.5413e-05, 7.4065e-05, 1.0686e-04, 6.6071e-05, 1.0064e-04, 5.9894e-05, 1.0766e-04], device='cuda:0') 2023-03-07 22:03:53,739 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.795e+02 3.023e+02 3.591e+02 4.363e+02 1.241e+03, threshold=7.183e+02, percent-clipped=4.0 2023-03-07 22:04:56,685 INFO [train2.py:809] (0/4) Epoch 7, batch 2750, loss[ctc_loss=0.1252, att_loss=0.2554, loss=0.2294, over 16286.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006823, over 43.00 utterances.], tot_loss[ctc_loss=0.1339, att_loss=0.2664, loss=0.2399, over 3273200.07 frames. utt_duration=1240 frames, utt_pad_proportion=0.057, over 10574.51 utterances.], batch size: 43, lr: 1.52e-02, grad_scale: 8.0 2023-03-07 22:06:15,389 INFO [train2.py:809] (0/4) Epoch 7, batch 2800, loss[ctc_loss=0.09917, att_loss=0.2499, loss=0.2198, over 16765.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.005938, over 48.00 utterances.], tot_loss[ctc_loss=0.133, att_loss=0.2659, loss=0.2393, over 3274748.39 frames. utt_duration=1263 frames, utt_pad_proportion=0.05196, over 10386.66 utterances.], batch size: 48, lr: 1.52e-02, grad_scale: 8.0 2023-03-07 22:06:31,174 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.919e+02 3.082e+02 3.761e+02 4.456e+02 1.465e+03, threshold=7.521e+02, percent-clipped=2.0 2023-03-07 22:06:36,233 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3356, 2.2175, 3.4219, 2.3452, 3.1439, 4.5061, 4.4711, 2.6531], device='cuda:0'), covar=tensor([0.0590, 0.2404, 0.1035, 0.2048, 0.1125, 0.0551, 0.0451, 0.2147], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0218, 0.0219, 0.0199, 0.0225, 0.0249, 0.0190, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 22:07:01,611 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8151, 5.2560, 4.6154, 5.2498, 4.5327, 4.9233, 5.4008, 5.1311], device='cuda:0'), covar=tensor([0.0525, 0.0211, 0.0871, 0.0238, 0.0524, 0.0199, 0.0206, 0.0167], device='cuda:0'), in_proj_covar=tensor([0.0268, 0.0209, 0.0264, 0.0189, 0.0216, 0.0165, 0.0195, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 22:07:13,155 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6079, 4.6573, 4.4185, 4.3822, 5.1069, 4.8032, 4.5022, 1.9018], device='cuda:0'), covar=tensor([0.0202, 0.0430, 0.0388, 0.0385, 0.1103, 0.0152, 0.0390, 0.2965], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0123, 0.0124, 0.0122, 0.0295, 0.0123, 0.0114, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 22:07:17,670 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6845, 4.6568, 4.7043, 2.6612, 4.5463, 4.2083, 3.8348, 2.4200], device='cuda:0'), covar=tensor([0.0108, 0.0066, 0.0103, 0.0999, 0.0074, 0.0203, 0.0341, 0.1323], device='cuda:0'), in_proj_covar=tensor([0.0054, 0.0071, 0.0059, 0.0098, 0.0063, 0.0082, 0.0084, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 22:07:34,539 INFO [train2.py:809] (0/4) Epoch 7, batch 2850, loss[ctc_loss=0.1595, att_loss=0.2863, loss=0.2609, over 16624.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005558, over 47.00 utterances.], tot_loss[ctc_loss=0.1345, att_loss=0.2666, loss=0.2402, over 3266463.61 frames. utt_duration=1229 frames, utt_pad_proportion=0.06268, over 10640.68 utterances.], batch size: 47, lr: 1.52e-02, grad_scale: 8.0 2023-03-07 22:07:34,972 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0010, 4.5947, 4.2395, 4.0439, 2.4818, 4.3012, 2.5309, 1.6782], device='cuda:0'), covar=tensor([0.0246, 0.0110, 0.0709, 0.0286, 0.1998, 0.0153, 0.1551, 0.1860], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0102, 0.0245, 0.0111, 0.0217, 0.0100, 0.0221, 0.0199], device='cuda:0'), out_proj_covar=tensor([1.1565e-04, 1.0342e-04, 2.2118e-04, 1.0346e-04, 2.0259e-04, 9.7584e-05, 1.9922e-04, 1.8110e-04], device='cuda:0') 2023-03-07 22:07:42,493 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26758.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:07:43,920 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0830, 6.1679, 5.7872, 6.0585, 5.9639, 5.5988, 5.6350, 5.6306], device='cuda:0'), covar=tensor([0.1136, 0.0838, 0.0767, 0.0806, 0.0616, 0.1429, 0.2145, 0.2046], device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0437, 0.0326, 0.0345, 0.0313, 0.0389, 0.0454, 0.0415], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 22:08:01,756 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=26770.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:08:48,496 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2186, 5.1974, 5.1869, 3.4277, 5.0462, 4.6499, 4.5019, 2.6626], device='cuda:0'), covar=tensor([0.0128, 0.0063, 0.0112, 0.0754, 0.0064, 0.0158, 0.0254, 0.1317], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0071, 0.0059, 0.0097, 0.0062, 0.0081, 0.0084, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 22:08:54,863 INFO [train2.py:809] (0/4) Epoch 7, batch 2900, loss[ctc_loss=0.1213, att_loss=0.2704, loss=0.2405, over 17325.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02155, over 59.00 utterances.], tot_loss[ctc_loss=0.1339, att_loss=0.2661, loss=0.2396, over 3274119.76 frames. utt_duration=1238 frames, utt_pad_proportion=0.05695, over 10589.34 utterances.], batch size: 59, lr: 1.52e-02, grad_scale: 8.0 2023-03-07 22:08:59,633 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26806.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:09:11,159 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 3.082e+02 3.745e+02 4.484e+02 1.236e+03, threshold=7.491e+02, percent-clipped=3.0 2023-03-07 22:09:30,275 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26824.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:10:16,146 INFO [train2.py:809] (0/4) Epoch 7, batch 2950, loss[ctc_loss=0.1462, att_loss=0.2836, loss=0.2561, over 17108.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01479, over 56.00 utterances.], tot_loss[ctc_loss=0.1337, att_loss=0.2666, loss=0.24, over 3277661.07 frames. utt_duration=1193 frames, utt_pad_proportion=0.0664, over 11007.81 utterances.], batch size: 56, lr: 1.52e-02, grad_scale: 8.0 2023-03-07 22:10:47,878 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26872.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:11:36,459 INFO [train2.py:809] (0/4) Epoch 7, batch 3000, loss[ctc_loss=0.1493, att_loss=0.2768, loss=0.2513, over 16130.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005941, over 42.00 utterances.], tot_loss[ctc_loss=0.1316, att_loss=0.2647, loss=0.2381, over 3273110.15 frames. utt_duration=1227 frames, utt_pad_proportion=0.05877, over 10685.45 utterances.], batch size: 42, lr: 1.52e-02, grad_scale: 8.0 2023-03-07 22:11:36,461 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 22:11:50,142 INFO [train2.py:843] (0/4) Epoch 7, validation: ctc_loss=0.06224, att_loss=0.2434, loss=0.2072, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 22:11:50,143 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16076MB 2023-03-07 22:11:52,050 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26904.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 22:11:55,080 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=26906.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:12:06,322 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.662e+02 2.833e+02 3.331e+02 4.099e+02 1.001e+03, threshold=6.663e+02, percent-clipped=1.0 2023-03-07 22:13:05,207 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=26950.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:13:09,496 INFO [train2.py:809] (0/4) Epoch 7, batch 3050, loss[ctc_loss=0.128, att_loss=0.254, loss=0.2288, over 15892.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008786, over 39.00 utterances.], tot_loss[ctc_loss=0.1326, att_loss=0.2653, loss=0.2387, over 3276601.34 frames. utt_duration=1237 frames, utt_pad_proportion=0.05625, over 10604.25 utterances.], batch size: 39, lr: 1.52e-02, grad_scale: 8.0 2023-03-07 22:13:11,180 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=26954.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:13:29,137 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=26965.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 22:14:29,788 INFO [train2.py:809] (0/4) Epoch 7, batch 3100, loss[ctc_loss=0.1217, att_loss=0.2607, loss=0.2329, over 17334.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02306, over 59.00 utterances.], tot_loss[ctc_loss=0.1327, att_loss=0.2654, loss=0.2389, over 3277746.93 frames. utt_duration=1220 frames, utt_pad_proportion=0.05936, over 10763.95 utterances.], batch size: 59, lr: 1.51e-02, grad_scale: 8.0 2023-03-07 22:14:43,208 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27011.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:14:45,949 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.079e+02 3.057e+02 3.764e+02 4.844e+02 1.301e+03, threshold=7.527e+02, percent-clipped=9.0 2023-03-07 22:15:49,014 INFO [train2.py:809] (0/4) Epoch 7, batch 3150, loss[ctc_loss=0.1633, att_loss=0.2885, loss=0.2634, over 17446.00 frames. utt_duration=1109 frames, utt_pad_proportion=0.0304, over 63.00 utterances.], tot_loss[ctc_loss=0.1333, att_loss=0.2662, loss=0.2396, over 3286703.47 frames. utt_duration=1226 frames, utt_pad_proportion=0.05553, over 10738.42 utterances.], batch size: 63, lr: 1.51e-02, grad_scale: 8.0 2023-03-07 22:16:17,074 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27070.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:17:09,172 INFO [train2.py:809] (0/4) Epoch 7, batch 3200, loss[ctc_loss=0.07594, att_loss=0.2184, loss=0.1899, over 15757.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009578, over 38.00 utterances.], tot_loss[ctc_loss=0.1322, att_loss=0.2653, loss=0.2387, over 3283395.63 frames. utt_duration=1218 frames, utt_pad_proportion=0.05915, over 10792.35 utterances.], batch size: 38, lr: 1.51e-02, grad_scale: 8.0 2023-03-07 22:17:25,181 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.984e+02 2.918e+02 3.542e+02 4.352e+02 6.869e+02, threshold=7.084e+02, percent-clipped=0.0 2023-03-07 22:17:25,694 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1785, 4.8879, 4.8872, 2.4242, 2.0860, 2.5383, 3.7318, 3.7298], device='cuda:0'), covar=tensor([0.0471, 0.0141, 0.0201, 0.3378, 0.5702, 0.2768, 0.0899, 0.1824], device='cuda:0'), in_proj_covar=tensor([0.0310, 0.0200, 0.0220, 0.0184, 0.0359, 0.0345, 0.0216, 0.0344], device='cuda:0'), out_proj_covar=tensor([1.5058e-04, 7.9944e-05, 9.8228e-05, 8.4905e-05, 1.6579e-04, 1.4798e-04, 8.8287e-05, 1.5586e-04], device='cuda:0') 2023-03-07 22:17:32,906 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27118.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:17:57,085 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.96 vs. limit=5.0 2023-03-07 22:18:21,538 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27148.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:18:29,006 INFO [train2.py:809] (0/4) Epoch 7, batch 3250, loss[ctc_loss=0.1228, att_loss=0.256, loss=0.2294, over 16418.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006619, over 44.00 utterances.], tot_loss[ctc_loss=0.132, att_loss=0.2652, loss=0.2386, over 3285326.34 frames. utt_duration=1238 frames, utt_pad_proportion=0.05449, over 10625.40 utterances.], batch size: 44, lr: 1.51e-02, grad_scale: 8.0 2023-03-07 22:19:48,583 INFO [train2.py:809] (0/4) Epoch 7, batch 3300, loss[ctc_loss=0.1373, att_loss=0.2405, loss=0.2199, over 14082.00 frames. utt_duration=1819 frames, utt_pad_proportion=0.04886, over 31.00 utterances.], tot_loss[ctc_loss=0.1316, att_loss=0.2648, loss=0.2382, over 3282199.90 frames. utt_duration=1243 frames, utt_pad_proportion=0.05303, over 10572.75 utterances.], batch size: 31, lr: 1.51e-02, grad_scale: 8.0 2023-03-07 22:19:59,156 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27209.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:20:05,127 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.634e+02 2.966e+02 3.461e+02 4.502e+02 1.067e+03, threshold=6.921e+02, percent-clipped=5.0 2023-03-07 22:20:59,445 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3959, 4.9967, 4.6445, 5.0080, 4.9887, 4.7007, 3.6998, 4.8409], device='cuda:0'), covar=tensor([0.0115, 0.0118, 0.0121, 0.0072, 0.0112, 0.0088, 0.0562, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0060, 0.0071, 0.0046, 0.0048, 0.0058, 0.0083, 0.0078], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-07 22:21:08,401 INFO [train2.py:809] (0/4) Epoch 7, batch 3350, loss[ctc_loss=0.1161, att_loss=0.253, loss=0.2256, over 16474.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006071, over 46.00 utterances.], tot_loss[ctc_loss=0.1316, att_loss=0.265, loss=0.2383, over 3275283.61 frames. utt_duration=1235 frames, utt_pad_proportion=0.05753, over 10623.43 utterances.], batch size: 46, lr: 1.51e-02, grad_scale: 8.0 2023-03-07 22:21:08,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9165, 4.7284, 4.7885, 4.5840, 5.0904, 4.9577, 4.5012, 2.2785], device='cuda:0'), covar=tensor([0.0254, 0.0347, 0.0221, 0.0289, 0.1189, 0.0215, 0.0393, 0.2593], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0126, 0.0127, 0.0126, 0.0303, 0.0127, 0.0115, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 22:21:20,182 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27260.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 22:22:28,310 INFO [train2.py:809] (0/4) Epoch 7, batch 3400, loss[ctc_loss=0.115, att_loss=0.2425, loss=0.217, over 15934.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.008254, over 41.00 utterances.], tot_loss[ctc_loss=0.1311, att_loss=0.2645, loss=0.2378, over 3271262.55 frames. utt_duration=1235 frames, utt_pad_proportion=0.05969, over 10611.45 utterances.], batch size: 41, lr: 1.51e-02, grad_scale: 8.0 2023-03-07 22:22:33,596 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27306.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:22:43,940 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.844e+02 2.932e+02 3.516e+02 4.279e+02 1.181e+03, threshold=7.032e+02, percent-clipped=5.0 2023-03-07 22:23:20,399 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-07 22:23:45,932 INFO [train2.py:809] (0/4) Epoch 7, batch 3450, loss[ctc_loss=0.1697, att_loss=0.2902, loss=0.2661, over 17274.00 frames. utt_duration=876.4 frames, utt_pad_proportion=0.08229, over 79.00 utterances.], tot_loss[ctc_loss=0.1318, att_loss=0.2649, loss=0.2383, over 3265313.56 frames. utt_duration=1235 frames, utt_pad_proportion=0.05981, over 10590.58 utterances.], batch size: 79, lr: 1.51e-02, grad_scale: 16.0 2023-03-07 22:25:06,080 INFO [train2.py:809] (0/4) Epoch 7, batch 3500, loss[ctc_loss=0.1031, att_loss=0.2273, loss=0.2025, over 15389.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.01014, over 35.00 utterances.], tot_loss[ctc_loss=0.133, att_loss=0.2657, loss=0.2392, over 3270219.27 frames. utt_duration=1235 frames, utt_pad_proportion=0.05781, over 10604.09 utterances.], batch size: 35, lr: 1.50e-02, grad_scale: 16.0 2023-03-07 22:25:22,143 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.067e+02 3.129e+02 3.937e+02 5.112e+02 1.151e+03, threshold=7.873e+02, percent-clipped=8.0 2023-03-07 22:26:26,694 INFO [train2.py:809] (0/4) Epoch 7, batch 3550, loss[ctc_loss=0.1167, att_loss=0.2514, loss=0.2245, over 16702.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005029, over 46.00 utterances.], tot_loss[ctc_loss=0.1335, att_loss=0.2666, loss=0.24, over 3268597.92 frames. utt_duration=1237 frames, utt_pad_proportion=0.05553, over 10585.03 utterances.], batch size: 46, lr: 1.50e-02, grad_scale: 16.0 2023-03-07 22:27:04,294 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3744, 2.5500, 3.5364, 2.5392, 3.3019, 4.4670, 4.2294, 2.8758], device='cuda:0'), covar=tensor([0.0398, 0.2132, 0.1102, 0.1761, 0.1108, 0.0699, 0.0654, 0.1706], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0217, 0.0219, 0.0199, 0.0223, 0.0252, 0.0190, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 22:27:46,264 INFO [train2.py:809] (0/4) Epoch 7, batch 3600, loss[ctc_loss=0.1126, att_loss=0.2309, loss=0.2073, over 14562.00 frames. utt_duration=1822 frames, utt_pad_proportion=0.04099, over 32.00 utterances.], tot_loss[ctc_loss=0.1321, att_loss=0.2656, loss=0.2389, over 3270016.22 frames. utt_duration=1264 frames, utt_pad_proportion=0.04916, over 10357.53 utterances.], batch size: 32, lr: 1.50e-02, grad_scale: 16.0 2023-03-07 22:27:48,043 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=27504.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:28:02,318 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.030e+02 2.874e+02 3.597e+02 4.562e+02 1.127e+03, threshold=7.194e+02, percent-clipped=4.0 2023-03-07 22:28:44,061 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7289, 4.9100, 4.9781, 4.7923, 5.0444, 5.0575, 4.7114, 4.5055], device='cuda:0'), covar=tensor([0.0889, 0.0636, 0.0234, 0.0570, 0.0266, 0.0263, 0.0321, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0412, 0.0252, 0.0193, 0.0234, 0.0290, 0.0313, 0.0242, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-07 22:28:54,947 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3169, 3.4237, 3.5792, 2.9028, 3.5597, 3.5245, 3.5939, 2.4017], device='cuda:0'), covar=tensor([0.1421, 0.1581, 0.3825, 0.6259, 0.1229, 0.3452, 0.0713, 0.9548], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0084, 0.0089, 0.0144, 0.0079, 0.0130, 0.0074, 0.0135], device='cuda:0'), out_proj_covar=tensor([6.5079e-05, 6.6475e-05, 7.5123e-05, 1.1046e-04, 6.6656e-05, 1.0215e-04, 6.0278e-05, 1.0734e-04], device='cuda:0') 2023-03-07 22:29:06,203 INFO [train2.py:809] (0/4) Epoch 7, batch 3650, loss[ctc_loss=0.09471, att_loss=0.2324, loss=0.2049, over 16016.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006815, over 40.00 utterances.], tot_loss[ctc_loss=0.1317, att_loss=0.265, loss=0.2383, over 3269114.86 frames. utt_duration=1278 frames, utt_pad_proportion=0.04708, over 10245.98 utterances.], batch size: 40, lr: 1.50e-02, grad_scale: 16.0 2023-03-07 22:29:18,182 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27560.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 22:30:27,716 INFO [train2.py:809] (0/4) Epoch 7, batch 3700, loss[ctc_loss=0.1184, att_loss=0.2679, loss=0.238, over 17411.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.04664, over 69.00 utterances.], tot_loss[ctc_loss=0.1318, att_loss=0.2645, loss=0.2379, over 3262644.06 frames. utt_duration=1249 frames, utt_pad_proportion=0.05497, over 10460.09 utterances.], batch size: 69, lr: 1.50e-02, grad_scale: 16.0 2023-03-07 22:30:33,188 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27606.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:30:36,280 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27608.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 22:30:44,494 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.774e+02 3.024e+02 4.004e+02 4.895e+02 1.157e+03, threshold=8.007e+02, percent-clipped=6.0 2023-03-07 22:30:44,843 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6011, 5.1535, 4.9078, 5.1963, 5.1502, 4.7742, 3.7810, 5.0282], device='cuda:0'), covar=tensor([0.0103, 0.0076, 0.0098, 0.0068, 0.0067, 0.0086, 0.0501, 0.0156], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0060, 0.0071, 0.0046, 0.0047, 0.0058, 0.0082, 0.0079], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-07 22:31:03,515 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9069, 4.7591, 4.6644, 2.9381, 4.2604, 4.2067, 4.0597, 2.2159], device='cuda:0'), covar=tensor([0.0102, 0.0107, 0.0187, 0.0930, 0.0144, 0.0196, 0.0306, 0.1566], device='cuda:0'), in_proj_covar=tensor([0.0053, 0.0074, 0.0060, 0.0101, 0.0064, 0.0081, 0.0085, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 22:31:12,037 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.48 vs. limit=5.0 2023-03-07 22:31:47,382 INFO [train2.py:809] (0/4) Epoch 7, batch 3750, loss[ctc_loss=0.08852, att_loss=0.2322, loss=0.2034, over 15995.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008055, over 40.00 utterances.], tot_loss[ctc_loss=0.1319, att_loss=0.2645, loss=0.238, over 3264127.71 frames. utt_duration=1246 frames, utt_pad_proportion=0.05582, over 10487.75 utterances.], batch size: 40, lr: 1.50e-02, grad_scale: 16.0 2023-03-07 22:31:49,611 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27654.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:33:07,742 INFO [train2.py:809] (0/4) Epoch 7, batch 3800, loss[ctc_loss=0.2023, att_loss=0.3145, loss=0.2921, over 14210.00 frames. utt_duration=391 frames, utt_pad_proportion=0.3189, over 146.00 utterances.], tot_loss[ctc_loss=0.1316, att_loss=0.2645, loss=0.2379, over 3256256.37 frames. utt_duration=1243 frames, utt_pad_proportion=0.05927, over 10493.29 utterances.], batch size: 146, lr: 1.50e-02, grad_scale: 16.0 2023-03-07 22:33:25,167 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.970e+02 3.032e+02 3.559e+02 4.389e+02 6.627e+02, threshold=7.117e+02, percent-clipped=0.0 2023-03-07 22:33:27,139 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=27714.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 22:34:28,354 INFO [train2.py:809] (0/4) Epoch 7, batch 3850, loss[ctc_loss=0.1284, att_loss=0.241, loss=0.2185, over 14548.00 frames. utt_duration=1820 frames, utt_pad_proportion=0.0339, over 32.00 utterances.], tot_loss[ctc_loss=0.1309, att_loss=0.2643, loss=0.2376, over 3263786.99 frames. utt_duration=1244 frames, utt_pad_proportion=0.05773, over 10504.35 utterances.], batch size: 32, lr: 1.49e-02, grad_scale: 16.0 2023-03-07 22:35:03,295 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=27775.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 22:35:46,402 INFO [train2.py:809] (0/4) Epoch 7, batch 3900, loss[ctc_loss=0.1047, att_loss=0.24, loss=0.2129, over 16020.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.007012, over 40.00 utterances.], tot_loss[ctc_loss=0.1313, att_loss=0.2649, loss=0.2382, over 3267171.55 frames. utt_duration=1242 frames, utt_pad_proportion=0.05742, over 10537.55 utterances.], batch size: 40, lr: 1.49e-02, grad_scale: 16.0 2023-03-07 22:35:48,214 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=27804.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:36:01,933 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.865e+02 3.007e+02 3.697e+02 4.485e+02 8.676e+02, threshold=7.394e+02, percent-clipped=2.0 2023-03-07 22:37:02,131 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=27852.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:37:03,593 INFO [train2.py:809] (0/4) Epoch 7, batch 3950, loss[ctc_loss=0.1096, att_loss=0.2481, loss=0.2204, over 16251.00 frames. utt_duration=1513 frames, utt_pad_proportion=0.008504, over 43.00 utterances.], tot_loss[ctc_loss=0.1314, att_loss=0.2651, loss=0.2384, over 3276609.40 frames. utt_duration=1241 frames, utt_pad_proportion=0.0547, over 10575.12 utterances.], batch size: 43, lr: 1.49e-02, grad_scale: 16.0 2023-03-07 22:37:48,136 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9814, 5.2485, 5.5851, 5.4994, 5.2778, 5.8695, 5.1178, 6.0271], device='cuda:0'), covar=tensor([0.0582, 0.0578, 0.0521, 0.0888, 0.1794, 0.0831, 0.0509, 0.0509], device='cuda:0'), in_proj_covar=tensor([0.0594, 0.0362, 0.0408, 0.0470, 0.0636, 0.0416, 0.0336, 0.0416], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 22:37:55,159 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-7.pt 2023-03-07 22:38:22,292 INFO [train2.py:809] (0/4) Epoch 8, batch 0, loss[ctc_loss=0.1506, att_loss=0.275, loss=0.2502, over 17011.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.008559, over 51.00 utterances.], tot_loss[ctc_loss=0.1506, att_loss=0.275, loss=0.2502, over 17011.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.008559, over 51.00 utterances.], batch size: 51, lr: 1.40e-02, grad_scale: 8.0 2023-03-07 22:38:22,294 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 22:38:34,610 INFO [train2.py:843] (0/4) Epoch 8, validation: ctc_loss=0.06098, att_loss=0.2435, loss=0.207, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 22:38:34,611 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16076MB 2023-03-07 22:39:19,192 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.946e+02 2.808e+02 3.476e+02 4.336e+02 8.495e+02, threshold=6.952e+02, percent-clipped=4.0 2023-03-07 22:39:49,751 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2975, 4.8063, 4.8167, 4.7065, 2.6073, 4.9535, 2.4826, 2.2222], device='cuda:0'), covar=tensor([0.0218, 0.0151, 0.0591, 0.0208, 0.2079, 0.0150, 0.1665, 0.1650], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0100, 0.0254, 0.0114, 0.0224, 0.0101, 0.0228, 0.0206], device='cuda:0'), out_proj_covar=tensor([1.2005e-04, 1.0292e-04, 2.3046e-04, 1.0715e-04, 2.1040e-04, 9.9420e-05, 2.0616e-04, 1.8771e-04], device='cuda:0') 2023-03-07 22:39:54,574 INFO [train2.py:809] (0/4) Epoch 8, batch 50, loss[ctc_loss=0.1239, att_loss=0.2683, loss=0.2394, over 16542.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006137, over 45.00 utterances.], tot_loss[ctc_loss=0.1299, att_loss=0.2659, loss=0.2387, over 745761.14 frames. utt_duration=1162 frames, utt_pad_proportion=0.06132, over 2569.46 utterances.], batch size: 45, lr: 1.40e-02, grad_scale: 4.0 2023-03-07 22:40:52,063 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0749, 4.8918, 4.8813, 2.4194, 1.8875, 2.8438, 4.0435, 3.7755], device='cuda:0'), covar=tensor([0.0596, 0.0140, 0.0163, 0.3412, 0.5855, 0.2263, 0.0657, 0.1654], device='cuda:0'), in_proj_covar=tensor([0.0308, 0.0194, 0.0219, 0.0182, 0.0350, 0.0340, 0.0212, 0.0339], device='cuda:0'), out_proj_covar=tensor([1.4831e-04, 7.6864e-05, 9.7208e-05, 8.3502e-05, 1.6112e-04, 1.4467e-04, 8.6274e-05, 1.5294e-04], device='cuda:0') 2023-03-07 22:41:14,412 INFO [train2.py:809] (0/4) Epoch 8, batch 100, loss[ctc_loss=0.09639, att_loss=0.2365, loss=0.2085, over 15505.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008354, over 36.00 utterances.], tot_loss[ctc_loss=0.1299, att_loss=0.2643, loss=0.2375, over 1294533.95 frames. utt_duration=1195 frames, utt_pad_proportion=0.0677, over 4338.91 utterances.], batch size: 36, lr: 1.40e-02, grad_scale: 4.0 2023-03-07 22:41:36,096 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-28000.pt 2023-03-07 22:41:43,034 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1312, 5.3556, 5.6440, 5.6095, 5.4932, 6.0525, 5.2459, 6.2029], device='cuda:0'), covar=tensor([0.0523, 0.0545, 0.0519, 0.0902, 0.1474, 0.0685, 0.0468, 0.0403], device='cuda:0'), in_proj_covar=tensor([0.0594, 0.0364, 0.0408, 0.0474, 0.0636, 0.0418, 0.0334, 0.0416], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-07 22:42:02,729 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.935e+02 2.962e+02 3.730e+02 4.479e+02 1.131e+03, threshold=7.459e+02, percent-clipped=8.0 2023-03-07 22:42:39,309 INFO [train2.py:809] (0/4) Epoch 8, batch 150, loss[ctc_loss=0.1066, att_loss=0.2363, loss=0.2104, over 14529.00 frames. utt_duration=1818 frames, utt_pad_proportion=0.04485, over 32.00 utterances.], tot_loss[ctc_loss=0.1288, att_loss=0.2632, loss=0.2363, over 1731314.02 frames. utt_duration=1230 frames, utt_pad_proportion=0.05874, over 5638.15 utterances.], batch size: 32, lr: 1.40e-02, grad_scale: 4.0 2023-03-07 22:43:02,853 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-03-07 22:43:31,915 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28070.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 22:44:00,280 INFO [train2.py:809] (0/4) Epoch 8, batch 200, loss[ctc_loss=0.1229, att_loss=0.2717, loss=0.242, over 16330.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006232, over 45.00 utterances.], tot_loss[ctc_loss=0.1297, att_loss=0.2636, loss=0.2368, over 2062861.44 frames. utt_duration=1189 frames, utt_pad_proportion=0.07327, over 6951.08 utterances.], batch size: 45, lr: 1.40e-02, grad_scale: 4.0 2023-03-07 22:44:44,919 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.776e+02 2.763e+02 3.553e+02 4.276e+02 6.955e+02, threshold=7.106e+02, percent-clipped=0.0 2023-03-07 22:44:52,118 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-03-07 22:45:21,331 INFO [train2.py:809] (0/4) Epoch 8, batch 250, loss[ctc_loss=0.1202, att_loss=0.2661, loss=0.2369, over 16769.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006377, over 48.00 utterances.], tot_loss[ctc_loss=0.1297, att_loss=0.2644, loss=0.2375, over 2331468.51 frames. utt_duration=1172 frames, utt_pad_proportion=0.07695, over 7969.19 utterances.], batch size: 48, lr: 1.40e-02, grad_scale: 4.0 2023-03-07 22:46:41,308 INFO [train2.py:809] (0/4) Epoch 8, batch 300, loss[ctc_loss=0.1989, att_loss=0.3046, loss=0.2835, over 14471.00 frames. utt_duration=398.1 frames, utt_pad_proportion=0.3077, over 146.00 utterances.], tot_loss[ctc_loss=0.1287, att_loss=0.2634, loss=0.2364, over 2531161.57 frames. utt_duration=1201 frames, utt_pad_proportion=0.07002, over 8439.28 utterances.], batch size: 146, lr: 1.40e-02, grad_scale: 4.0 2023-03-07 22:47:27,067 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.936e+02 2.948e+02 3.357e+02 4.007e+02 1.001e+03, threshold=6.714e+02, percent-clipped=3.0 2023-03-07 22:47:33,663 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8912, 4.1304, 3.7581, 4.1562, 3.7584, 3.6798, 4.1583, 4.0865], device='cuda:0'), covar=tensor([0.0486, 0.0276, 0.0752, 0.0243, 0.0432, 0.0831, 0.0276, 0.0189], device='cuda:0'), in_proj_covar=tensor([0.0275, 0.0213, 0.0271, 0.0197, 0.0223, 0.0170, 0.0199, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0004, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 22:47:51,023 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28229.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:48:03,885 INFO [train2.py:809] (0/4) Epoch 8, batch 350, loss[ctc_loss=0.1346, att_loss=0.2779, loss=0.2492, over 16972.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007188, over 50.00 utterances.], tot_loss[ctc_loss=0.1286, att_loss=0.2638, loss=0.2368, over 2694539.36 frames. utt_duration=1186 frames, utt_pad_proportion=0.07227, over 9098.12 utterances.], batch size: 50, lr: 1.40e-02, grad_scale: 4.0 2023-03-07 22:48:33,608 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28255.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 22:49:25,165 INFO [train2.py:809] (0/4) Epoch 8, batch 400, loss[ctc_loss=0.1323, att_loss=0.2493, loss=0.2259, over 16002.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.005602, over 40.00 utterances.], tot_loss[ctc_loss=0.128, att_loss=0.2634, loss=0.2363, over 2826825.94 frames. utt_duration=1203 frames, utt_pad_proportion=0.06628, over 9413.48 utterances.], batch size: 40, lr: 1.40e-02, grad_scale: 8.0 2023-03-07 22:49:25,532 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28287.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:49:30,661 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28290.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:50:09,511 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.601e+02 3.135e+02 4.187e+02 9.825e+02, threshold=6.270e+02, percent-clipped=5.0 2023-03-07 22:50:11,477 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28316.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 22:50:37,334 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-07 22:50:45,304 INFO [train2.py:809] (0/4) Epoch 8, batch 450, loss[ctc_loss=0.1482, att_loss=0.2746, loss=0.2493, over 16461.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.007693, over 46.00 utterances.], tot_loss[ctc_loss=0.1266, att_loss=0.2624, loss=0.2353, over 2926351.66 frames. utt_duration=1228 frames, utt_pad_proportion=0.06019, over 9541.90 utterances.], batch size: 46, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 22:51:03,314 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28348.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:51:37,018 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28370.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 22:52:04,593 INFO [train2.py:809] (0/4) Epoch 8, batch 500, loss[ctc_loss=0.1466, att_loss=0.2887, loss=0.2603, over 17306.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01158, over 55.00 utterances.], tot_loss[ctc_loss=0.1261, att_loss=0.2623, loss=0.235, over 3005759.63 frames. utt_duration=1246 frames, utt_pad_proportion=0.05503, over 9660.94 utterances.], batch size: 55, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 22:52:36,980 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3760, 2.5294, 3.3204, 4.5042, 4.1853, 3.9595, 2.9235, 1.8906], device='cuda:0'), covar=tensor([0.0719, 0.2550, 0.1294, 0.0513, 0.0780, 0.0450, 0.1696, 0.2946], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0199, 0.0192, 0.0176, 0.0164, 0.0136, 0.0190, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 22:52:38,342 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8577, 6.0390, 5.4595, 5.9348, 5.7020, 5.2390, 5.4544, 5.3108], device='cuda:0'), covar=tensor([0.1259, 0.0891, 0.0902, 0.0633, 0.0724, 0.1444, 0.2324, 0.1974], device='cuda:0'), in_proj_covar=tensor([0.0384, 0.0432, 0.0331, 0.0341, 0.0318, 0.0392, 0.0456, 0.0405], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 22:52:49,177 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.916e+02 2.823e+02 3.661e+02 4.475e+02 7.657e+02, threshold=7.322e+02, percent-clipped=5.0 2023-03-07 22:52:54,234 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28418.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 22:53:25,908 INFO [train2.py:809] (0/4) Epoch 8, batch 550, loss[ctc_loss=0.1179, att_loss=0.2647, loss=0.2354, over 17046.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.009202, over 52.00 utterances.], tot_loss[ctc_loss=0.1276, att_loss=0.2628, loss=0.2357, over 3063286.78 frames. utt_duration=1202 frames, utt_pad_proportion=0.0663, over 10208.00 utterances.], batch size: 52, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 22:53:45,930 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4053, 4.9106, 4.4649, 4.9925, 2.9780, 4.8305, 2.5630, 2.5890], device='cuda:0'), covar=tensor([0.0187, 0.0116, 0.0781, 0.0152, 0.1901, 0.0115, 0.1811, 0.1599], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0099, 0.0255, 0.0112, 0.0220, 0.0100, 0.0226, 0.0203], device='cuda:0'), out_proj_covar=tensor([1.2145e-04, 1.0206e-04, 2.3120e-04, 1.0523e-04, 2.0797e-04, 9.7912e-05, 2.0470e-04, 1.8602e-04], device='cuda:0') 2023-03-07 22:54:43,094 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28486.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:54:44,262 INFO [train2.py:809] (0/4) Epoch 8, batch 600, loss[ctc_loss=0.1259, att_loss=0.2863, loss=0.2542, over 17033.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009281, over 52.00 utterances.], tot_loss[ctc_loss=0.1281, att_loss=0.2632, loss=0.2362, over 3106651.12 frames. utt_duration=1206 frames, utt_pad_proportion=0.06711, over 10317.56 utterances.], batch size: 52, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 22:54:52,839 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28492.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:55:27,679 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.904e+02 2.636e+02 3.439e+02 4.247e+02 1.142e+03, threshold=6.879e+02, percent-clipped=2.0 2023-03-07 22:56:03,434 INFO [train2.py:809] (0/4) Epoch 8, batch 650, loss[ctc_loss=0.1134, att_loss=0.2429, loss=0.217, over 16002.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007605, over 40.00 utterances.], tot_loss[ctc_loss=0.1282, att_loss=0.2633, loss=0.2362, over 3146847.65 frames. utt_duration=1220 frames, utt_pad_proportion=0.06357, over 10332.57 utterances.], batch size: 40, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 22:56:19,717 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28547.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:56:29,166 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28553.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:56:57,639 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2727, 3.5644, 3.4599, 2.8081, 3.5162, 3.3459, 3.2097, 2.3229], device='cuda:0'), covar=tensor([0.1362, 0.0872, 0.2586, 0.6552, 0.1229, 0.2001, 0.1040, 0.8715], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0088, 0.0091, 0.0150, 0.0084, 0.0134, 0.0082, 0.0140], device='cuda:0'), out_proj_covar=tensor([6.8979e-05, 7.0105e-05, 7.7302e-05, 1.1587e-04, 7.1573e-05, 1.0650e-04, 6.6655e-05, 1.1202e-04], device='cuda:0') 2023-03-07 22:57:21,184 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28585.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 22:57:24,651 INFO [train2.py:809] (0/4) Epoch 8, batch 700, loss[ctc_loss=0.1458, att_loss=0.2742, loss=0.2485, over 16118.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.00616, over 42.00 utterances.], tot_loss[ctc_loss=0.1276, att_loss=0.2628, loss=0.2358, over 3176099.26 frames. utt_duration=1225 frames, utt_pad_proportion=0.0609, over 10380.13 utterances.], batch size: 42, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 22:57:36,860 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-03-07 22:58:02,976 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28611.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 22:58:08,929 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.741e+02 2.921e+02 3.565e+02 4.512e+02 9.961e+02, threshold=7.130e+02, percent-clipped=4.0 2023-03-07 22:58:29,164 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6658, 3.9042, 3.4681, 3.5672, 2.4453, 3.5357, 2.6358, 1.9515], device='cuda:0'), covar=tensor([0.0270, 0.0141, 0.0836, 0.0230, 0.1938, 0.0187, 0.1326, 0.1624], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0098, 0.0252, 0.0112, 0.0219, 0.0100, 0.0225, 0.0203], device='cuda:0'), out_proj_covar=tensor([1.1996e-04, 1.0187e-04, 2.2921e-04, 1.0567e-04, 2.0675e-04, 9.7816e-05, 2.0441e-04, 1.8514e-04], device='cuda:0') 2023-03-07 22:58:44,785 INFO [train2.py:809] (0/4) Epoch 8, batch 750, loss[ctc_loss=0.1338, att_loss=0.2683, loss=0.2414, over 17361.00 frames. utt_duration=880.3 frames, utt_pad_proportion=0.07918, over 79.00 utterances.], tot_loss[ctc_loss=0.1267, att_loss=0.2624, loss=0.2352, over 3193445.69 frames. utt_duration=1241 frames, utt_pad_proportion=0.05715, over 10303.73 utterances.], batch size: 79, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 22:58:55,032 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28643.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:00:04,388 INFO [train2.py:809] (0/4) Epoch 8, batch 800, loss[ctc_loss=0.1307, att_loss=0.2512, loss=0.2271, over 15487.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.009062, over 36.00 utterances.], tot_loss[ctc_loss=0.127, att_loss=0.2627, loss=0.2356, over 3211978.69 frames. utt_duration=1244 frames, utt_pad_proportion=0.05517, over 10337.29 utterances.], batch size: 36, lr: 1.39e-02, grad_scale: 8.0 2023-03-07 23:00:47,880 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.805e+02 2.816e+02 3.750e+02 4.467e+02 1.351e+03, threshold=7.500e+02, percent-clipped=6.0 2023-03-07 23:00:49,851 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8036, 1.9463, 1.6083, 1.5004, 3.3936, 1.6194, 1.6725, 1.1353], device='cuda:0'), covar=tensor([0.0901, 0.2926, 0.2506, 0.1832, 0.0368, 0.2054, 0.2913, 0.2722], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0079, 0.0084, 0.0074, 0.0071, 0.0071, 0.0081, 0.0065], device='cuda:0'), out_proj_covar=tensor([3.9749e-05, 5.0084e-05, 4.9933e-05, 4.2177e-05, 3.7064e-05, 4.3418e-05, 4.9968e-05, 4.1688e-05], device='cuda:0') 2023-03-07 23:01:24,435 INFO [train2.py:809] (0/4) Epoch 8, batch 850, loss[ctc_loss=0.1008, att_loss=0.2438, loss=0.2152, over 16170.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006872, over 41.00 utterances.], tot_loss[ctc_loss=0.1261, att_loss=0.2626, loss=0.2353, over 3227150.63 frames. utt_duration=1233 frames, utt_pad_proportion=0.05844, over 10478.72 utterances.], batch size: 41, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:01:30,790 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8065, 4.8018, 4.6956, 2.4055, 4.6295, 4.2402, 3.9980, 2.4832], device='cuda:0'), covar=tensor([0.0111, 0.0075, 0.0176, 0.1260, 0.0083, 0.0221, 0.0352, 0.1445], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0073, 0.0061, 0.0098, 0.0063, 0.0081, 0.0084, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:02:43,214 INFO [train2.py:809] (0/4) Epoch 8, batch 900, loss[ctc_loss=0.13, att_loss=0.2693, loss=0.2414, over 16391.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.00753, over 44.00 utterances.], tot_loss[ctc_loss=0.1267, att_loss=0.2631, loss=0.2358, over 3242341.52 frames. utt_duration=1229 frames, utt_pad_proportion=0.05704, over 10563.87 utterances.], batch size: 44, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:03:27,129 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.754e+02 2.976e+02 3.721e+02 4.391e+02 9.491e+02, threshold=7.443e+02, percent-clipped=5.0 2023-03-07 23:03:45,719 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-03-07 23:03:54,178 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6895, 5.2289, 4.8330, 5.0967, 5.2862, 4.8459, 3.6765, 5.1819], device='cuda:0'), covar=tensor([0.0094, 0.0100, 0.0097, 0.0098, 0.0056, 0.0086, 0.0573, 0.0134], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0062, 0.0074, 0.0047, 0.0049, 0.0060, 0.0085, 0.0083], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-07 23:04:03,681 INFO [train2.py:809] (0/4) Epoch 8, batch 950, loss[ctc_loss=0.128, att_loss=0.2563, loss=0.2306, over 16002.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007574, over 40.00 utterances.], tot_loss[ctc_loss=0.1265, att_loss=0.2626, loss=0.2354, over 3246423.70 frames. utt_duration=1225 frames, utt_pad_proportion=0.05955, over 10617.02 utterances.], batch size: 40, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:04:12,007 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28842.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:04:21,442 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=28848.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:04:42,040 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=28861.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:05:16,426 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7567, 3.7580, 3.0950, 3.4018, 3.9772, 3.5027, 2.4546, 4.3393], device='cuda:0'), covar=tensor([0.1121, 0.0417, 0.0984, 0.0759, 0.0494, 0.0678, 0.1100, 0.0465], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0157, 0.0191, 0.0161, 0.0198, 0.0192, 0.0168, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 23:05:21,458 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28885.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:05:24,288 INFO [train2.py:809] (0/4) Epoch 8, batch 1000, loss[ctc_loss=0.1455, att_loss=0.2847, loss=0.2568, over 17062.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008389, over 52.00 utterances.], tot_loss[ctc_loss=0.1274, att_loss=0.2636, loss=0.2363, over 3250389.35 frames. utt_duration=1200 frames, utt_pad_proportion=0.06602, over 10844.06 utterances.], batch size: 52, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:05:26,127 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2686, 5.5834, 5.1087, 5.6464, 5.0680, 5.2166, 5.6799, 5.4798], device='cuda:0'), covar=tensor([0.0409, 0.0199, 0.0643, 0.0146, 0.0404, 0.0151, 0.0168, 0.0122], device='cuda:0'), in_proj_covar=tensor([0.0283, 0.0214, 0.0279, 0.0202, 0.0228, 0.0176, 0.0200, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 23:05:58,527 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1746, 4.5027, 4.3146, 4.3627, 2.1307, 4.3016, 2.3942, 1.4198], device='cuda:0'), covar=tensor([0.0292, 0.0134, 0.0760, 0.0239, 0.2420, 0.0202, 0.1775, 0.2225], device='cuda:0'), in_proj_covar=tensor([0.0119, 0.0099, 0.0250, 0.0114, 0.0218, 0.0101, 0.0225, 0.0201], device='cuda:0'), out_proj_covar=tensor([1.1962e-04, 1.0230e-04, 2.2698e-04, 1.0767e-04, 2.0591e-04, 9.9453e-05, 2.0410e-04, 1.8381e-04], device='cuda:0') 2023-03-07 23:06:01,434 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28911.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 23:06:07,293 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.707e+02 2.634e+02 3.393e+02 4.335e+02 1.819e+03, threshold=6.785e+02, percent-clipped=6.0 2023-03-07 23:06:20,094 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=28922.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:06:36,964 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28933.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:06:40,325 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4924, 2.4387, 2.9439, 4.2181, 3.8420, 4.0671, 2.8685, 1.9269], device='cuda:0'), covar=tensor([0.0610, 0.2319, 0.1402, 0.0628, 0.0624, 0.0298, 0.1574, 0.2613], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0192, 0.0185, 0.0172, 0.0162, 0.0133, 0.0185, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 23:06:43,109 INFO [train2.py:809] (0/4) Epoch 8, batch 1050, loss[ctc_loss=0.1002, att_loss=0.2364, loss=0.2091, over 15639.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009194, over 37.00 utterances.], tot_loss[ctc_loss=0.126, att_loss=0.2621, loss=0.2349, over 3253360.12 frames. utt_duration=1236 frames, utt_pad_proportion=0.05756, over 10537.61 utterances.], batch size: 37, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:06:53,349 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=28943.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:07:03,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-03-07 23:07:17,997 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28959.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 23:08:03,726 INFO [train2.py:809] (0/4) Epoch 8, batch 1100, loss[ctc_loss=0.1106, att_loss=0.2559, loss=0.2268, over 16323.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006642, over 45.00 utterances.], tot_loss[ctc_loss=0.1248, att_loss=0.2614, loss=0.2341, over 3257920.47 frames. utt_duration=1241 frames, utt_pad_proportion=0.05676, over 10510.44 utterances.], batch size: 45, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:08:10,573 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=28991.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:08:28,598 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29002.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:08:48,755 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.591e+02 2.718e+02 3.264e+02 3.994e+02 7.593e+02, threshold=6.528e+02, percent-clipped=2.0 2023-03-07 23:09:25,649 INFO [train2.py:809] (0/4) Epoch 8, batch 1150, loss[ctc_loss=0.1078, att_loss=0.2594, loss=0.2291, over 16698.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005074, over 46.00 utterances.], tot_loss[ctc_loss=0.1243, att_loss=0.2614, loss=0.234, over 3266110.34 frames. utt_duration=1238 frames, utt_pad_proportion=0.05624, over 10566.55 utterances.], batch size: 46, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:10:04,164 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-07 23:10:06,625 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29063.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:10:25,873 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5203, 4.6961, 4.5559, 4.5302, 5.0517, 4.8847, 4.2200, 2.1345], device='cuda:0'), covar=tensor([0.0308, 0.0266, 0.0287, 0.0231, 0.1237, 0.0256, 0.0408, 0.2605], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0123, 0.0127, 0.0127, 0.0310, 0.0125, 0.0116, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 23:10:38,480 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-07 23:10:46,210 INFO [train2.py:809] (0/4) Epoch 8, batch 1200, loss[ctc_loss=0.09017, att_loss=0.245, loss=0.214, over 16486.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005499, over 46.00 utterances.], tot_loss[ctc_loss=0.1243, att_loss=0.2613, loss=0.2339, over 3266607.10 frames. utt_duration=1236 frames, utt_pad_proportion=0.05702, over 10585.38 utterances.], batch size: 46, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:11:20,545 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6972, 4.9994, 4.8987, 4.9701, 5.1260, 5.0655, 4.7575, 4.5550], device='cuda:0'), covar=tensor([0.1150, 0.0549, 0.0249, 0.0422, 0.0276, 0.0333, 0.0275, 0.0359], device='cuda:0'), in_proj_covar=tensor([0.0431, 0.0264, 0.0203, 0.0242, 0.0305, 0.0334, 0.0252, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:11:30,217 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.598e+02 2.794e+02 3.105e+02 3.921e+02 8.857e+02, threshold=6.210e+02, percent-clipped=2.0 2023-03-07 23:12:06,183 INFO [train2.py:809] (0/4) Epoch 8, batch 1250, loss[ctc_loss=0.1195, att_loss=0.2759, loss=0.2446, over 17294.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01216, over 55.00 utterances.], tot_loss[ctc_loss=0.1252, att_loss=0.2616, loss=0.2343, over 3271378.41 frames. utt_duration=1242 frames, utt_pad_proportion=0.0542, over 10550.16 utterances.], batch size: 55, lr: 1.38e-02, grad_scale: 8.0 2023-03-07 23:12:14,250 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29142.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:12:23,503 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29148.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:12:28,246 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6391, 4.7308, 4.5624, 4.5146, 5.1825, 5.0147, 4.3157, 2.1440], device='cuda:0'), covar=tensor([0.0251, 0.0278, 0.0313, 0.0278, 0.0985, 0.0185, 0.0395, 0.2460], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0123, 0.0129, 0.0127, 0.0308, 0.0125, 0.0115, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-07 23:13:04,874 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-07 23:13:25,930 INFO [train2.py:809] (0/4) Epoch 8, batch 1300, loss[ctc_loss=0.1401, att_loss=0.2507, loss=0.2286, over 16279.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007204, over 43.00 utterances.], tot_loss[ctc_loss=0.1251, att_loss=0.2613, loss=0.2341, over 3268375.49 frames. utt_duration=1253 frames, utt_pad_proportion=0.05377, over 10447.70 utterances.], batch size: 43, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:13:30,501 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29190.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:13:39,805 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29196.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:14:09,580 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.817e+02 2.738e+02 3.321e+02 4.360e+02 1.126e+03, threshold=6.642e+02, percent-clipped=8.0 2023-03-07 23:14:13,580 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29217.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:14:44,863 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29236.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:14:46,055 INFO [train2.py:809] (0/4) Epoch 8, batch 1350, loss[ctc_loss=0.1502, att_loss=0.2718, loss=0.2475, over 16622.00 frames. utt_duration=673.1 frames, utt_pad_proportion=0.1555, over 99.00 utterances.], tot_loss[ctc_loss=0.1258, att_loss=0.2615, loss=0.2343, over 3265468.45 frames. utt_duration=1247 frames, utt_pad_proportion=0.05418, over 10490.21 utterances.], batch size: 99, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:16:06,778 INFO [train2.py:809] (0/4) Epoch 8, batch 1400, loss[ctc_loss=0.1074, att_loss=0.2392, loss=0.2129, over 15764.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.00917, over 38.00 utterances.], tot_loss[ctc_loss=0.1259, att_loss=0.2619, loss=0.2347, over 3269926.09 frames. utt_duration=1235 frames, utt_pad_proportion=0.05602, over 10607.50 utterances.], batch size: 38, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:16:23,170 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29297.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:16:51,999 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.810e+02 3.377e+02 4.369e+02 8.474e+02, threshold=6.754e+02, percent-clipped=4.0 2023-03-07 23:17:12,845 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8757, 5.1635, 5.0398, 5.0894, 5.2439, 5.2050, 4.9731, 4.6723], device='cuda:0'), covar=tensor([0.0959, 0.0413, 0.0225, 0.0425, 0.0232, 0.0250, 0.0272, 0.0305], device='cuda:0'), in_proj_covar=tensor([0.0428, 0.0262, 0.0202, 0.0245, 0.0302, 0.0334, 0.0252, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:17:27,564 INFO [train2.py:809] (0/4) Epoch 8, batch 1450, loss[ctc_loss=0.1209, att_loss=0.2597, loss=0.232, over 16171.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006872, over 41.00 utterances.], tot_loss[ctc_loss=0.1257, att_loss=0.2621, loss=0.2348, over 3277464.56 frames. utt_duration=1241 frames, utt_pad_proportion=0.05287, over 10581.06 utterances.], batch size: 41, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:17:45,067 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4919, 3.7538, 3.5435, 2.8872, 3.6565, 3.6619, 3.4934, 2.3730], device='cuda:0'), covar=tensor([0.1143, 0.1351, 0.4195, 0.7241, 0.1416, 0.3863, 0.0930, 0.9695], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0088, 0.0094, 0.0145, 0.0083, 0.0138, 0.0081, 0.0140], device='cuda:0'), out_proj_covar=tensor([6.8192e-05, 7.0693e-05, 7.9883e-05, 1.1362e-04, 7.0958e-05, 1.0955e-04, 6.6761e-05, 1.1214e-04], device='cuda:0') 2023-03-07 23:18:00,607 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29358.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:18:47,770 INFO [train2.py:809] (0/4) Epoch 8, batch 1500, loss[ctc_loss=0.1322, att_loss=0.2645, loss=0.238, over 16998.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.009818, over 51.00 utterances.], tot_loss[ctc_loss=0.1247, att_loss=0.2615, loss=0.2341, over 3280120.24 frames. utt_duration=1256 frames, utt_pad_proportion=0.04954, over 10457.32 utterances.], batch size: 51, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:19:23,502 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7652, 3.6910, 3.0344, 3.2548, 3.8520, 3.4626, 2.7935, 4.2724], device='cuda:0'), covar=tensor([0.1028, 0.0477, 0.1022, 0.0693, 0.0624, 0.0708, 0.0894, 0.0406], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0160, 0.0193, 0.0164, 0.0203, 0.0194, 0.0167, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 23:19:30,617 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.613e+02 2.592e+02 3.615e+02 4.521e+02 1.886e+03, threshold=7.229e+02, percent-clipped=4.0 2023-03-07 23:20:07,837 INFO [train2.py:809] (0/4) Epoch 8, batch 1550, loss[ctc_loss=0.199, att_loss=0.3046, loss=0.2835, over 17056.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008795, over 52.00 utterances.], tot_loss[ctc_loss=0.1258, att_loss=0.2622, loss=0.2349, over 3284401.98 frames. utt_duration=1251 frames, utt_pad_proportion=0.04948, over 10517.09 utterances.], batch size: 52, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:21:28,329 INFO [train2.py:809] (0/4) Epoch 8, batch 1600, loss[ctc_loss=0.09342, att_loss=0.2204, loss=0.195, over 14504.00 frames. utt_duration=1815 frames, utt_pad_proportion=0.03272, over 32.00 utterances.], tot_loss[ctc_loss=0.1259, att_loss=0.2618, loss=0.2346, over 3264227.96 frames. utt_duration=1230 frames, utt_pad_proportion=0.06092, over 10629.16 utterances.], batch size: 32, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:21:48,633 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9331, 5.3191, 4.7553, 5.3276, 4.7502, 5.0372, 5.4174, 5.2062], device='cuda:0'), covar=tensor([0.0432, 0.0205, 0.0751, 0.0148, 0.0412, 0.0180, 0.0161, 0.0152], device='cuda:0'), in_proj_covar=tensor([0.0285, 0.0216, 0.0280, 0.0204, 0.0229, 0.0178, 0.0203, 0.0199], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 23:22:09,646 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.44 vs. limit=5.0 2023-03-07 23:22:11,634 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.666e+02 2.839e+02 3.566e+02 4.248e+02 8.470e+02, threshold=7.133e+02, percent-clipped=2.0 2023-03-07 23:22:15,559 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29517.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:22:31,307 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.07 vs. limit=5.0 2023-03-07 23:22:48,442 INFO [train2.py:809] (0/4) Epoch 8, batch 1650, loss[ctc_loss=0.1484, att_loss=0.2889, loss=0.2608, over 17316.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02215, over 59.00 utterances.], tot_loss[ctc_loss=0.1262, att_loss=0.2616, loss=0.2345, over 3262515.05 frames. utt_duration=1219 frames, utt_pad_proportion=0.06467, over 10716.55 utterances.], batch size: 59, lr: 1.37e-02, grad_scale: 8.0 2023-03-07 23:23:10,606 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29551.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:23:32,755 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29565.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:24:08,550 INFO [train2.py:809] (0/4) Epoch 8, batch 1700, loss[ctc_loss=0.1322, att_loss=0.2665, loss=0.2396, over 17466.00 frames. utt_duration=1110 frames, utt_pad_proportion=0.02848, over 63.00 utterances.], tot_loss[ctc_loss=0.1269, att_loss=0.2624, loss=0.2353, over 3267026.82 frames. utt_duration=1186 frames, utt_pad_proportion=0.07243, over 11034.78 utterances.], batch size: 63, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:24:11,521 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-07 23:24:16,260 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29592.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:24:47,182 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29612.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:24:49,218 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9779, 5.2779, 5.2003, 5.1907, 5.2801, 5.2195, 4.9607, 4.7149], device='cuda:0'), covar=tensor([0.0995, 0.0422, 0.0240, 0.0438, 0.0268, 0.0318, 0.0326, 0.0359], device='cuda:0'), in_proj_covar=tensor([0.0420, 0.0260, 0.0204, 0.0242, 0.0300, 0.0327, 0.0251, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:24:50,955 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.1204, 3.6890, 3.5003, 2.8646, 3.4839, 3.5824, 3.5541, 1.9400], device='cuda:0'), covar=tensor([0.1204, 0.1050, 0.1867, 0.5866, 0.2271, 0.3403, 0.0727, 0.9640], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0086, 0.0093, 0.0146, 0.0082, 0.0135, 0.0079, 0.0141], device='cuda:0'), out_proj_covar=tensor([6.9080e-05, 6.9868e-05, 7.9701e-05, 1.1378e-04, 7.0784e-05, 1.0798e-04, 6.5065e-05, 1.1247e-04], device='cuda:0') 2023-03-07 23:24:52,831 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.894e+02 2.662e+02 3.188e+02 4.272e+02 1.218e+03, threshold=6.376e+02, percent-clipped=1.0 2023-03-07 23:24:55,358 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1752, 4.8021, 4.5210, 4.4268, 2.1247, 4.6046, 2.3121, 1.3983], device='cuda:0'), covar=tensor([0.0280, 0.0102, 0.0668, 0.0202, 0.2463, 0.0127, 0.1907, 0.2172], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0097, 0.0249, 0.0112, 0.0219, 0.0101, 0.0225, 0.0199], device='cuda:0'), out_proj_covar=tensor([1.2238e-04, 1.0172e-04, 2.2715e-04, 1.0625e-04, 2.0719e-04, 9.8850e-05, 2.0477e-04, 1.8223e-04], device='cuda:0') 2023-03-07 23:25:28,578 INFO [train2.py:809] (0/4) Epoch 8, batch 1750, loss[ctc_loss=0.1091, att_loss=0.2345, loss=0.2094, over 14539.00 frames. utt_duration=1819 frames, utt_pad_proportion=0.04112, over 32.00 utterances.], tot_loss[ctc_loss=0.1262, att_loss=0.2623, loss=0.2351, over 3273464.17 frames. utt_duration=1207 frames, utt_pad_proportion=0.06457, over 10863.30 utterances.], batch size: 32, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:26:01,594 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29658.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:26:07,745 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29662.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:26:48,812 INFO [train2.py:809] (0/4) Epoch 8, batch 1800, loss[ctc_loss=0.1115, att_loss=0.2386, loss=0.2132, over 15631.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009231, over 37.00 utterances.], tot_loss[ctc_loss=0.1259, att_loss=0.2622, loss=0.235, over 3275320.64 frames. utt_duration=1213 frames, utt_pad_proportion=0.06274, over 10813.26 utterances.], batch size: 37, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:26:53,346 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-03-07 23:27:00,888 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-03-07 23:27:15,983 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9056, 2.9852, 3.5647, 4.5644, 4.3539, 4.2108, 3.1128, 2.3359], device='cuda:0'), covar=tensor([0.0524, 0.1931, 0.1040, 0.0567, 0.0541, 0.0341, 0.1365, 0.2315], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0192, 0.0180, 0.0174, 0.0164, 0.0131, 0.0184, 0.0175], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 23:27:19,005 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29706.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:27:33,857 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.769e+02 3.006e+02 3.671e+02 4.359e+02 7.069e+02, threshold=7.341e+02, percent-clipped=3.0 2023-03-07 23:27:47,120 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29723.0, num_to_drop=1, layers_to_drop={1} 2023-03-07 23:28:06,112 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3795, 5.2173, 5.2843, 3.1536, 5.1885, 4.7004, 4.5760, 2.8096], device='cuda:0'), covar=tensor([0.0076, 0.0083, 0.0126, 0.0934, 0.0062, 0.0139, 0.0270, 0.1284], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0074, 0.0060, 0.0098, 0.0064, 0.0082, 0.0085, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:28:08,901 INFO [train2.py:809] (0/4) Epoch 8, batch 1850, loss[ctc_loss=0.151, att_loss=0.2762, loss=0.2512, over 16771.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.00651, over 48.00 utterances.], tot_loss[ctc_loss=0.1257, att_loss=0.2623, loss=0.235, over 3278890.31 frames. utt_duration=1214 frames, utt_pad_proportion=0.06203, over 10820.02 utterances.], batch size: 48, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:28:23,187 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3733, 4.6778, 4.7798, 4.6753, 2.5208, 4.9992, 2.9975, 1.8171], device='cuda:0'), covar=tensor([0.0238, 0.0191, 0.0539, 0.0166, 0.1918, 0.0085, 0.1345, 0.1775], device='cuda:0'), in_proj_covar=tensor([0.0121, 0.0098, 0.0247, 0.0111, 0.0217, 0.0098, 0.0222, 0.0198], device='cuda:0'), out_proj_covar=tensor([1.2086e-04, 1.0171e-04, 2.2543e-04, 1.0557e-04, 2.0510e-04, 9.6236e-05, 2.0222e-04, 1.8105e-04], device='cuda:0') 2023-03-07 23:29:17,151 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3997, 1.4671, 1.5444, 1.7876, 2.0626, 1.7268, 1.5351, 2.0776], device='cuda:0'), covar=tensor([0.0841, 0.3584, 0.2565, 0.2018, 0.0999, 0.1875, 0.3410, 0.1553], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0084, 0.0087, 0.0077, 0.0074, 0.0071, 0.0083, 0.0065], device='cuda:0'), out_proj_covar=tensor([4.1133e-05, 5.2598e-05, 5.2404e-05, 4.4929e-05, 3.9798e-05, 4.4657e-05, 5.1521e-05, 4.2410e-05], device='cuda:0') 2023-03-07 23:29:29,188 INFO [train2.py:809] (0/4) Epoch 8, batch 1900, loss[ctc_loss=0.1153, att_loss=0.2519, loss=0.2246, over 16188.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.00627, over 41.00 utterances.], tot_loss[ctc_loss=0.1262, att_loss=0.2629, loss=0.2355, over 3280804.06 frames. utt_duration=1192 frames, utt_pad_proportion=0.06645, over 11025.61 utterances.], batch size: 41, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:29:53,389 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=29802.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:30:14,334 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.846e+02 2.848e+02 3.489e+02 4.183e+02 9.529e+02, threshold=6.977e+02, percent-clipped=3.0 2023-03-07 23:30:49,807 INFO [train2.py:809] (0/4) Epoch 8, batch 1950, loss[ctc_loss=0.1333, att_loss=0.2689, loss=0.2418, over 16546.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005807, over 45.00 utterances.], tot_loss[ctc_loss=0.1259, att_loss=0.2623, loss=0.235, over 3271460.00 frames. utt_duration=1210 frames, utt_pad_proportion=0.06466, over 10824.15 utterances.], batch size: 45, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:31:26,790 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9929, 4.9531, 4.8691, 2.9553, 4.7386, 4.4961, 4.2041, 2.2027], device='cuda:0'), covar=tensor([0.0099, 0.0082, 0.0137, 0.1014, 0.0092, 0.0160, 0.0354, 0.1648], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0074, 0.0061, 0.0100, 0.0065, 0.0083, 0.0086, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:31:32,150 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=29863.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:31:38,692 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9807, 5.3197, 5.1474, 5.2315, 5.4490, 5.3455, 5.1940, 4.7994], device='cuda:0'), covar=tensor([0.1048, 0.0466, 0.0248, 0.0457, 0.0253, 0.0280, 0.0241, 0.0296], device='cuda:0'), in_proj_covar=tensor([0.0435, 0.0268, 0.0211, 0.0249, 0.0315, 0.0336, 0.0254, 0.0289], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:32:10,085 INFO [train2.py:809] (0/4) Epoch 8, batch 2000, loss[ctc_loss=0.1111, att_loss=0.2402, loss=0.2143, over 15659.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.007959, over 37.00 utterances.], tot_loss[ctc_loss=0.1254, att_loss=0.2622, loss=0.2348, over 3274964.20 frames. utt_duration=1232 frames, utt_pad_proportion=0.05845, over 10649.59 utterances.], batch size: 37, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:32:18,040 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=29892.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:32:41,259 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=29907.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:32:54,605 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.764e+02 2.836e+02 3.233e+02 3.942e+02 8.276e+02, threshold=6.465e+02, percent-clipped=1.0 2023-03-07 23:33:30,282 INFO [train2.py:809] (0/4) Epoch 8, batch 2050, loss[ctc_loss=0.1169, att_loss=0.2502, loss=0.2235, over 16410.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006381, over 44.00 utterances.], tot_loss[ctc_loss=0.1241, att_loss=0.2613, loss=0.2339, over 3282974.71 frames. utt_duration=1250 frames, utt_pad_proportion=0.05185, over 10519.18 utterances.], batch size: 44, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:33:34,836 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=29940.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:33:40,666 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-03-07 23:33:51,321 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-03-07 23:34:29,552 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8578, 5.2710, 4.6261, 5.2898, 4.5800, 4.9368, 5.3842, 5.1723], device='cuda:0'), covar=tensor([0.0463, 0.0239, 0.0888, 0.0181, 0.0451, 0.0219, 0.0165, 0.0148], device='cuda:0'), in_proj_covar=tensor([0.0286, 0.0217, 0.0282, 0.0203, 0.0230, 0.0177, 0.0201, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-07 23:34:51,160 INFO [train2.py:809] (0/4) Epoch 8, batch 2100, loss[ctc_loss=0.1401, att_loss=0.2747, loss=0.2478, over 17322.00 frames. utt_duration=878.7 frames, utt_pad_proportion=0.07892, over 79.00 utterances.], tot_loss[ctc_loss=0.124, att_loss=0.2614, loss=0.2339, over 3281997.89 frames. utt_duration=1234 frames, utt_pad_proportion=0.05641, over 10651.09 utterances.], batch size: 79, lr: 1.36e-02, grad_scale: 8.0 2023-03-07 23:35:12,213 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-30000.pt 2023-03-07 23:35:43,064 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.761e+02 2.671e+02 3.420e+02 4.365e+02 1.034e+03, threshold=6.841e+02, percent-clipped=5.0 2023-03-07 23:35:46,378 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30018.0, num_to_drop=1, layers_to_drop={2} 2023-03-07 23:36:16,101 INFO [train2.py:809] (0/4) Epoch 8, batch 2150, loss[ctc_loss=0.1082, att_loss=0.2385, loss=0.2125, over 16020.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007183, over 40.00 utterances.], tot_loss[ctc_loss=0.124, att_loss=0.2612, loss=0.2338, over 3284315.25 frames. utt_duration=1246 frames, utt_pad_proportion=0.0528, over 10552.49 utterances.], batch size: 40, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:37:36,070 INFO [train2.py:809] (0/4) Epoch 8, batch 2200, loss[ctc_loss=0.1267, att_loss=0.2473, loss=0.2232, over 16179.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.005787, over 41.00 utterances.], tot_loss[ctc_loss=0.1238, att_loss=0.2616, loss=0.2341, over 3290167.19 frames. utt_duration=1259 frames, utt_pad_proportion=0.04768, over 10462.95 utterances.], batch size: 41, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:38:22,431 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.913e+02 2.675e+02 3.232e+02 4.079e+02 9.910e+02, threshold=6.464e+02, percent-clipped=1.0 2023-03-07 23:38:37,595 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30125.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:38:55,632 INFO [train2.py:809] (0/4) Epoch 8, batch 2250, loss[ctc_loss=0.1017, att_loss=0.2364, loss=0.2095, over 16389.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.007712, over 44.00 utterances.], tot_loss[ctc_loss=0.123, att_loss=0.2606, loss=0.2331, over 3281924.99 frames. utt_duration=1288 frames, utt_pad_proportion=0.04379, over 10206.43 utterances.], batch size: 44, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:39:28,571 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30158.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:39:50,127 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1069, 5.0824, 4.9564, 2.5832, 1.8412, 2.4776, 3.6810, 3.7056], device='cuda:0'), covar=tensor([0.0592, 0.0146, 0.0208, 0.2960, 0.6208, 0.2928, 0.1080, 0.1875], device='cuda:0'), in_proj_covar=tensor([0.0310, 0.0193, 0.0219, 0.0183, 0.0344, 0.0329, 0.0209, 0.0334], device='cuda:0'), out_proj_covar=tensor([1.4821e-04, 7.6182e-05, 9.7187e-05, 8.4840e-05, 1.5632e-04, 1.3980e-04, 8.4448e-05, 1.4968e-04], device='cuda:0') 2023-03-07 23:40:13,306 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30186.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:40:14,496 INFO [train2.py:809] (0/4) Epoch 8, batch 2300, loss[ctc_loss=0.1224, att_loss=0.2317, loss=0.2098, over 15781.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.007961, over 38.00 utterances.], tot_loss[ctc_loss=0.1233, att_loss=0.2605, loss=0.233, over 3276630.11 frames. utt_duration=1312 frames, utt_pad_proportion=0.03855, over 9999.02 utterances.], batch size: 38, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:40:31,780 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7111, 3.6985, 2.8457, 3.4615, 3.8244, 3.4592, 2.8593, 4.2767], device='cuda:0'), covar=tensor([0.1111, 0.0503, 0.1357, 0.0629, 0.0645, 0.0704, 0.0922, 0.0412], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0164, 0.0195, 0.0165, 0.0208, 0.0194, 0.0169, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:40:32,680 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-03-07 23:40:46,669 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30207.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:40:50,442 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30209.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:41:00,842 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.528e+02 2.649e+02 3.337e+02 4.141e+02 1.009e+03, threshold=6.675e+02, percent-clipped=1.0 2023-03-07 23:41:34,430 INFO [train2.py:809] (0/4) Epoch 8, batch 2350, loss[ctc_loss=0.1161, att_loss=0.2384, loss=0.2139, over 15746.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.01006, over 38.00 utterances.], tot_loss[ctc_loss=0.1235, att_loss=0.2602, loss=0.2328, over 3264621.18 frames. utt_duration=1304 frames, utt_pad_proportion=0.04294, over 10024.21 utterances.], batch size: 38, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:41:39,213 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7881, 6.0590, 5.4124, 5.9391, 5.7209, 5.3223, 5.4336, 5.3099], device='cuda:0'), covar=tensor([0.1347, 0.0905, 0.0889, 0.0685, 0.0780, 0.1242, 0.2463, 0.2131], device='cuda:0'), in_proj_covar=tensor([0.0398, 0.0461, 0.0353, 0.0358, 0.0329, 0.0403, 0.0487, 0.0431], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-07 23:41:58,243 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2023-03-07 23:42:02,808 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30255.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:42:04,730 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8945, 3.8251, 3.1745, 3.4717, 3.9813, 3.5397, 2.7732, 4.3348], device='cuda:0'), covar=tensor([0.0942, 0.0342, 0.0878, 0.0547, 0.0461, 0.0570, 0.0864, 0.0406], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0164, 0.0191, 0.0163, 0.0206, 0.0192, 0.0167, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:42:28,164 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30270.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:42:42,128 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6911, 2.9075, 3.7897, 3.1158, 3.5281, 4.7421, 4.5585, 3.6671], device='cuda:0'), covar=tensor([0.0370, 0.1801, 0.1078, 0.1410, 0.1041, 0.0594, 0.0477, 0.1121], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0218, 0.0228, 0.0196, 0.0225, 0.0261, 0.0196, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-07 23:42:54,197 INFO [train2.py:809] (0/4) Epoch 8, batch 2400, loss[ctc_loss=0.1475, att_loss=0.2785, loss=0.2523, over 17383.00 frames. utt_duration=1105 frames, utt_pad_proportion=0.03381, over 63.00 utterances.], tot_loss[ctc_loss=0.123, att_loss=0.26, loss=0.2326, over 3265796.54 frames. utt_duration=1304 frames, utt_pad_proportion=0.04242, over 10032.51 utterances.], batch size: 63, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:43:00,492 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4559, 2.5241, 3.1384, 4.5010, 4.1776, 4.0807, 2.8917, 1.9727], device='cuda:0'), covar=tensor([0.0646, 0.2467, 0.1415, 0.0439, 0.0625, 0.0305, 0.1607, 0.2664], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0193, 0.0183, 0.0174, 0.0167, 0.0136, 0.0186, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 23:43:12,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-07 23:43:14,259 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2552, 2.6307, 3.1405, 4.3918, 3.9876, 3.9252, 2.6977, 1.8057], device='cuda:0'), covar=tensor([0.0707, 0.2007, 0.1140, 0.0478, 0.0632, 0.0338, 0.1746, 0.2672], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0194, 0.0183, 0.0175, 0.0167, 0.0136, 0.0187, 0.0176], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-07 23:43:40,599 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.713e+02 2.833e+02 3.441e+02 4.475e+02 1.318e+03, threshold=6.883e+02, percent-clipped=7.0 2023-03-07 23:43:44,653 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30318.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 23:44:13,408 INFO [train2.py:809] (0/4) Epoch 8, batch 2450, loss[ctc_loss=0.1375, att_loss=0.268, loss=0.2419, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007508, over 43.00 utterances.], tot_loss[ctc_loss=0.1217, att_loss=0.2592, loss=0.2317, over 3267263.88 frames. utt_duration=1310 frames, utt_pad_proportion=0.03968, over 9991.04 utterances.], batch size: 43, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:44:50,938 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30360.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:45:00,766 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30366.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:45:34,116 INFO [train2.py:809] (0/4) Epoch 8, batch 2500, loss[ctc_loss=0.1433, att_loss=0.2798, loss=0.2525, over 17444.00 frames. utt_duration=1013 frames, utt_pad_proportion=0.04457, over 69.00 utterances.], tot_loss[ctc_loss=0.1219, att_loss=0.2592, loss=0.2318, over 3265186.03 frames. utt_duration=1291 frames, utt_pad_proportion=0.04431, over 10124.67 utterances.], batch size: 69, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:45:35,906 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9811, 5.3576, 5.2056, 5.1787, 5.4273, 5.4034, 5.1231, 4.8875], device='cuda:0'), covar=tensor([0.0985, 0.0370, 0.0254, 0.0493, 0.0222, 0.0238, 0.0245, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0430, 0.0262, 0.0210, 0.0251, 0.0308, 0.0331, 0.0250, 0.0284], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:46:21,433 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.959e+02 2.650e+02 3.302e+02 4.215e+02 1.043e+03, threshold=6.603e+02, percent-clipped=3.0 2023-03-07 23:46:30,133 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30421.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:46:41,658 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-07 23:46:54,579 INFO [train2.py:809] (0/4) Epoch 8, batch 2550, loss[ctc_loss=0.08814, att_loss=0.2228, loss=0.1959, over 15627.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009811, over 37.00 utterances.], tot_loss[ctc_loss=0.1219, att_loss=0.2598, loss=0.2322, over 3268691.44 frames. utt_duration=1294 frames, utt_pad_proportion=0.04452, over 10113.26 utterances.], batch size: 37, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:47:10,454 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7298, 3.8812, 3.6316, 3.2464, 3.7253, 3.8967, 3.7752, 2.5011], device='cuda:0'), covar=tensor([0.1022, 0.1131, 0.4545, 0.5788, 0.1155, 0.6299, 0.0720, 0.9224], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0091, 0.0096, 0.0153, 0.0087, 0.0141, 0.0084, 0.0148], device='cuda:0'), out_proj_covar=tensor([7.3478e-05, 7.4766e-05, 8.3139e-05, 1.2048e-04, 7.5626e-05, 1.1345e-04, 6.9537e-05, 1.1802e-04], device='cuda:0') 2023-03-07 23:47:28,394 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30458.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:48:04,552 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30481.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:48:04,577 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0750, 4.9107, 4.9695, 4.9863, 4.7952, 5.0148, 4.9014, 4.5145], device='cuda:0'), covar=tensor([0.2116, 0.0852, 0.0333, 0.0599, 0.0916, 0.0456, 0.0318, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0433, 0.0265, 0.0210, 0.0251, 0.0311, 0.0335, 0.0251, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:48:13,626 INFO [train2.py:809] (0/4) Epoch 8, batch 2600, loss[ctc_loss=0.1032, att_loss=0.2499, loss=0.2205, over 17496.00 frames. utt_duration=887.4 frames, utt_pad_proportion=0.07082, over 79.00 utterances.], tot_loss[ctc_loss=0.1213, att_loss=0.2596, loss=0.2319, over 3266657.45 frames. utt_duration=1307 frames, utt_pad_proportion=0.04198, over 10006.60 utterances.], batch size: 79, lr: 1.35e-02, grad_scale: 8.0 2023-03-07 23:48:44,080 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30506.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:48:58,005 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6986, 4.8752, 5.1928, 5.2155, 4.9975, 5.6977, 4.9988, 5.7675], device='cuda:0'), covar=tensor([0.0737, 0.0687, 0.0722, 0.0997, 0.2132, 0.0727, 0.0746, 0.0574], device='cuda:0'), in_proj_covar=tensor([0.0627, 0.0375, 0.0433, 0.0495, 0.0673, 0.0439, 0.0351, 0.0428], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 23:49:00,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.127e+02 2.736e+02 3.358e+02 3.851e+02 8.673e+02, threshold=6.715e+02, percent-clipped=1.0 2023-03-07 23:49:18,396 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9717, 5.0972, 5.4377, 5.3961, 5.3090, 5.9280, 5.1127, 6.0307], device='cuda:0'), covar=tensor([0.0624, 0.0623, 0.0675, 0.0982, 0.1789, 0.0792, 0.0587, 0.0469], device='cuda:0'), in_proj_covar=tensor([0.0626, 0.0374, 0.0431, 0.0495, 0.0671, 0.0440, 0.0351, 0.0427], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-07 23:49:33,582 INFO [train2.py:809] (0/4) Epoch 8, batch 2650, loss[ctc_loss=0.1334, att_loss=0.2577, loss=0.2328, over 15945.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007617, over 41.00 utterances.], tot_loss[ctc_loss=0.121, att_loss=0.2591, loss=0.2314, over 3267990.21 frames. utt_duration=1298 frames, utt_pad_proportion=0.04444, over 10084.71 utterances.], batch size: 41, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:50:18,930 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30565.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:50:52,903 INFO [train2.py:809] (0/4) Epoch 8, batch 2700, loss[ctc_loss=0.1481, att_loss=0.286, loss=0.2585, over 17401.00 frames. utt_duration=1181 frames, utt_pad_proportion=0.01881, over 59.00 utterances.], tot_loss[ctc_loss=0.1215, att_loss=0.2594, loss=0.2318, over 3265569.41 frames. utt_duration=1283 frames, utt_pad_proportion=0.04938, over 10190.62 utterances.], batch size: 59, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:51:39,696 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.853e+02 2.974e+02 3.669e+02 4.665e+02 1.299e+03, threshold=7.338e+02, percent-clipped=5.0 2023-03-07 23:52:12,702 INFO [train2.py:809] (0/4) Epoch 8, batch 2750, loss[ctc_loss=0.1024, att_loss=0.2388, loss=0.2115, over 14571.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.03243, over 32.00 utterances.], tot_loss[ctc_loss=0.1211, att_loss=0.259, loss=0.2314, over 3267792.15 frames. utt_duration=1290 frames, utt_pad_proportion=0.04725, over 10147.48 utterances.], batch size: 32, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:52:42,251 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30655.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:53:33,179 INFO [train2.py:809] (0/4) Epoch 8, batch 2800, loss[ctc_loss=0.1273, att_loss=0.2765, loss=0.2467, over 17534.00 frames. utt_duration=1018 frames, utt_pad_proportion=0.03968, over 69.00 utterances.], tot_loss[ctc_loss=0.1209, att_loss=0.2586, loss=0.231, over 3268268.11 frames. utt_duration=1290 frames, utt_pad_proportion=0.04817, over 10144.94 utterances.], batch size: 69, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:54:19,315 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.00 vs. limit=5.0 2023-03-07 23:54:21,208 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.747e+02 2.729e+02 3.286e+02 4.018e+02 8.527e+02, threshold=6.573e+02, percent-clipped=3.0 2023-03-07 23:54:21,438 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=30716.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:54:21,627 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30716.0, num_to_drop=1, layers_to_drop={0} 2023-03-07 23:54:53,844 INFO [train2.py:809] (0/4) Epoch 8, batch 2850, loss[ctc_loss=0.1219, att_loss=0.2362, loss=0.2134, over 15763.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009028, over 38.00 utterances.], tot_loss[ctc_loss=0.1212, att_loss=0.2585, loss=0.231, over 3270101.74 frames. utt_duration=1268 frames, utt_pad_proportion=0.05087, over 10327.55 utterances.], batch size: 38, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:56:04,074 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30781.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:56:13,169 INFO [train2.py:809] (0/4) Epoch 8, batch 2900, loss[ctc_loss=0.1145, att_loss=0.2506, loss=0.2234, over 16158.00 frames. utt_duration=1578 frames, utt_pad_proportion=0.007624, over 41.00 utterances.], tot_loss[ctc_loss=0.1198, att_loss=0.2578, loss=0.2302, over 3261366.18 frames. utt_duration=1279 frames, utt_pad_proportion=0.04947, over 10213.92 utterances.], batch size: 41, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:56:24,162 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6920, 2.4893, 5.0209, 3.8080, 3.0117, 4.4402, 4.7403, 4.6718], device='cuda:0'), covar=tensor([0.0160, 0.1739, 0.0103, 0.1117, 0.1885, 0.0230, 0.0107, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0242, 0.0124, 0.0305, 0.0284, 0.0182, 0.0110, 0.0143], device='cuda:0'), out_proj_covar=tensor([1.3758e-04, 2.0268e-04, 1.1066e-04, 2.5164e-04, 2.5115e-04, 1.6212e-04, 9.9556e-05, 1.3446e-04], device='cuda:0') 2023-03-07 23:56:42,954 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-03-07 23:56:56,971 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.2116, 3.7653, 3.0048, 3.2097, 3.8882, 3.5387, 2.3964, 4.0849], device='cuda:0'), covar=tensor([0.1337, 0.0396, 0.0895, 0.0697, 0.0509, 0.0550, 0.1162, 0.0476], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0168, 0.0194, 0.0165, 0.0209, 0.0195, 0.0171, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-07 23:56:59,560 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.721e+02 2.826e+02 3.398e+02 4.214e+02 7.798e+02, threshold=6.796e+02, percent-clipped=3.0 2023-03-07 23:57:09,852 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1274, 4.6741, 4.6146, 4.6998, 2.2285, 4.7818, 2.6053, 2.0460], device='cuda:0'), covar=tensor([0.0301, 0.0142, 0.0721, 0.0181, 0.2471, 0.0130, 0.1741, 0.1791], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0100, 0.0253, 0.0109, 0.0223, 0.0099, 0.0226, 0.0199], device='cuda:0'), out_proj_covar=tensor([1.2452e-04, 1.0439e-04, 2.3153e-04, 1.0478e-04, 2.1096e-04, 9.8613e-05, 2.0677e-04, 1.8276e-04], device='cuda:0') 2023-03-07 23:57:20,110 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30829.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:57:32,268 INFO [train2.py:809] (0/4) Epoch 8, batch 2950, loss[ctc_loss=0.1223, att_loss=0.2712, loss=0.2414, over 16470.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006267, over 46.00 utterances.], tot_loss[ctc_loss=0.1204, att_loss=0.2578, loss=0.2303, over 3259632.92 frames. utt_duration=1290 frames, utt_pad_proportion=0.04698, over 10118.79 utterances.], batch size: 46, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:57:57,935 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-07 23:58:17,822 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=30865.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:58:38,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7225, 2.4791, 5.1150, 4.0395, 3.0794, 4.6776, 5.0251, 4.8223], device='cuda:0'), covar=tensor([0.0215, 0.1782, 0.0180, 0.1092, 0.2022, 0.0219, 0.0094, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0147, 0.0240, 0.0124, 0.0300, 0.0281, 0.0181, 0.0109, 0.0142], device='cuda:0'), out_proj_covar=tensor([1.3596e-04, 2.0095e-04, 1.0985e-04, 2.4763e-04, 2.4829e-04, 1.6057e-04, 9.8624e-05, 1.3327e-04], device='cuda:0') 2023-03-07 23:58:52,464 INFO [train2.py:809] (0/4) Epoch 8, batch 3000, loss[ctc_loss=0.1132, att_loss=0.2726, loss=0.2407, over 16880.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.006876, over 49.00 utterances.], tot_loss[ctc_loss=0.1218, att_loss=0.2595, loss=0.232, over 3270330.85 frames. utt_duration=1254 frames, utt_pad_proportion=0.05322, over 10442.37 utterances.], batch size: 49, lr: 1.34e-02, grad_scale: 8.0 2023-03-07 23:58:52,466 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-07 23:59:06,434 INFO [train2.py:843] (0/4) Epoch 8, validation: ctc_loss=0.05812, att_loss=0.2422, loss=0.2054, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-07 23:59:06,434 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16076MB 2023-03-07 23:59:32,869 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30903.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:59:48,119 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=30913.0, num_to_drop=0, layers_to_drop=set() 2023-03-07 23:59:53,217 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.005e+02 2.876e+02 3.540e+02 4.400e+02 7.249e+02, threshold=7.079e+02, percent-clipped=3.0 2023-03-08 00:00:07,386 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30925.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:00:21,864 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=30934.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:00:26,213 INFO [train2.py:809] (0/4) Epoch 8, batch 3050, loss[ctc_loss=0.154, att_loss=0.2869, loss=0.2603, over 17437.00 frames. utt_duration=884.3 frames, utt_pad_proportion=0.07407, over 79.00 utterances.], tot_loss[ctc_loss=0.1213, att_loss=0.2594, loss=0.2318, over 3273547.70 frames. utt_duration=1256 frames, utt_pad_proportion=0.05259, over 10441.75 utterances.], batch size: 79, lr: 1.34e-02, grad_scale: 8.0 2023-03-08 00:00:30,360 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.36 vs. limit=5.0 2023-03-08 00:00:50,337 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9379, 6.1460, 5.5747, 5.9205, 5.7535, 5.4724, 5.6491, 5.3942], device='cuda:0'), covar=tensor([0.1110, 0.0751, 0.0748, 0.0746, 0.0747, 0.1346, 0.1797, 0.2071], device='cuda:0'), in_proj_covar=tensor([0.0398, 0.0446, 0.0347, 0.0359, 0.0331, 0.0397, 0.0468, 0.0420], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 00:01:01,328 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-08 00:01:09,788 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30964.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:01:26,239 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.34 vs. limit=5.0 2023-03-08 00:01:44,859 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30986.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:01:46,066 INFO [train2.py:809] (0/4) Epoch 8, batch 3100, loss[ctc_loss=0.1825, att_loss=0.3017, loss=0.2778, over 17304.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02361, over 59.00 utterances.], tot_loss[ctc_loss=0.1218, att_loss=0.2599, loss=0.2323, over 3274135.28 frames. utt_duration=1235 frames, utt_pad_proportion=0.05768, over 10614.53 utterances.], batch size: 59, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:01:58,991 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=30995.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:02:24,714 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31011.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 00:02:24,852 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31011.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 00:02:32,585 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.786e+02 2.541e+02 3.093e+02 3.936e+02 7.180e+02, threshold=6.186e+02, percent-clipped=1.0 2023-03-08 00:02:32,935 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31016.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:02:54,580 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.10 vs. limit=2.0 2023-03-08 00:03:05,879 INFO [train2.py:809] (0/4) Epoch 8, batch 3150, loss[ctc_loss=0.08766, att_loss=0.2459, loss=0.2143, over 16338.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005654, over 45.00 utterances.], tot_loss[ctc_loss=0.1207, att_loss=0.2594, loss=0.2317, over 3274842.67 frames. utt_duration=1260 frames, utt_pad_proportion=0.0523, over 10411.46 utterances.], batch size: 45, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:03:49,190 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31064.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:04:01,734 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31072.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 00:04:13,002 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8058, 4.9730, 5.3578, 5.2736, 5.1244, 5.7367, 5.0089, 5.7946], device='cuda:0'), covar=tensor([0.0619, 0.0632, 0.0659, 0.0939, 0.1914, 0.0823, 0.0638, 0.0576], device='cuda:0'), in_proj_covar=tensor([0.0616, 0.0377, 0.0432, 0.0490, 0.0662, 0.0432, 0.0350, 0.0429], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-08 00:04:25,394 INFO [train2.py:809] (0/4) Epoch 8, batch 3200, loss[ctc_loss=0.1204, att_loss=0.2335, loss=0.2109, over 15392.00 frames. utt_duration=1761 frames, utt_pad_proportion=0.00977, over 35.00 utterances.], tot_loss[ctc_loss=0.1214, att_loss=0.2593, loss=0.2317, over 3269974.91 frames. utt_duration=1250 frames, utt_pad_proportion=0.05551, over 10474.89 utterances.], batch size: 35, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:04:59,743 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-08 00:05:11,435 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.012e+02 2.783e+02 3.320e+02 3.859e+02 7.988e+02, threshold=6.640e+02, percent-clipped=5.0 2023-03-08 00:05:12,558 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-08 00:05:21,977 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.00 vs. limit=5.0 2023-03-08 00:05:44,552 INFO [train2.py:809] (0/4) Epoch 8, batch 3250, loss[ctc_loss=0.1308, att_loss=0.2789, loss=0.2493, over 16879.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007067, over 49.00 utterances.], tot_loss[ctc_loss=0.122, att_loss=0.2601, loss=0.2325, over 3277796.72 frames. utt_duration=1238 frames, utt_pad_proportion=0.05631, over 10601.65 utterances.], batch size: 49, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:06:26,809 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6572, 1.8488, 4.8909, 4.0956, 3.0588, 4.4178, 4.4976, 4.5919], device='cuda:0'), covar=tensor([0.0102, 0.1814, 0.0102, 0.0767, 0.1805, 0.0180, 0.0109, 0.0151], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0241, 0.0126, 0.0303, 0.0281, 0.0183, 0.0108, 0.0145], device='cuda:0'), out_proj_covar=tensor([1.3552e-04, 2.0182e-04, 1.1189e-04, 2.4990e-04, 2.4883e-04, 1.6227e-04, 9.8637e-05, 1.3594e-04], device='cuda:0') 2023-03-08 00:07:03,806 INFO [train2.py:809] (0/4) Epoch 8, batch 3300, loss[ctc_loss=0.1182, att_loss=0.2507, loss=0.2242, over 16550.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005552, over 45.00 utterances.], tot_loss[ctc_loss=0.1214, att_loss=0.2598, loss=0.2321, over 3283310.69 frames. utt_duration=1254 frames, utt_pad_proportion=0.05009, over 10487.36 utterances.], batch size: 45, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:07:40,363 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6295, 2.2963, 4.9396, 3.8150, 3.1708, 4.3566, 4.7498, 4.5303], device='cuda:0'), covar=tensor([0.0177, 0.1777, 0.0119, 0.1176, 0.1805, 0.0226, 0.0107, 0.0201], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0239, 0.0125, 0.0300, 0.0278, 0.0181, 0.0109, 0.0143], device='cuda:0'), out_proj_covar=tensor([1.3391e-04, 2.0015e-04, 1.1140e-04, 2.4763e-04, 2.4625e-04, 1.6041e-04, 9.8765e-05, 1.3356e-04], device='cuda:0') 2023-03-08 00:07:51,301 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.702e+02 2.610e+02 3.147e+02 4.010e+02 7.285e+02, threshold=6.294e+02, percent-clipped=3.0 2023-03-08 00:08:01,701 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-03-08 00:08:24,606 INFO [train2.py:809] (0/4) Epoch 8, batch 3350, loss[ctc_loss=0.1156, att_loss=0.2655, loss=0.2355, over 17337.00 frames. utt_duration=1102 frames, utt_pad_proportion=0.03645, over 63.00 utterances.], tot_loss[ctc_loss=0.1202, att_loss=0.2592, loss=0.2314, over 3279450.34 frames. utt_duration=1265 frames, utt_pad_proportion=0.04772, over 10382.12 utterances.], batch size: 63, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:08:27,326 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-08 00:08:41,725 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([1.9262, 3.4358, 3.4175, 2.8381, 3.3959, 3.3990, 3.2878, 2.2093], device='cuda:0'), covar=tensor([0.1190, 0.1374, 0.1846, 0.5631, 0.1624, 0.5570, 0.0827, 0.8586], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0092, 0.0099, 0.0159, 0.0086, 0.0145, 0.0083, 0.0150], device='cuda:0'), out_proj_covar=tensor([7.6330e-05, 7.6300e-05, 8.6507e-05, 1.2557e-04, 7.5856e-05, 1.1644e-04, 7.0164e-05, 1.2055e-04], device='cuda:0') 2023-03-08 00:09:00,686 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31259.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:09:35,237 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31281.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:09:44,164 INFO [train2.py:809] (0/4) Epoch 8, batch 3400, loss[ctc_loss=0.1224, att_loss=0.2782, loss=0.2471, over 17047.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008497, over 52.00 utterances.], tot_loss[ctc_loss=0.1212, att_loss=0.2593, loss=0.2317, over 3272162.78 frames. utt_duration=1294 frames, utt_pad_proportion=0.04183, over 10124.38 utterances.], batch size: 52, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:09:49,566 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31290.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:10:23,008 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31311.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:10:30,543 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.874e+02 2.871e+02 3.632e+02 4.520e+02 1.169e+03, threshold=7.263e+02, percent-clipped=5.0 2023-03-08 00:11:04,305 INFO [train2.py:809] (0/4) Epoch 8, batch 3450, loss[ctc_loss=0.1157, att_loss=0.2517, loss=0.2245, over 16016.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007337, over 40.00 utterances.], tot_loss[ctc_loss=0.1212, att_loss=0.2595, loss=0.2318, over 3271861.85 frames. utt_duration=1300 frames, utt_pad_proportion=0.03921, over 10075.20 utterances.], batch size: 40, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:11:34,335 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-03-08 00:11:40,440 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31359.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:11:52,615 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31367.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 00:12:25,031 INFO [train2.py:809] (0/4) Epoch 8, batch 3500, loss[ctc_loss=0.08201, att_loss=0.2161, loss=0.1893, over 15613.00 frames. utt_duration=1689 frames, utt_pad_proportion=0.01093, over 37.00 utterances.], tot_loss[ctc_loss=0.1217, att_loss=0.2599, loss=0.2322, over 3272494.34 frames. utt_duration=1277 frames, utt_pad_proportion=0.04401, over 10262.52 utterances.], batch size: 37, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:13:02,192 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31409.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:13:12,728 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.860e+02 2.623e+02 3.276e+02 4.275e+02 9.794e+02, threshold=6.553e+02, percent-clipped=4.0 2023-03-08 00:13:47,408 INFO [train2.py:809] (0/4) Epoch 8, batch 3550, loss[ctc_loss=0.1208, att_loss=0.2687, loss=0.2391, over 16810.00 frames. utt_duration=680.8 frames, utt_pad_proportion=0.1469, over 99.00 utterances.], tot_loss[ctc_loss=0.1211, att_loss=0.2598, loss=0.232, over 3271936.25 frames. utt_duration=1265 frames, utt_pad_proportion=0.0464, over 10356.98 utterances.], batch size: 99, lr: 1.33e-02, grad_scale: 8.0 2023-03-08 00:13:54,011 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-03-08 00:14:33,223 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.99 vs. limit=5.0 2023-03-08 00:14:40,153 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31470.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:15:08,415 INFO [train2.py:809] (0/4) Epoch 8, batch 3600, loss[ctc_loss=0.121, att_loss=0.2659, loss=0.237, over 16910.00 frames. utt_duration=684.8 frames, utt_pad_proportion=0.1397, over 99.00 utterances.], tot_loss[ctc_loss=0.1222, att_loss=0.2607, loss=0.233, over 3269596.22 frames. utt_duration=1229 frames, utt_pad_proportion=0.05842, over 10658.76 utterances.], batch size: 99, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:15:15,033 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31491.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 00:15:34,274 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6527, 5.9535, 5.3936, 5.7665, 5.5547, 5.2318, 5.3544, 5.2166], device='cuda:0'), covar=tensor([0.1285, 0.0910, 0.0772, 0.0773, 0.0793, 0.1523, 0.2249, 0.2249], device='cuda:0'), in_proj_covar=tensor([0.0396, 0.0447, 0.0344, 0.0357, 0.0328, 0.0397, 0.0470, 0.0423], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 00:15:54,465 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.609e+02 2.998e+02 4.010e+02 6.869e+02, threshold=5.995e+02, percent-clipped=2.0 2023-03-08 00:15:54,782 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4311, 4.9071, 4.6790, 4.8129, 4.9425, 4.6832, 3.5469, 4.7819], device='cuda:0'), covar=tensor([0.0118, 0.0096, 0.0125, 0.0083, 0.0091, 0.0097, 0.0582, 0.0177], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0061, 0.0073, 0.0046, 0.0050, 0.0060, 0.0082, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:16:29,576 INFO [train2.py:809] (0/4) Epoch 8, batch 3650, loss[ctc_loss=0.0917, att_loss=0.2304, loss=0.2027, over 15362.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.008992, over 35.00 utterances.], tot_loss[ctc_loss=0.1212, att_loss=0.26, loss=0.2322, over 3269654.58 frames. utt_duration=1239 frames, utt_pad_proportion=0.05558, over 10569.07 utterances.], batch size: 35, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:16:54,247 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31552.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 00:17:04,954 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31559.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:17:40,199 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31581.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:17:50,149 INFO [train2.py:809] (0/4) Epoch 8, batch 3700, loss[ctc_loss=0.1476, att_loss=0.2815, loss=0.2547, over 16984.00 frames. utt_duration=694.6 frames, utt_pad_proportion=0.1296, over 98.00 utterances.], tot_loss[ctc_loss=0.1212, att_loss=0.2599, loss=0.2322, over 3267345.65 frames. utt_duration=1232 frames, utt_pad_proportion=0.05756, over 10618.77 utterances.], batch size: 98, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:17:55,128 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31590.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:18:17,631 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9720, 6.1933, 5.5773, 6.0043, 5.8723, 5.3967, 5.6339, 5.4934], device='cuda:0'), covar=tensor([0.1169, 0.0813, 0.0746, 0.0716, 0.0796, 0.1456, 0.2202, 0.2077], device='cuda:0'), in_proj_covar=tensor([0.0400, 0.0449, 0.0351, 0.0359, 0.0332, 0.0398, 0.0472, 0.0424], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 00:18:22,255 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31607.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:18:35,885 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.785e+02 2.776e+02 3.358e+02 4.277e+02 8.903e+02, threshold=6.717e+02, percent-clipped=4.0 2023-03-08 00:18:57,076 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31629.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:19:01,072 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-03-08 00:19:10,111 INFO [train2.py:809] (0/4) Epoch 8, batch 3750, loss[ctc_loss=0.08828, att_loss=0.2248, loss=0.1975, over 15628.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.008068, over 37.00 utterances.], tot_loss[ctc_loss=0.1209, att_loss=0.2591, loss=0.2315, over 3261983.56 frames. utt_duration=1257 frames, utt_pad_proportion=0.05364, over 10390.09 utterances.], batch size: 37, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:19:11,662 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31638.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:19:57,353 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=31667.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 00:20:08,941 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0551, 4.8831, 4.7115, 4.8735, 5.1789, 4.8443, 4.5941, 2.1079], device='cuda:0'), covar=tensor([0.0126, 0.0140, 0.0170, 0.0105, 0.0962, 0.0145, 0.0214, 0.2494], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0124, 0.0127, 0.0131, 0.0314, 0.0126, 0.0114, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 00:20:29,568 INFO [train2.py:809] (0/4) Epoch 8, batch 3800, loss[ctc_loss=0.1045, att_loss=0.2338, loss=0.2079, over 13169.00 frames. utt_duration=1818 frames, utt_pad_proportion=0.09221, over 29.00 utterances.], tot_loss[ctc_loss=0.1213, att_loss=0.2596, loss=0.232, over 3267528.66 frames. utt_duration=1267 frames, utt_pad_proportion=0.05015, over 10326.78 utterances.], batch size: 29, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:20:56,747 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31704.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:21:13,679 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=31715.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 00:21:14,931 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.603e+02 2.740e+02 3.298e+02 4.235e+02 1.019e+03, threshold=6.595e+02, percent-clipped=5.0 2023-03-08 00:21:49,309 INFO [train2.py:809] (0/4) Epoch 8, batch 3850, loss[ctc_loss=0.1017, att_loss=0.2199, loss=0.1963, over 15528.00 frames. utt_duration=1727 frames, utt_pad_proportion=0.006709, over 36.00 utterances.], tot_loss[ctc_loss=0.121, att_loss=0.2589, loss=0.2313, over 3261910.72 frames. utt_duration=1246 frames, utt_pad_proportion=0.05531, over 10487.04 utterances.], batch size: 36, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:21:56,446 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3021, 5.1572, 5.1788, 3.0894, 4.9579, 4.6388, 4.4027, 2.5752], device='cuda:0'), covar=tensor([0.0078, 0.0082, 0.0134, 0.0918, 0.0086, 0.0158, 0.0274, 0.1322], device='cuda:0'), in_proj_covar=tensor([0.0055, 0.0074, 0.0064, 0.0099, 0.0064, 0.0085, 0.0086, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 00:22:32,472 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:22:32,678 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31765.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:22:58,795 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7575, 5.0171, 5.2685, 5.1991, 5.1265, 5.7072, 4.9531, 5.8690], device='cuda:0'), covar=tensor([0.0637, 0.0648, 0.0621, 0.0889, 0.1703, 0.0713, 0.0737, 0.0427], device='cuda:0'), in_proj_covar=tensor([0.0634, 0.0387, 0.0438, 0.0501, 0.0671, 0.0437, 0.0356, 0.0431], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:23:05,942 INFO [train2.py:809] (0/4) Epoch 8, batch 3900, loss[ctc_loss=0.1046, att_loss=0.236, loss=0.2097, over 15473.00 frames. utt_duration=1721 frames, utt_pad_proportion=0.01037, over 36.00 utterances.], tot_loss[ctc_loss=0.12, att_loss=0.2585, loss=0.2308, over 3268783.21 frames. utt_duration=1268 frames, utt_pad_proportion=0.04828, over 10326.05 utterances.], batch size: 36, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:23:07,745 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31788.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:23:24,496 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-03-08 00:23:50,995 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.828e+02 2.672e+02 3.345e+02 4.002e+02 8.198e+02, threshold=6.690e+02, percent-clipped=4.0 2023-03-08 00:24:01,898 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31823.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:24:22,884 INFO [train2.py:809] (0/4) Epoch 8, batch 3950, loss[ctc_loss=0.1235, att_loss=0.2752, loss=0.2449, over 17368.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03227, over 63.00 utterances.], tot_loss[ctc_loss=0.12, att_loss=0.2588, loss=0.2311, over 3262218.80 frames. utt_duration=1257 frames, utt_pad_proportion=0.05347, over 10396.91 utterances.], batch size: 63, lr: 1.32e-02, grad_scale: 8.0 2023-03-08 00:24:38,313 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=31847.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 00:24:41,462 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31849.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:25:13,413 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-8.pt 2023-03-08 00:25:39,121 INFO [train2.py:809] (0/4) Epoch 9, batch 0, loss[ctc_loss=0.1272, att_loss=0.2671, loss=0.2391, over 17301.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01117, over 55.00 utterances.], tot_loss[ctc_loss=0.1272, att_loss=0.2671, loss=0.2391, over 17301.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01117, over 55.00 utterances.], batch size: 55, lr: 1.25e-02, grad_scale: 8.0 2023-03-08 00:25:39,123 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 00:25:51,861 INFO [train2.py:843] (0/4) Epoch 9, validation: ctc_loss=0.05701, att_loss=0.2418, loss=0.2048, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 00:25:51,862 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16076MB 2023-03-08 00:26:04,748 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31878.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:26:14,977 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31884.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:27:04,456 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31915.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:27:05,620 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.740e+02 3.193e+02 4.001e+02 8.553e+02, threshold=6.386e+02, percent-clipped=3.0 2023-03-08 00:27:12,465 INFO [train2.py:809] (0/4) Epoch 9, batch 50, loss[ctc_loss=0.1601, att_loss=0.2949, loss=0.2679, over 17048.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008955, over 52.00 utterances.], tot_loss[ctc_loss=0.1216, att_loss=0.2634, loss=0.235, over 745255.85 frames. utt_duration=1265 frames, utt_pad_proportion=0.0399, over 2359.71 utterances.], batch size: 52, lr: 1.25e-02, grad_scale: 16.0 2023-03-08 00:27:13,545 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.10 vs. limit=2.0 2023-03-08 00:27:43,122 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31939.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:28:00,012 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=31950.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:28:31,524 INFO [train2.py:809] (0/4) Epoch 9, batch 100, loss[ctc_loss=0.1279, att_loss=0.2668, loss=0.239, over 17329.00 frames. utt_duration=1006 frames, utt_pad_proportion=0.04905, over 69.00 utterances.], tot_loss[ctc_loss=0.1197, att_loss=0.2605, loss=0.2324, over 1311707.79 frames. utt_duration=1274 frames, utt_pad_proportion=0.03871, over 4123.14 utterances.], batch size: 69, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:28:41,607 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=31976.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:29:19,197 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-32000.pt 2023-03-08 00:29:41,555 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=32011.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:29:48,727 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 3.075e+02 3.496e+02 4.231e+02 7.573e+02, threshold=6.993e+02, percent-clipped=4.0 2023-03-08 00:29:54,994 INFO [train2.py:809] (0/4) Epoch 9, batch 150, loss[ctc_loss=0.1113, att_loss=0.2673, loss=0.2361, over 17418.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.04606, over 69.00 utterances.], tot_loss[ctc_loss=0.1192, att_loss=0.2585, loss=0.2306, over 1744171.24 frames. utt_duration=1324 frames, utt_pad_proportion=0.03179, over 5276.93 utterances.], batch size: 69, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:30:59,436 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32060.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:31:07,684 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32065.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:31:15,195 INFO [train2.py:809] (0/4) Epoch 9, batch 200, loss[ctc_loss=0.09727, att_loss=0.2353, loss=0.2077, over 15388.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.009695, over 35.00 utterances.], tot_loss[ctc_loss=0.1195, att_loss=0.2591, loss=0.2312, over 2088170.18 frames. utt_duration=1263 frames, utt_pad_proportion=0.04526, over 6622.03 utterances.], batch size: 35, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:32:23,943 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32113.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:32:28,428 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.810e+02 2.737e+02 3.348e+02 3.979e+02 1.144e+03, threshold=6.695e+02, percent-clipped=2.0 2023-03-08 00:32:34,684 INFO [train2.py:809] (0/4) Epoch 9, batch 250, loss[ctc_loss=0.09734, att_loss=0.2503, loss=0.2197, over 17282.00 frames. utt_duration=1003 frames, utt_pad_proportion=0.05426, over 69.00 utterances.], tot_loss[ctc_loss=0.1183, att_loss=0.2585, loss=0.2305, over 2354130.37 frames. utt_duration=1284 frames, utt_pad_proportion=0.04146, over 7339.54 utterances.], batch size: 69, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:33:13,060 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32144.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:33:17,805 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32147.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 00:33:54,277 INFO [train2.py:809] (0/4) Epoch 9, batch 300, loss[ctc_loss=0.201, att_loss=0.3021, loss=0.2818, over 14013.00 frames. utt_duration=385.4 frames, utt_pad_proportion=0.3273, over 146.00 utterances.], tot_loss[ctc_loss=0.1198, att_loss=0.2596, loss=0.2316, over 2557407.18 frames. utt_duration=1255 frames, utt_pad_proportion=0.0507, over 8159.63 utterances.], batch size: 146, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:34:09,518 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32179.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:34:34,193 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32195.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 00:34:55,291 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5444, 5.1061, 4.8494, 5.1614, 5.1451, 4.7932, 3.5201, 4.9507], device='cuda:0'), covar=tensor([0.0106, 0.0098, 0.0088, 0.0073, 0.0068, 0.0095, 0.0617, 0.0224], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0063, 0.0073, 0.0047, 0.0049, 0.0060, 0.0084, 0.0082], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:35:08,279 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.524e+02 3.099e+02 4.059e+02 8.225e+02, threshold=6.197e+02, percent-clipped=2.0 2023-03-08 00:35:14,578 INFO [train2.py:809] (0/4) Epoch 9, batch 350, loss[ctc_loss=0.1241, att_loss=0.2724, loss=0.2427, over 17349.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03581, over 63.00 utterances.], tot_loss[ctc_loss=0.1187, att_loss=0.2582, loss=0.2303, over 2707212.11 frames. utt_duration=1247 frames, utt_pad_proportion=0.05656, over 8695.16 utterances.], batch size: 63, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:35:37,832 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32234.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:35:42,773 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3797, 3.7606, 3.6078, 3.0480, 3.5336, 3.7050, 3.5399, 2.4224], device='cuda:0'), covar=tensor([0.1555, 0.1039, 0.2704, 0.6353, 0.1340, 0.5157, 0.0949, 1.0135], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0096, 0.0102, 0.0167, 0.0088, 0.0152, 0.0087, 0.0153], device='cuda:0'), out_proj_covar=tensor([7.9445e-05, 8.0112e-05, 8.9655e-05, 1.3258e-04, 7.8253e-05, 1.2210e-04, 7.3862e-05, 1.2287e-04], device='cuda:0') 2023-03-08 00:36:35,197 INFO [train2.py:809] (0/4) Epoch 9, batch 400, loss[ctc_loss=0.07573, att_loss=0.2211, loss=0.192, over 15769.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009447, over 38.00 utterances.], tot_loss[ctc_loss=0.1184, att_loss=0.2582, loss=0.2303, over 2833568.64 frames. utt_duration=1223 frames, utt_pad_proportion=0.06219, over 9276.00 utterances.], batch size: 38, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:36:36,904 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32271.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:36:49,943 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8702, 6.1043, 5.5108, 5.9077, 5.8487, 5.4352, 5.5630, 5.3867], device='cuda:0'), covar=tensor([0.1111, 0.0828, 0.0761, 0.0669, 0.0665, 0.1288, 0.1827, 0.1918], device='cuda:0'), in_proj_covar=tensor([0.0390, 0.0446, 0.0341, 0.0353, 0.0327, 0.0386, 0.0461, 0.0418], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:37:32,181 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=32306.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:37:36,000 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6429, 3.9581, 3.8425, 4.0158, 3.9962, 3.7590, 2.9559, 3.8377], device='cuda:0'), covar=tensor([0.0129, 0.0107, 0.0117, 0.0076, 0.0079, 0.0115, 0.0570, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0064, 0.0074, 0.0048, 0.0050, 0.0061, 0.0084, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:37:48,587 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.723e+02 2.567e+02 3.123e+02 3.914e+02 5.946e+02, threshold=6.245e+02, percent-clipped=0.0 2023-03-08 00:37:54,621 INFO [train2.py:809] (0/4) Epoch 9, batch 450, loss[ctc_loss=0.1316, att_loss=0.2743, loss=0.2458, over 17284.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01234, over 55.00 utterances.], tot_loss[ctc_loss=0.1202, att_loss=0.2592, loss=0.2314, over 2930394.96 frames. utt_duration=1203 frames, utt_pad_proportion=0.06743, over 9752.05 utterances.], batch size: 55, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:38:07,534 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1735, 5.2009, 5.0285, 2.4930, 4.8979, 4.5842, 4.5186, 2.5737], device='cuda:0'), covar=tensor([0.0122, 0.0067, 0.0225, 0.1202, 0.0082, 0.0159, 0.0274, 0.1503], device='cuda:0'), in_proj_covar=tensor([0.0056, 0.0074, 0.0065, 0.0099, 0.0065, 0.0086, 0.0086, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 00:38:29,807 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8477, 4.6182, 4.7471, 4.7797, 5.2454, 4.9255, 4.5025, 2.3749], device='cuda:0'), covar=tensor([0.0208, 0.0258, 0.0208, 0.0199, 0.0789, 0.0193, 0.0291, 0.2297], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0122, 0.0126, 0.0130, 0.0311, 0.0124, 0.0113, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 00:38:59,103 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32360.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:39:14,592 INFO [train2.py:809] (0/4) Epoch 9, batch 500, loss[ctc_loss=0.09959, att_loss=0.2271, loss=0.2016, over 15631.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009136, over 37.00 utterances.], tot_loss[ctc_loss=0.1187, att_loss=0.2581, loss=0.2302, over 3008110.34 frames. utt_duration=1234 frames, utt_pad_proportion=0.05828, over 9764.82 utterances.], batch size: 37, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:39:23,110 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7527, 5.9997, 5.2624, 5.8221, 5.7033, 5.2489, 5.3576, 5.1490], device='cuda:0'), covar=tensor([0.1277, 0.0930, 0.1170, 0.0739, 0.0702, 0.1480, 0.2203, 0.2424], device='cuda:0'), in_proj_covar=tensor([0.0396, 0.0452, 0.0345, 0.0354, 0.0331, 0.0392, 0.0465, 0.0420], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:40:04,265 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7308, 5.0908, 4.6016, 5.2009, 4.5078, 4.9224, 5.2761, 5.0067], device='cuda:0'), covar=tensor([0.0535, 0.0273, 0.0773, 0.0220, 0.0482, 0.0208, 0.0217, 0.0169], device='cuda:0'), in_proj_covar=tensor([0.0293, 0.0223, 0.0284, 0.0216, 0.0237, 0.0185, 0.0211, 0.0207], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-08 00:40:15,769 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32408.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:40:28,732 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.910e+02 2.692e+02 3.196e+02 3.958e+02 9.983e+02, threshold=6.392e+02, percent-clipped=1.0 2023-03-08 00:40:35,575 INFO [train2.py:809] (0/4) Epoch 9, batch 550, loss[ctc_loss=0.1442, att_loss=0.2505, loss=0.2292, over 15757.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009421, over 38.00 utterances.], tot_loss[ctc_loss=0.1188, att_loss=0.2581, loss=0.2303, over 3062061.22 frames. utt_duration=1217 frames, utt_pad_proportion=0.06495, over 10078.35 utterances.], batch size: 38, lr: 1.24e-02, grad_scale: 16.0 2023-03-08 00:41:13,812 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32444.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:41:55,742 INFO [train2.py:809] (0/4) Epoch 9, batch 600, loss[ctc_loss=0.1176, att_loss=0.2655, loss=0.2359, over 16763.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006658, over 48.00 utterances.], tot_loss[ctc_loss=0.1184, att_loss=0.2578, loss=0.2299, over 3107746.35 frames. utt_duration=1225 frames, utt_pad_proportion=0.06291, over 10161.73 utterances.], batch size: 48, lr: 1.23e-02, grad_scale: 16.0 2023-03-08 00:42:05,236 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 00:42:10,442 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32479.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:42:29,845 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32492.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:42:38,373 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-03-08 00:43:06,490 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.5229, 5.7742, 5.0569, 5.5947, 5.4135, 5.0634, 5.2268, 5.0479], device='cuda:0'), covar=tensor([0.1242, 0.1000, 0.1127, 0.0810, 0.0944, 0.1403, 0.2234, 0.2416], device='cuda:0'), in_proj_covar=tensor([0.0397, 0.0453, 0.0346, 0.0361, 0.0334, 0.0393, 0.0470, 0.0423], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 00:43:07,804 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 2.723e+02 3.556e+02 4.400e+02 8.502e+02, threshold=7.112e+02, percent-clipped=9.0 2023-03-08 00:43:14,504 INFO [train2.py:809] (0/4) Epoch 9, batch 650, loss[ctc_loss=0.1537, att_loss=0.2647, loss=0.2425, over 16539.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006452, over 45.00 utterances.], tot_loss[ctc_loss=0.1179, att_loss=0.2577, loss=0.2298, over 3149684.54 frames. utt_duration=1261 frames, utt_pad_proportion=0.05272, over 10003.41 utterances.], batch size: 45, lr: 1.23e-02, grad_scale: 16.0 2023-03-08 00:43:25,725 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32527.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:43:36,657 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32534.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:44:08,141 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7484, 2.9632, 5.0838, 4.2224, 3.3048, 4.7017, 4.9870, 4.8583], device='cuda:0'), covar=tensor([0.0224, 0.1355, 0.0219, 0.1068, 0.1792, 0.0172, 0.0113, 0.0193], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0240, 0.0124, 0.0298, 0.0278, 0.0181, 0.0108, 0.0144], device='cuda:0'), out_proj_covar=tensor([1.3302e-04, 2.0160e-04, 1.1172e-04, 2.4825e-04, 2.4649e-04, 1.6144e-04, 9.9406e-05, 1.3667e-04], device='cuda:0') 2023-03-08 00:44:33,111 INFO [train2.py:809] (0/4) Epoch 9, batch 700, loss[ctc_loss=0.09063, att_loss=0.2541, loss=0.2214, over 16461.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.007693, over 46.00 utterances.], tot_loss[ctc_loss=0.1191, att_loss=0.2589, loss=0.231, over 3177037.82 frames. utt_duration=1210 frames, utt_pad_proportion=0.06388, over 10512.46 utterances.], batch size: 46, lr: 1.23e-02, grad_scale: 16.0 2023-03-08 00:44:35,671 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32571.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:44:52,242 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32582.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:45:01,042 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-03-08 00:45:06,790 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0718, 5.3140, 5.6328, 5.5361, 5.4575, 5.9958, 5.2946, 6.1184], device='cuda:0'), covar=tensor([0.0648, 0.0631, 0.0593, 0.0849, 0.1745, 0.0780, 0.0458, 0.0559], device='cuda:0'), in_proj_covar=tensor([0.0644, 0.0392, 0.0447, 0.0504, 0.0679, 0.0447, 0.0364, 0.0440], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:45:31,367 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=32606.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:45:47,000 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.724e+02 2.817e+02 3.384e+02 3.932e+02 1.069e+03, threshold=6.768e+02, percent-clipped=2.0 2023-03-08 00:45:51,659 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32619.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:45:51,788 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5409, 5.1080, 4.0264, 5.2400, 4.4341, 4.9734, 5.0403, 4.9867], device='cuda:0'), covar=tensor([0.0540, 0.0305, 0.1422, 0.0199, 0.0485, 0.0219, 0.0356, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0294, 0.0226, 0.0288, 0.0217, 0.0238, 0.0186, 0.0210, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-08 00:45:53,045 INFO [train2.py:809] (0/4) Epoch 9, batch 750, loss[ctc_loss=0.1314, att_loss=0.2853, loss=0.2545, over 17291.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02422, over 59.00 utterances.], tot_loss[ctc_loss=0.1193, att_loss=0.2595, loss=0.2315, over 3199936.40 frames. utt_duration=1206 frames, utt_pad_proportion=0.0656, over 10628.93 utterances.], batch size: 59, lr: 1.23e-02, grad_scale: 16.0 2023-03-08 00:46:47,230 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=32654.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:47:12,885 INFO [train2.py:809] (0/4) Epoch 9, batch 800, loss[ctc_loss=0.1063, att_loss=0.2585, loss=0.228, over 17298.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01222, over 55.00 utterances.], tot_loss[ctc_loss=0.118, att_loss=0.2582, loss=0.2301, over 3214523.20 frames. utt_duration=1249 frames, utt_pad_proportion=0.05592, over 10310.09 utterances.], batch size: 55, lr: 1.23e-02, grad_scale: 16.0 2023-03-08 00:48:11,767 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.70 vs. limit=5.0 2023-03-08 00:48:28,947 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.641e+02 2.761e+02 3.783e+02 4.545e+02 8.141e+02, threshold=7.567e+02, percent-clipped=8.0 2023-03-08 00:48:33,636 INFO [train2.py:809] (0/4) Epoch 9, batch 850, loss[ctc_loss=0.1251, att_loss=0.2648, loss=0.2369, over 16686.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006528, over 46.00 utterances.], tot_loss[ctc_loss=0.1172, att_loss=0.2582, loss=0.23, over 3233786.95 frames. utt_duration=1253 frames, utt_pad_proportion=0.05369, over 10337.65 utterances.], batch size: 46, lr: 1.23e-02, grad_scale: 8.0 2023-03-08 00:49:05,274 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0541, 5.2220, 5.5956, 5.4964, 5.4258, 5.9753, 5.1825, 6.0682], device='cuda:0'), covar=tensor([0.0588, 0.0630, 0.0674, 0.0822, 0.1864, 0.0801, 0.0558, 0.0614], device='cuda:0'), in_proj_covar=tensor([0.0645, 0.0391, 0.0445, 0.0503, 0.0679, 0.0449, 0.0363, 0.0444], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:49:54,782 INFO [train2.py:809] (0/4) Epoch 9, batch 900, loss[ctc_loss=0.1499, att_loss=0.2656, loss=0.2424, over 16317.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.007053, over 45.00 utterances.], tot_loss[ctc_loss=0.1171, att_loss=0.2581, loss=0.2299, over 3249775.48 frames. utt_duration=1266 frames, utt_pad_proportion=0.04772, over 10281.89 utterances.], batch size: 45, lr: 1.23e-02, grad_scale: 8.0 2023-03-08 00:51:11,048 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.656e+02 2.630e+02 3.149e+02 3.930e+02 7.487e+02, threshold=6.299e+02, percent-clipped=0.0 2023-03-08 00:51:15,680 INFO [train2.py:809] (0/4) Epoch 9, batch 950, loss[ctc_loss=0.141, att_loss=0.2867, loss=0.2575, over 17407.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03266, over 63.00 utterances.], tot_loss[ctc_loss=0.1182, att_loss=0.2588, loss=0.2307, over 3246548.94 frames. utt_duration=1255 frames, utt_pad_proportion=0.0525, over 10357.26 utterances.], batch size: 63, lr: 1.23e-02, grad_scale: 8.0 2023-03-08 00:51:42,590 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1787, 5.2319, 5.0052, 2.2880, 1.9606, 2.8425, 3.6517, 3.8229], device='cuda:0'), covar=tensor([0.0619, 0.0161, 0.0207, 0.4458, 0.6017, 0.2500, 0.1310, 0.1854], device='cuda:0'), in_proj_covar=tensor([0.0331, 0.0210, 0.0240, 0.0195, 0.0363, 0.0347, 0.0228, 0.0358], device='cuda:0'), out_proj_covar=tensor([1.5642e-04, 8.0549e-05, 1.0530e-04, 9.0603e-05, 1.6349e-04, 1.4587e-04, 9.0724e-05, 1.5678e-04], device='cuda:0') 2023-03-08 00:52:14,139 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-03-08 00:52:35,534 INFO [train2.py:809] (0/4) Epoch 9, batch 1000, loss[ctc_loss=0.1151, att_loss=0.2511, loss=0.2239, over 16333.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005943, over 45.00 utterances.], tot_loss[ctc_loss=0.1179, att_loss=0.2582, loss=0.2301, over 3238906.41 frames. utt_duration=1234 frames, utt_pad_proportion=0.0612, over 10509.21 utterances.], batch size: 45, lr: 1.23e-02, grad_scale: 8.0 2023-03-08 00:53:50,524 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.656e+02 3.223e+02 3.962e+02 9.648e+02, threshold=6.446e+02, percent-clipped=5.0 2023-03-08 00:53:55,326 INFO [train2.py:809] (0/4) Epoch 9, batch 1050, loss[ctc_loss=0.1222, att_loss=0.2722, loss=0.2422, over 16955.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008137, over 50.00 utterances.], tot_loss[ctc_loss=0.1186, att_loss=0.2584, loss=0.2304, over 3239981.92 frames. utt_duration=1221 frames, utt_pad_proportion=0.06628, over 10629.86 utterances.], batch size: 50, lr: 1.23e-02, grad_scale: 8.0 2023-03-08 00:54:47,727 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-08 00:55:10,113 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=32966.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:55:15,878 INFO [train2.py:809] (0/4) Epoch 9, batch 1100, loss[ctc_loss=0.09458, att_loss=0.2356, loss=0.2074, over 15770.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008714, over 38.00 utterances.], tot_loss[ctc_loss=0.1187, att_loss=0.2594, loss=0.2312, over 3261527.89 frames. utt_duration=1199 frames, utt_pad_proportion=0.06552, over 10894.94 utterances.], batch size: 38, lr: 1.23e-02, grad_scale: 8.0 2023-03-08 00:56:07,602 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0720, 5.1197, 5.0047, 2.5566, 1.9831, 2.8034, 3.8272, 3.8766], device='cuda:0'), covar=tensor([0.0704, 0.0220, 0.0230, 0.3386, 0.5803, 0.2598, 0.1156, 0.1793], device='cuda:0'), in_proj_covar=tensor([0.0324, 0.0207, 0.0232, 0.0187, 0.0351, 0.0336, 0.0222, 0.0346], device='cuda:0'), out_proj_covar=tensor([1.5329e-04, 7.9095e-05, 1.0185e-04, 8.6146e-05, 1.5821e-04, 1.4130e-04, 8.8248e-05, 1.5135e-04], device='cuda:0') 2023-03-08 00:56:31,654 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.884e+02 3.040e+02 3.595e+02 4.615e+02 9.742e+02, threshold=7.190e+02, percent-clipped=7.0 2023-03-08 00:56:36,387 INFO [train2.py:809] (0/4) Epoch 9, batch 1150, loss[ctc_loss=0.1175, att_loss=0.2735, loss=0.2423, over 17027.00 frames. utt_duration=689.5 frames, utt_pad_proportion=0.1327, over 99.00 utterances.], tot_loss[ctc_loss=0.1186, att_loss=0.2591, loss=0.231, over 3261915.02 frames. utt_duration=1213 frames, utt_pad_proportion=0.06206, over 10770.43 utterances.], batch size: 99, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 00:56:47,498 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33027.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:57:07,629 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-03-08 00:57:43,528 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-08 00:57:54,909 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33069.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:57:56,104 INFO [train2.py:809] (0/4) Epoch 9, batch 1200, loss[ctc_loss=0.1044, att_loss=0.2363, loss=0.2099, over 15359.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01137, over 35.00 utterances.], tot_loss[ctc_loss=0.1195, att_loss=0.2596, loss=0.2316, over 3264736.04 frames. utt_duration=1206 frames, utt_pad_proportion=0.06481, over 10842.04 utterances.], batch size: 35, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 00:58:01,034 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4853, 2.5654, 3.6276, 2.5642, 3.3868, 4.6151, 4.4528, 3.0109], device='cuda:0'), covar=tensor([0.0399, 0.1996, 0.1052, 0.1629, 0.1087, 0.0679, 0.0463, 0.1548], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0226, 0.0237, 0.0201, 0.0232, 0.0276, 0.0206, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 00:58:41,563 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7237, 5.0991, 5.2312, 5.2581, 5.1448, 5.6988, 5.0592, 5.7991], device='cuda:0'), covar=tensor([0.0645, 0.0598, 0.0643, 0.1014, 0.1822, 0.0750, 0.0640, 0.0503], device='cuda:0'), in_proj_covar=tensor([0.0643, 0.0391, 0.0444, 0.0502, 0.0681, 0.0444, 0.0360, 0.0441], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0002, 0.0003], device='cuda:0') 2023-03-08 00:58:56,442 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-03-08 00:59:07,459 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 00:59:12,584 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.824e+02 2.465e+02 3.116e+02 3.703e+02 8.772e+02, threshold=6.232e+02, percent-clipped=2.0 2023-03-08 00:59:17,164 INFO [train2.py:809] (0/4) Epoch 9, batch 1250, loss[ctc_loss=0.1378, att_loss=0.2777, loss=0.2498, over 17342.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03537, over 63.00 utterances.], tot_loss[ctc_loss=0.1186, att_loss=0.2589, loss=0.2308, over 3263891.24 frames. utt_duration=1186 frames, utt_pad_proportion=0.07107, over 11017.78 utterances.], batch size: 63, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 00:59:28,308 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1127, 4.6856, 4.5773, 4.5957, 2.6260, 4.4301, 2.6156, 1.5880], device='cuda:0'), covar=tensor([0.0260, 0.0110, 0.0647, 0.0162, 0.1925, 0.0171, 0.1688, 0.1922], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0102, 0.0252, 0.0108, 0.0221, 0.0105, 0.0229, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 00:59:32,915 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33130.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 00:59:45,244 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 01:00:14,116 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33155.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:00:36,806 INFO [train2.py:809] (0/4) Epoch 9, batch 1300, loss[ctc_loss=0.1087, att_loss=0.2523, loss=0.2236, over 16128.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006249, over 42.00 utterances.], tot_loss[ctc_loss=0.1185, att_loss=0.2588, loss=0.2307, over 3265191.06 frames. utt_duration=1209 frames, utt_pad_proportion=0.06518, over 10815.80 utterances.], batch size: 42, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:01:39,463 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.69 vs. limit=5.0 2023-03-08 01:01:51,086 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33216.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:01:52,243 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.789e+02 2.474e+02 3.066e+02 3.872e+02 7.426e+02, threshold=6.132e+02, percent-clipped=4.0 2023-03-08 01:01:56,793 INFO [train2.py:809] (0/4) Epoch 9, batch 1350, loss[ctc_loss=0.1526, att_loss=0.2818, loss=0.256, over 16665.00 frames. utt_duration=674.8 frames, utt_pad_proportion=0.1533, over 99.00 utterances.], tot_loss[ctc_loss=0.1185, att_loss=0.2583, loss=0.2304, over 3253276.94 frames. utt_duration=1193 frames, utt_pad_proportion=0.07212, over 10923.64 utterances.], batch size: 99, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:02:04,864 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1388, 5.1175, 4.9572, 2.5981, 1.9147, 2.6918, 3.7564, 3.8818], device='cuda:0'), covar=tensor([0.0628, 0.0230, 0.0263, 0.3707, 0.6067, 0.2958, 0.1318, 0.1783], device='cuda:0'), in_proj_covar=tensor([0.0322, 0.0209, 0.0233, 0.0187, 0.0352, 0.0340, 0.0221, 0.0349], device='cuda:0'), out_proj_covar=tensor([1.5193e-04, 8.0523e-05, 1.0225e-04, 8.5926e-05, 1.5855e-04, 1.4265e-04, 8.8299e-05, 1.5247e-04], device='cuda:0') 2023-03-08 01:03:16,720 INFO [train2.py:809] (0/4) Epoch 9, batch 1400, loss[ctc_loss=0.1122, att_loss=0.2705, loss=0.2389, over 16963.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007772, over 50.00 utterances.], tot_loss[ctc_loss=0.1164, att_loss=0.2566, loss=0.2286, over 3256584.24 frames. utt_duration=1226 frames, utt_pad_proportion=0.06444, over 10640.30 utterances.], batch size: 50, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:04:32,132 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.539e+02 2.521e+02 2.872e+02 3.748e+02 7.311e+02, threshold=5.745e+02, percent-clipped=4.0 2023-03-08 01:04:37,046 INFO [train2.py:809] (0/4) Epoch 9, batch 1450, loss[ctc_loss=0.1158, att_loss=0.2569, loss=0.2287, over 16755.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006367, over 48.00 utterances.], tot_loss[ctc_loss=0.116, att_loss=0.2564, loss=0.2283, over 3264847.01 frames. utt_duration=1229 frames, utt_pad_proportion=0.062, over 10640.48 utterances.], batch size: 48, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:04:40,491 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33322.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:04:41,405 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-03-08 01:05:56,598 INFO [train2.py:809] (0/4) Epoch 9, batch 1500, loss[ctc_loss=0.1118, att_loss=0.2747, loss=0.2421, over 17029.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.008088, over 51.00 utterances.], tot_loss[ctc_loss=0.117, att_loss=0.2572, loss=0.2292, over 3271413.73 frames. utt_duration=1229 frames, utt_pad_proportion=0.06044, over 10659.07 utterances.], batch size: 51, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:07:11,748 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.661e+02 2.556e+02 3.254e+02 4.022e+02 8.037e+02, threshold=6.507e+02, percent-clipped=3.0 2023-03-08 01:07:16,446 INFO [train2.py:809] (0/4) Epoch 9, batch 1550, loss[ctc_loss=0.0931, att_loss=0.2462, loss=0.2156, over 16271.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007224, over 43.00 utterances.], tot_loss[ctc_loss=0.117, att_loss=0.2573, loss=0.2292, over 3269826.09 frames. utt_duration=1218 frames, utt_pad_proportion=0.06486, over 10753.54 utterances.], batch size: 43, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:07:21,718 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1125, 5.1449, 5.0280, 2.5465, 1.9495, 2.4905, 3.5067, 3.8389], device='cuda:0'), covar=tensor([0.0599, 0.0197, 0.0206, 0.3196, 0.6347, 0.3167, 0.1487, 0.1845], device='cuda:0'), in_proj_covar=tensor([0.0321, 0.0207, 0.0230, 0.0186, 0.0349, 0.0335, 0.0218, 0.0345], device='cuda:0'), out_proj_covar=tensor([1.5092e-04, 7.9593e-05, 1.0066e-04, 8.5163e-05, 1.5622e-04, 1.4058e-04, 8.7186e-05, 1.5065e-04], device='cuda:0') 2023-03-08 01:07:24,576 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33425.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:08:06,191 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-08 01:08:37,541 INFO [train2.py:809] (0/4) Epoch 9, batch 1600, loss[ctc_loss=0.1283, att_loss=0.2708, loss=0.2423, over 16311.00 frames. utt_duration=1451 frames, utt_pad_proportion=0.007205, over 45.00 utterances.], tot_loss[ctc_loss=0.1168, att_loss=0.2577, loss=0.2295, over 3274150.74 frames. utt_duration=1227 frames, utt_pad_proportion=0.06165, over 10686.75 utterances.], batch size: 45, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:08:40,904 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33472.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:09:11,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0650, 4.5392, 4.7357, 5.0847, 2.2348, 4.7398, 2.5929, 2.0468], device='cuda:0'), covar=tensor([0.0310, 0.0141, 0.0605, 0.0084, 0.2441, 0.0167, 0.1924, 0.1794], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0103, 0.0257, 0.0107, 0.0223, 0.0104, 0.0229, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 01:09:43,890 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33511.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:09:53,210 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.888e+02 2.746e+02 3.273e+02 4.611e+02 8.460e+02, threshold=6.545e+02, percent-clipped=8.0 2023-03-08 01:09:57,949 INFO [train2.py:809] (0/4) Epoch 9, batch 1650, loss[ctc_loss=0.1242, att_loss=0.2723, loss=0.2427, over 17123.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01467, over 56.00 utterances.], tot_loss[ctc_loss=0.1168, att_loss=0.2581, loss=0.2299, over 3276694.26 frames. utt_duration=1240 frames, utt_pad_proportion=0.05677, over 10579.56 utterances.], batch size: 56, lr: 1.22e-02, grad_scale: 8.0 2023-03-08 01:10:05,574 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.62 vs. limit=2.0 2023-03-08 01:10:08,262 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-03-08 01:10:18,949 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33533.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:10:45,618 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8162, 3.9026, 4.0404, 3.9375, 4.0735, 4.0682, 3.8140, 3.7102], device='cuda:0'), covar=tensor([0.1142, 0.1026, 0.0315, 0.0502, 0.0353, 0.0362, 0.0390, 0.0392], device='cuda:0'), in_proj_covar=tensor([0.0435, 0.0271, 0.0215, 0.0254, 0.0319, 0.0340, 0.0259, 0.0298], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 01:11:18,159 INFO [train2.py:809] (0/4) Epoch 9, batch 1700, loss[ctc_loss=0.1086, att_loss=0.2358, loss=0.2104, over 15866.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.009758, over 39.00 utterances.], tot_loss[ctc_loss=0.1167, att_loss=0.2581, loss=0.2298, over 3279458.87 frames. utt_duration=1236 frames, utt_pad_proportion=0.0573, over 10628.94 utterances.], batch size: 39, lr: 1.21e-02, grad_scale: 8.0 2023-03-08 01:12:35,922 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.664e+02 3.328e+02 4.191e+02 7.435e+02, threshold=6.656e+02, percent-clipped=3.0 2023-03-08 01:12:39,100 INFO [train2.py:809] (0/4) Epoch 9, batch 1750, loss[ctc_loss=0.1057, att_loss=0.2476, loss=0.2192, over 16136.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005679, over 42.00 utterances.], tot_loss[ctc_loss=0.1167, att_loss=0.2585, loss=0.2302, over 3286823.02 frames. utt_duration=1239 frames, utt_pad_proportion=0.05438, over 10623.63 utterances.], batch size: 42, lr: 1.21e-02, grad_scale: 4.0 2023-03-08 01:12:42,484 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33622.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:13:58,113 INFO [train2.py:809] (0/4) Epoch 9, batch 1800, loss[ctc_loss=0.1224, att_loss=0.2716, loss=0.2417, over 16872.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007987, over 49.00 utterances.], tot_loss[ctc_loss=0.117, att_loss=0.2587, loss=0.2304, over 3290730.68 frames. utt_duration=1257 frames, utt_pad_proportion=0.04808, over 10482.63 utterances.], batch size: 49, lr: 1.21e-02, grad_scale: 4.0 2023-03-08 01:13:58,193 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33670.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:14:09,846 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33677.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:14:12,671 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8198, 5.1282, 5.0902, 5.1307, 5.2045, 5.1866, 4.8652, 4.6442], device='cuda:0'), covar=tensor([0.1040, 0.0500, 0.0195, 0.0428, 0.0273, 0.0253, 0.0272, 0.0309], device='cuda:0'), in_proj_covar=tensor([0.0431, 0.0267, 0.0215, 0.0252, 0.0320, 0.0339, 0.0258, 0.0295], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 01:15:01,271 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2023-03-08 01:15:10,288 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7653, 2.2111, 4.9761, 4.0299, 3.2859, 4.6734, 4.7017, 4.8255], device='cuda:0'), covar=tensor([0.0105, 0.1786, 0.0119, 0.0795, 0.1528, 0.0124, 0.0089, 0.0116], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0249, 0.0129, 0.0308, 0.0286, 0.0184, 0.0113, 0.0147], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0003, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-03-08 01:15:14,676 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.773e+02 3.319e+02 4.110e+02 8.879e+02, threshold=6.639e+02, percent-clipped=6.0 2023-03-08 01:15:17,761 INFO [train2.py:809] (0/4) Epoch 9, batch 1850, loss[ctc_loss=0.1245, att_loss=0.2622, loss=0.2347, over 17040.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.009797, over 53.00 utterances.], tot_loss[ctc_loss=0.1162, att_loss=0.2576, loss=0.2294, over 3291431.27 frames. utt_duration=1282 frames, utt_pad_proportion=0.04199, over 10284.23 utterances.], batch size: 53, lr: 1.21e-02, grad_scale: 4.0 2023-03-08 01:15:23,526 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 01:15:26,852 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33725.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:15:47,512 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=33738.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 01:16:17,901 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3222, 5.0637, 5.0972, 2.2988, 2.0193, 2.5658, 4.0999, 3.7084], device='cuda:0'), covar=tensor([0.0625, 0.0368, 0.0320, 0.3584, 0.7131, 0.3436, 0.0726, 0.2211], device='cuda:0'), in_proj_covar=tensor([0.0318, 0.0204, 0.0228, 0.0188, 0.0350, 0.0336, 0.0218, 0.0346], device='cuda:0'), out_proj_covar=tensor([1.5036e-04, 7.8456e-05, 9.9907e-05, 8.5819e-05, 1.5617e-04, 1.4061e-04, 8.6681e-05, 1.5068e-04], device='cuda:0') 2023-03-08 01:16:37,508 INFO [train2.py:809] (0/4) Epoch 9, batch 1900, loss[ctc_loss=0.09471, att_loss=0.243, loss=0.2133, over 16174.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.007129, over 41.00 utterances.], tot_loss[ctc_loss=0.1168, att_loss=0.2579, loss=0.2297, over 3277492.14 frames. utt_duration=1259 frames, utt_pad_proportion=0.05068, over 10429.09 utterances.], batch size: 41, lr: 1.21e-02, grad_scale: 4.0 2023-03-08 01:16:42,269 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33773.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:16:53,188 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.79 vs. limit=5.0 2023-03-08 01:17:06,066 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-03-08 01:17:29,480 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8177, 6.0474, 5.4902, 5.9382, 5.6892, 5.3602, 5.4637, 5.3739], device='cuda:0'), covar=tensor([0.1085, 0.0843, 0.0820, 0.0628, 0.0840, 0.1137, 0.1824, 0.1933], device='cuda:0'), in_proj_covar=tensor([0.0403, 0.0462, 0.0355, 0.0366, 0.0333, 0.0398, 0.0479, 0.0432], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 01:17:44,818 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=33811.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:17:55,252 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.768e+02 2.711e+02 3.222e+02 4.216e+02 1.081e+03, threshold=6.444e+02, percent-clipped=7.0 2023-03-08 01:17:58,263 INFO [train2.py:809] (0/4) Epoch 9, batch 1950, loss[ctc_loss=0.1244, att_loss=0.2594, loss=0.2324, over 17060.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.00764, over 52.00 utterances.], tot_loss[ctc_loss=0.1178, att_loss=0.2588, loss=0.2306, over 3273735.25 frames. utt_duration=1225 frames, utt_pad_proportion=0.06081, over 10705.53 utterances.], batch size: 52, lr: 1.21e-02, grad_scale: 4.0 2023-03-08 01:18:11,626 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=33828.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:19:01,164 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=33859.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:19:02,911 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4579, 2.4290, 3.1787, 4.4375, 4.1501, 4.1033, 2.8788, 2.3292], device='cuda:0'), covar=tensor([0.0543, 0.2359, 0.1240, 0.0376, 0.0484, 0.0247, 0.1421, 0.2095], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0200, 0.0188, 0.0182, 0.0172, 0.0139, 0.0190, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 01:19:17,451 INFO [train2.py:809] (0/4) Epoch 9, batch 2000, loss[ctc_loss=0.1324, att_loss=0.2865, loss=0.2557, over 17049.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009862, over 53.00 utterances.], tot_loss[ctc_loss=0.1173, att_loss=0.2583, loss=0.2301, over 3267966.61 frames. utt_duration=1252 frames, utt_pad_proportion=0.05431, over 10455.36 utterances.], batch size: 53, lr: 1.21e-02, grad_scale: 8.0 2023-03-08 01:20:34,522 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.651e+02 2.627e+02 3.033e+02 3.756e+02 1.034e+03, threshold=6.065e+02, percent-clipped=4.0 2023-03-08 01:20:37,623 INFO [train2.py:809] (0/4) Epoch 9, batch 2050, loss[ctc_loss=0.1448, att_loss=0.2668, loss=0.2424, over 16557.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005342, over 45.00 utterances.], tot_loss[ctc_loss=0.1171, att_loss=0.2579, loss=0.2297, over 3262887.14 frames. utt_duration=1236 frames, utt_pad_proportion=0.06006, over 10569.62 utterances.], batch size: 45, lr: 1.21e-02, grad_scale: 8.0 2023-03-08 01:20:42,380 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-03-08 01:20:57,564 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9100, 4.9015, 4.8578, 4.5889, 5.3971, 5.1122, 4.7812, 2.7564], device='cuda:0'), covar=tensor([0.0185, 0.0212, 0.0153, 0.0327, 0.0974, 0.0144, 0.0204, 0.1885], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0125, 0.0125, 0.0131, 0.0316, 0.0124, 0.0116, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 01:21:08,841 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=33939.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:21:58,312 INFO [train2.py:809] (0/4) Epoch 9, batch 2100, loss[ctc_loss=0.1202, att_loss=0.2691, loss=0.2393, over 17118.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01438, over 56.00 utterances.], tot_loss[ctc_loss=0.1171, att_loss=0.258, loss=0.2298, over 3264156.31 frames. utt_duration=1245 frames, utt_pad_proportion=0.05627, over 10501.54 utterances.], batch size: 56, lr: 1.21e-02, grad_scale: 8.0 2023-03-08 01:22:45,748 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-34000.pt 2023-03-08 01:22:50,653 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34000.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:23:18,325 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.708e+02 2.606e+02 3.129e+02 4.184e+02 1.323e+03, threshold=6.258e+02, percent-clipped=9.0 2023-03-08 01:23:22,286 INFO [train2.py:809] (0/4) Epoch 9, batch 2150, loss[ctc_loss=0.1215, att_loss=0.27, loss=0.2403, over 17297.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02388, over 59.00 utterances.], tot_loss[ctc_loss=0.1173, att_loss=0.258, loss=0.2299, over 3265799.79 frames. utt_duration=1256 frames, utt_pad_proportion=0.05131, over 10413.44 utterances.], batch size: 59, lr: 1.21e-02, grad_scale: 8.0 2023-03-08 01:23:42,958 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34033.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 01:23:44,413 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-03-08 01:24:02,282 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34044.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:24:42,986 INFO [train2.py:809] (0/4) Epoch 9, batch 2200, loss[ctc_loss=0.105, att_loss=0.2604, loss=0.2293, over 16781.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005741, over 48.00 utterances.], tot_loss[ctc_loss=0.1176, att_loss=0.2579, loss=0.2299, over 3263213.13 frames. utt_duration=1243 frames, utt_pad_proportion=0.05668, over 10513.39 utterances.], batch size: 48, lr: 1.21e-02, grad_scale: 8.0 2023-03-08 01:25:00,274 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2090, 5.2799, 5.0766, 2.6597, 2.1132, 2.7753, 3.8407, 3.8485], device='cuda:0'), covar=tensor([0.0607, 0.0229, 0.0236, 0.3171, 0.5903, 0.2556, 0.1139, 0.1866], device='cuda:0'), in_proj_covar=tensor([0.0319, 0.0208, 0.0226, 0.0189, 0.0349, 0.0335, 0.0220, 0.0347], device='cuda:0'), out_proj_covar=tensor([1.5013e-04, 7.9061e-05, 9.9255e-05, 8.5821e-05, 1.5583e-04, 1.3975e-04, 8.7288e-05, 1.5059e-04], device='cuda:0') 2023-03-08 01:25:38,271 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34105.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:25:58,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.871e+02 2.581e+02 3.160e+02 3.967e+02 1.179e+03, threshold=6.319e+02, percent-clipped=5.0 2023-03-08 01:26:01,794 INFO [train2.py:809] (0/4) Epoch 9, batch 2250, loss[ctc_loss=0.115, att_loss=0.2675, loss=0.237, over 17472.00 frames. utt_duration=886.2 frames, utt_pad_proportion=0.07205, over 79.00 utterances.], tot_loss[ctc_loss=0.1184, att_loss=0.2584, loss=0.2304, over 3268383.40 frames. utt_duration=1219 frames, utt_pad_proportion=0.06122, over 10736.29 utterances.], batch size: 79, lr: 1.21e-02, grad_scale: 8.0 2023-03-08 01:26:14,012 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34128.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:26:20,074 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5083, 3.7581, 3.6416, 2.8536, 3.6197, 3.6678, 3.4444, 2.1095], device='cuda:0'), covar=tensor([0.1224, 0.1438, 0.1985, 0.7086, 0.2884, 0.3866, 0.0818, 0.9155], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0104, 0.0109, 0.0176, 0.0092, 0.0161, 0.0091, 0.0158], device='cuda:0'), out_proj_covar=tensor([8.0438e-05, 8.7970e-05, 9.6050e-05, 1.4004e-04, 8.2872e-05, 1.3113e-04, 7.8764e-05, 1.2776e-04], device='cuda:0') 2023-03-08 01:26:42,322 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34145.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:26:43,791 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2240, 4.8537, 4.6252, 4.8234, 4.8590, 4.5364, 3.4211, 4.7928], device='cuda:0'), covar=tensor([0.0103, 0.0094, 0.0112, 0.0065, 0.0079, 0.0094, 0.0586, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0067, 0.0065, 0.0078, 0.0049, 0.0053, 0.0063, 0.0086, 0.0084], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 01:27:21,330 INFO [train2.py:809] (0/4) Epoch 9, batch 2300, loss[ctc_loss=0.1117, att_loss=0.2549, loss=0.2263, over 16879.00 frames. utt_duration=683.4 frames, utt_pad_proportion=0.1425, over 99.00 utterances.], tot_loss[ctc_loss=0.1179, att_loss=0.2578, loss=0.2298, over 3261718.11 frames. utt_duration=1217 frames, utt_pad_proportion=0.06351, over 10735.12 utterances.], batch size: 99, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:27:30,919 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34176.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:28:20,189 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34206.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:28:39,045 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.844e+02 2.559e+02 3.119e+02 3.946e+02 1.160e+03, threshold=6.239e+02, percent-clipped=6.0 2023-03-08 01:28:42,205 INFO [train2.py:809] (0/4) Epoch 9, batch 2350, loss[ctc_loss=0.1111, att_loss=0.2592, loss=0.2296, over 15999.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007714, over 40.00 utterances.], tot_loss[ctc_loss=0.1173, att_loss=0.2575, loss=0.2295, over 3272116.46 frames. utt_duration=1229 frames, utt_pad_proportion=0.05833, over 10663.33 utterances.], batch size: 40, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:29:16,778 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2023-03-08 01:30:03,494 INFO [train2.py:809] (0/4) Epoch 9, batch 2400, loss[ctc_loss=0.1532, att_loss=0.2808, loss=0.2552, over 16487.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005333, over 46.00 utterances.], tot_loss[ctc_loss=0.1166, att_loss=0.2573, loss=0.2291, over 3270941.29 frames. utt_duration=1254 frames, utt_pad_proportion=0.05314, over 10442.40 utterances.], batch size: 46, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:30:45,155 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34295.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:31:18,996 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6249, 4.6335, 4.6500, 4.7015, 5.1372, 4.6959, 4.5899, 2.4128], device='cuda:0'), covar=tensor([0.0197, 0.0365, 0.0220, 0.0248, 0.0953, 0.0184, 0.0307, 0.2058], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0126, 0.0125, 0.0133, 0.0319, 0.0127, 0.0115, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 01:31:21,639 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.665e+02 2.539e+02 3.254e+02 3.979e+02 8.947e+02, threshold=6.508e+02, percent-clipped=3.0 2023-03-08 01:31:24,804 INFO [train2.py:809] (0/4) Epoch 9, batch 2450, loss[ctc_loss=0.1668, att_loss=0.2855, loss=0.2618, over 17174.00 frames. utt_duration=695.3 frames, utt_pad_proportion=0.1265, over 99.00 utterances.], tot_loss[ctc_loss=0.1174, att_loss=0.2572, loss=0.2293, over 3261419.62 frames. utt_duration=1220 frames, utt_pad_proportion=0.06459, over 10710.25 utterances.], batch size: 99, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:31:32,951 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8466, 6.0928, 5.4174, 5.9694, 5.7569, 5.4053, 5.5109, 5.3651], device='cuda:0'), covar=tensor([0.1115, 0.0891, 0.0794, 0.0709, 0.0673, 0.1237, 0.2019, 0.1956], device='cuda:0'), in_proj_covar=tensor([0.0409, 0.0472, 0.0364, 0.0372, 0.0337, 0.0410, 0.0494, 0.0440], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 01:31:46,968 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34333.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:32:45,724 INFO [train2.py:809] (0/4) Epoch 9, batch 2500, loss[ctc_loss=0.1135, att_loss=0.2579, loss=0.229, over 17376.00 frames. utt_duration=1009 frames, utt_pad_proportion=0.04917, over 69.00 utterances.], tot_loss[ctc_loss=0.1178, att_loss=0.2572, loss=0.2294, over 3257956.10 frames. utt_duration=1205 frames, utt_pad_proportion=0.06811, over 10824.09 utterances.], batch size: 69, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:33:03,103 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34381.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:33:35,233 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34400.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:34:03,745 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.575e+02 3.136e+02 3.976e+02 7.316e+02, threshold=6.272e+02, percent-clipped=1.0 2023-03-08 01:34:06,905 INFO [train2.py:809] (0/4) Epoch 9, batch 2550, loss[ctc_loss=0.1481, att_loss=0.28, loss=0.2536, over 17136.00 frames. utt_duration=693.9 frames, utt_pad_proportion=0.1283, over 99.00 utterances.], tot_loss[ctc_loss=0.1162, att_loss=0.2562, loss=0.2282, over 3260986.04 frames. utt_duration=1229 frames, utt_pad_proportion=0.06075, over 10628.16 utterances.], batch size: 99, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:34:29,298 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9296, 4.4144, 4.2002, 4.8882, 2.0673, 4.6628, 2.2880, 1.6152], device='cuda:0'), covar=tensor([0.0332, 0.0178, 0.0818, 0.0107, 0.2559, 0.0129, 0.2053, 0.2027], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0107, 0.0258, 0.0109, 0.0223, 0.0103, 0.0229, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 01:34:41,386 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.66 vs. limit=5.0 2023-03-08 01:35:06,205 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1499, 5.4033, 5.6444, 5.5800, 5.5359, 6.0852, 5.2623, 6.1232], device='cuda:0'), covar=tensor([0.0645, 0.0676, 0.0695, 0.0940, 0.1994, 0.0894, 0.0498, 0.0662], device='cuda:0'), in_proj_covar=tensor([0.0658, 0.0400, 0.0458, 0.0525, 0.0700, 0.0464, 0.0377, 0.0450], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 01:35:28,479 INFO [train2.py:809] (0/4) Epoch 9, batch 2600, loss[ctc_loss=0.09105, att_loss=0.2341, loss=0.2055, over 15627.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.008761, over 37.00 utterances.], tot_loss[ctc_loss=0.1157, att_loss=0.2559, loss=0.2279, over 3257827.21 frames. utt_duration=1226 frames, utt_pad_proportion=0.06311, over 10639.26 utterances.], batch size: 37, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:35:28,746 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5298, 2.7562, 3.2497, 4.4256, 4.0361, 4.0238, 3.0703, 2.0141], device='cuda:0'), covar=tensor([0.0588, 0.2042, 0.1012, 0.0494, 0.0515, 0.0285, 0.1488, 0.2533], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0200, 0.0184, 0.0177, 0.0169, 0.0139, 0.0189, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 01:36:19,394 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34501.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:36:46,601 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.544e+02 3.202e+02 3.743e+02 1.159e+03, threshold=6.403e+02, percent-clipped=3.0 2023-03-08 01:36:49,751 INFO [train2.py:809] (0/4) Epoch 9, batch 2650, loss[ctc_loss=0.1273, att_loss=0.282, loss=0.2511, over 17051.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.008897, over 53.00 utterances.], tot_loss[ctc_loss=0.1148, att_loss=0.2559, loss=0.2277, over 3268700.30 frames. utt_duration=1238 frames, utt_pad_proportion=0.05766, over 10571.46 utterances.], batch size: 53, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:36:51,668 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2629, 4.4328, 4.7687, 5.0219, 2.1274, 4.6909, 2.7825, 1.8516], device='cuda:0'), covar=tensor([0.0263, 0.0216, 0.0594, 0.0101, 0.2461, 0.0132, 0.1725, 0.1982], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0107, 0.0256, 0.0109, 0.0221, 0.0102, 0.0228, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 01:36:52,340 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-08 01:37:14,253 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-08 01:38:10,640 INFO [train2.py:809] (0/4) Epoch 9, batch 2700, loss[ctc_loss=0.1541, att_loss=0.263, loss=0.2412, over 15867.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.0105, over 39.00 utterances.], tot_loss[ctc_loss=0.1146, att_loss=0.2557, loss=0.2275, over 3271482.97 frames. utt_duration=1252 frames, utt_pad_proportion=0.05398, over 10461.20 utterances.], batch size: 39, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:38:47,888 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34593.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:38:50,979 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34595.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:39:27,954 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 2.068e+02 2.571e+02 3.320e+02 3.912e+02 1.043e+03, threshold=6.640e+02, percent-clipped=3.0 2023-03-08 01:39:30,981 INFO [train2.py:809] (0/4) Epoch 9, batch 2750, loss[ctc_loss=0.1022, att_loss=0.2409, loss=0.2132, over 15624.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009579, over 37.00 utterances.], tot_loss[ctc_loss=0.1157, att_loss=0.2562, loss=0.2281, over 3269140.37 frames. utt_duration=1236 frames, utt_pad_proportion=0.05926, over 10591.54 utterances.], batch size: 37, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:39:34,341 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34622.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:39:40,634 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34626.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:40:02,781 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.89 vs. limit=5.0 2023-03-08 01:40:08,092 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34643.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:40:25,627 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34654.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:40:50,934 INFO [train2.py:809] (0/4) Epoch 9, batch 2800, loss[ctc_loss=0.1657, att_loss=0.2903, loss=0.2654, over 13636.00 frames. utt_duration=377.7 frames, utt_pad_proportion=0.3432, over 145.00 utterances.], tot_loss[ctc_loss=0.1162, att_loss=0.2569, loss=0.2288, over 3271349.82 frames. utt_duration=1214 frames, utt_pad_proportion=0.06328, over 10791.52 utterances.], batch size: 145, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:41:12,998 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34683.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 01:41:19,692 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34687.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:41:24,608 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1227, 5.1411, 5.0379, 2.4670, 1.8690, 2.6829, 3.2152, 3.6396], device='cuda:0'), covar=tensor([0.0633, 0.0187, 0.0215, 0.3631, 0.6225, 0.2827, 0.1671, 0.2075], device='cuda:0'), in_proj_covar=tensor([0.0322, 0.0209, 0.0224, 0.0190, 0.0351, 0.0337, 0.0223, 0.0347], device='cuda:0'), out_proj_covar=tensor([1.5103e-04, 8.0367e-05, 9.8147e-05, 8.5759e-05, 1.5612e-04, 1.4019e-04, 8.8589e-05, 1.5080e-04], device='cuda:0') 2023-03-08 01:41:40,031 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34700.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:41:55,150 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=34710.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 01:42:07,899 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.689e+02 3.286e+02 4.031e+02 7.048e+02, threshold=6.572e+02, percent-clipped=5.0 2023-03-08 01:42:11,089 INFO [train2.py:809] (0/4) Epoch 9, batch 2850, loss[ctc_loss=0.08355, att_loss=0.2362, loss=0.2057, over 16178.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.007037, over 41.00 utterances.], tot_loss[ctc_loss=0.1152, att_loss=0.2564, loss=0.2282, over 3275749.28 frames. utt_duration=1226 frames, utt_pad_proportion=0.05961, over 10699.39 utterances.], batch size: 41, lr: 1.20e-02, grad_scale: 8.0 2023-03-08 01:42:26,694 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1524, 2.1159, 2.9926, 2.1980, 2.8943, 3.4303, 3.3833, 2.3887], device='cuda:0'), covar=tensor([0.0664, 0.2055, 0.1071, 0.1643, 0.0927, 0.0980, 0.0624, 0.1707], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0220, 0.0238, 0.0202, 0.0232, 0.0277, 0.0207, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-03-08 01:42:56,383 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34748.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:42:58,134 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3483, 1.4380, 2.3110, 1.9030, 3.1076, 2.1544, 1.8513, 2.7038], device='cuda:0'), covar=tensor([0.0575, 0.4601, 0.2645, 0.1517, 0.0697, 0.1511, 0.3036, 0.1199], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0086, 0.0085, 0.0075, 0.0072, 0.0068, 0.0081, 0.0061], device='cuda:0'), out_proj_covar=tensor([4.3794e-05, 5.5234e-05, 5.4388e-05, 4.6671e-05, 4.2754e-05, 4.5604e-05, 5.2455e-05, 4.1705e-05], device='cuda:0') 2023-03-08 01:43:12,608 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2023-03-08 01:43:31,237 INFO [train2.py:809] (0/4) Epoch 9, batch 2900, loss[ctc_loss=0.09999, att_loss=0.2237, loss=0.1989, over 15519.00 frames. utt_duration=1726 frames, utt_pad_proportion=0.007635, over 36.00 utterances.], tot_loss[ctc_loss=0.1135, att_loss=0.2552, loss=0.2269, over 3269475.98 frames. utt_duration=1255 frames, utt_pad_proportion=0.05402, over 10431.18 utterances.], batch size: 36, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:43:33,173 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=34771.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 01:44:11,324 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 01:44:22,022 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=34801.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:44:48,782 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.507e+02 2.823e+02 3.310e+02 4.171e+02 9.231e+02, threshold=6.619e+02, percent-clipped=4.0 2023-03-08 01:44:51,949 INFO [train2.py:809] (0/4) Epoch 9, batch 2950, loss[ctc_loss=0.1134, att_loss=0.26, loss=0.2306, over 16686.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.005848, over 46.00 utterances.], tot_loss[ctc_loss=0.1146, att_loss=0.2555, loss=0.2273, over 3265546.05 frames. utt_duration=1247 frames, utt_pad_proportion=0.05647, over 10489.29 utterances.], batch size: 46, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:44:54,665 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-08 01:45:40,203 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=34849.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:45:56,340 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4979, 3.8164, 3.5764, 2.9034, 3.5369, 3.5782, 3.4246, 2.2467], device='cuda:0'), covar=tensor([0.1527, 0.1375, 0.6810, 0.8068, 1.0601, 0.6176, 0.1128, 1.0154], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0102, 0.0108, 0.0171, 0.0093, 0.0158, 0.0089, 0.0155], device='cuda:0'), out_proj_covar=tensor([7.9971e-05, 8.6817e-05, 9.5951e-05, 1.3741e-04, 8.3179e-05, 1.2908e-04, 7.7490e-05, 1.2557e-04], device='cuda:0') 2023-03-08 01:46:01,081 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0531, 5.4362, 4.8843, 5.5321, 4.8257, 5.1030, 5.5830, 5.2400], device='cuda:0'), covar=tensor([0.0491, 0.0243, 0.0748, 0.0181, 0.0370, 0.0201, 0.0171, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0311, 0.0240, 0.0301, 0.0232, 0.0246, 0.0192, 0.0222, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 01:46:13,923 INFO [train2.py:809] (0/4) Epoch 9, batch 3000, loss[ctc_loss=0.1154, att_loss=0.2428, loss=0.2173, over 16172.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007389, over 41.00 utterances.], tot_loss[ctc_loss=0.1147, att_loss=0.2556, loss=0.2274, over 3258910.62 frames. utt_duration=1234 frames, utt_pad_proportion=0.06161, over 10578.96 utterances.], batch size: 41, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:46:13,926 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 01:46:32,377 INFO [train2.py:843] (0/4) Epoch 9, validation: ctc_loss=0.05401, att_loss=0.2408, loss=0.2035, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 01:46:32,378 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16076MB 2023-03-08 01:46:47,566 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6353, 4.8992, 4.3250, 4.7788, 4.5080, 4.2901, 4.4255, 4.2128], device='cuda:0'), covar=tensor([0.1212, 0.1121, 0.0848, 0.0910, 0.0916, 0.1469, 0.2379, 0.2529], device='cuda:0'), in_proj_covar=tensor([0.0414, 0.0471, 0.0362, 0.0371, 0.0341, 0.0406, 0.0490, 0.0436], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 01:47:49,369 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.566e+02 2.760e+02 3.196e+02 3.997e+02 8.726e+02, threshold=6.392e+02, percent-clipped=4.0 2023-03-08 01:47:52,527 INFO [train2.py:809] (0/4) Epoch 9, batch 3050, loss[ctc_loss=0.1113, att_loss=0.2473, loss=0.2201, over 16120.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006741, over 42.00 utterances.], tot_loss[ctc_loss=0.1145, att_loss=0.2552, loss=0.227, over 3255290.57 frames. utt_duration=1240 frames, utt_pad_proportion=0.06089, over 10517.32 utterances.], batch size: 42, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:48:14,360 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2795, 2.6055, 3.5882, 2.7225, 3.4407, 4.6289, 4.3354, 3.1717], device='cuda:0'), covar=tensor([0.0525, 0.1920, 0.1076, 0.1519, 0.1119, 0.0549, 0.0484, 0.1336], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0223, 0.0237, 0.0203, 0.0233, 0.0279, 0.0207, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 01:48:39,996 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34949.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:48:59,977 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5685, 1.5764, 1.7046, 1.7593, 1.9314, 1.9311, 1.4963, 2.6741], device='cuda:0'), covar=tensor([0.1357, 0.5733, 0.4846, 0.1296, 0.2719, 0.1915, 0.2397, 0.1098], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0083, 0.0084, 0.0072, 0.0071, 0.0067, 0.0079, 0.0059], device='cuda:0'), out_proj_covar=tensor([4.3185e-05, 5.3599e-05, 5.3663e-05, 4.5153e-05, 4.1872e-05, 4.4896e-05, 5.1412e-05, 4.0934e-05], device='cuda:0') 2023-03-08 01:49:12,665 INFO [train2.py:809] (0/4) Epoch 9, batch 3100, loss[ctc_loss=0.113, att_loss=0.2563, loss=0.2276, over 16533.00 frames. utt_duration=676.5 frames, utt_pad_proportion=0.1522, over 98.00 utterances.], tot_loss[ctc_loss=0.1145, att_loss=0.2556, loss=0.2274, over 3257626.61 frames. utt_duration=1243 frames, utt_pad_proportion=0.05848, over 10496.29 utterances.], batch size: 98, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:49:24,964 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34978.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 01:49:31,859 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=34982.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:50:30,866 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.761e+02 2.700e+02 3.373e+02 4.220e+02 7.587e+02, threshold=6.745e+02, percent-clipped=2.0 2023-03-08 01:50:34,012 INFO [train2.py:809] (0/4) Epoch 9, batch 3150, loss[ctc_loss=0.1036, att_loss=0.2586, loss=0.2276, over 17035.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.009826, over 53.00 utterances.], tot_loss[ctc_loss=0.1141, att_loss=0.2554, loss=0.2271, over 3248991.87 frames. utt_duration=1219 frames, utt_pad_proportion=0.06842, over 10674.33 utterances.], batch size: 53, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:51:24,733 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6291, 5.8800, 5.2321, 5.6885, 5.5273, 5.1861, 5.2313, 5.1847], device='cuda:0'), covar=tensor([0.1224, 0.0974, 0.0912, 0.0838, 0.0780, 0.1336, 0.2587, 0.2499], device='cuda:0'), in_proj_covar=tensor([0.0419, 0.0480, 0.0365, 0.0374, 0.0347, 0.0406, 0.0499, 0.0442], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 01:51:48,788 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35066.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 01:51:54,778 INFO [train2.py:809] (0/4) Epoch 9, batch 3200, loss[ctc_loss=0.1313, att_loss=0.2612, loss=0.2352, over 16459.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.007603, over 46.00 utterances.], tot_loss[ctc_loss=0.1142, att_loss=0.2554, loss=0.2272, over 3249495.88 frames. utt_duration=1217 frames, utt_pad_proportion=0.06925, over 10691.43 utterances.], batch size: 46, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:52:31,380 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35092.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:53:04,471 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-03-08 01:53:13,426 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.527e+02 2.431e+02 2.913e+02 3.671e+02 9.523e+02, threshold=5.826e+02, percent-clipped=2.0 2023-03-08 01:53:16,524 INFO [train2.py:809] (0/4) Epoch 9, batch 3250, loss[ctc_loss=0.1044, att_loss=0.2503, loss=0.2211, over 16181.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006259, over 41.00 utterances.], tot_loss[ctc_loss=0.1139, att_loss=0.2555, loss=0.2272, over 3255436.09 frames. utt_duration=1218 frames, utt_pad_proportion=0.06661, over 10700.93 utterances.], batch size: 41, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:54:10,608 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35153.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:54:13,152 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-03-08 01:54:37,145 INFO [train2.py:809] (0/4) Epoch 9, batch 3300, loss[ctc_loss=0.1056, att_loss=0.2641, loss=0.2324, over 16859.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.008751, over 49.00 utterances.], tot_loss[ctc_loss=0.1138, att_loss=0.2553, loss=0.227, over 3262632.43 frames. utt_duration=1237 frames, utt_pad_proportion=0.05919, over 10566.63 utterances.], batch size: 49, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:55:55,871 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.545e+02 2.550e+02 3.164e+02 3.869e+02 8.354e+02, threshold=6.328e+02, percent-clipped=5.0 2023-03-08 01:55:58,989 INFO [train2.py:809] (0/4) Epoch 9, batch 3350, loss[ctc_loss=0.08412, att_loss=0.221, loss=0.1936, over 16019.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006707, over 40.00 utterances.], tot_loss[ctc_loss=0.1136, att_loss=0.2556, loss=0.2272, over 3262939.87 frames. utt_duration=1244 frames, utt_pad_proportion=0.05677, over 10500.76 utterances.], batch size: 40, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:56:46,136 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35249.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:56:53,051 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-03-08 01:57:18,719 INFO [train2.py:809] (0/4) Epoch 9, batch 3400, loss[ctc_loss=0.119, att_loss=0.2597, loss=0.2315, over 17060.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008592, over 52.00 utterances.], tot_loss[ctc_loss=0.1145, att_loss=0.2557, loss=0.2274, over 3258561.40 frames. utt_duration=1231 frames, utt_pad_proportion=0.06082, over 10598.72 utterances.], batch size: 52, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:57:32,786 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35278.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 01:57:38,861 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35282.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:58:03,205 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35297.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:58:36,339 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.749e+02 3.298e+02 4.168e+02 1.046e+03, threshold=6.595e+02, percent-clipped=7.0 2023-03-08 01:58:39,566 INFO [train2.py:809] (0/4) Epoch 9, batch 3450, loss[ctc_loss=0.09179, att_loss=0.2484, loss=0.217, over 17027.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007971, over 51.00 utterances.], tot_loss[ctc_loss=0.1139, att_loss=0.2555, loss=0.2272, over 3258780.73 frames. utt_duration=1223 frames, utt_pad_proportion=0.06345, over 10673.10 utterances.], batch size: 51, lr: 1.19e-02, grad_scale: 8.0 2023-03-08 01:58:50,367 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35326.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:58:56,814 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35330.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 01:59:16,043 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-03-08 01:59:21,254 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35345.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 01:59:54,982 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35366.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 02:00:01,142 INFO [train2.py:809] (0/4) Epoch 9, batch 3500, loss[ctc_loss=0.1086, att_loss=0.2389, loss=0.2128, over 15626.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009199, over 37.00 utterances.], tot_loss[ctc_loss=0.1132, att_loss=0.2547, loss=0.2264, over 3259877.04 frames. utt_duration=1235 frames, utt_pad_proportion=0.05977, over 10569.21 utterances.], batch size: 37, lr: 1.18e-02, grad_scale: 8.0 2023-03-08 02:00:08,979 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35374.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:00:32,209 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8542, 4.6184, 4.5767, 4.6948, 5.0079, 4.6486, 4.4486, 2.2985], device='cuda:0'), covar=tensor([0.0166, 0.0240, 0.0216, 0.0165, 0.0967, 0.0219, 0.0274, 0.2341], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0124, 0.0126, 0.0131, 0.0319, 0.0123, 0.0115, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 02:00:59,410 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35406.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:01:12,410 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35414.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 02:01:18,490 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.648e+02 2.480e+02 3.195e+02 3.962e+02 8.803e+02, threshold=6.389e+02, percent-clipped=2.0 2023-03-08 02:01:21,703 INFO [train2.py:809] (0/4) Epoch 9, batch 3550, loss[ctc_loss=0.1183, att_loss=0.2681, loss=0.2381, over 17286.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01189, over 55.00 utterances.], tot_loss[ctc_loss=0.1149, att_loss=0.2559, loss=0.2277, over 3261295.70 frames. utt_duration=1205 frames, utt_pad_proportion=0.0669, over 10841.83 utterances.], batch size: 55, lr: 1.18e-02, grad_scale: 8.0 2023-03-08 02:01:46,247 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35435.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:02:06,747 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35448.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:02:41,168 INFO [train2.py:809] (0/4) Epoch 9, batch 3600, loss[ctc_loss=0.08654, att_loss=0.2183, loss=0.192, over 15785.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.007725, over 38.00 utterances.], tot_loss[ctc_loss=0.1162, att_loss=0.2564, loss=0.2283, over 3266291.57 frames. utt_duration=1228 frames, utt_pad_proportion=0.0608, over 10654.84 utterances.], batch size: 38, lr: 1.18e-02, grad_scale: 8.0 2023-03-08 02:03:58,103 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.720e+02 2.714e+02 3.545e+02 4.370e+02 8.238e+02, threshold=7.091e+02, percent-clipped=8.0 2023-03-08 02:04:01,307 INFO [train2.py:809] (0/4) Epoch 9, batch 3650, loss[ctc_loss=0.0944, att_loss=0.2511, loss=0.2197, over 16389.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007606, over 44.00 utterances.], tot_loss[ctc_loss=0.1153, att_loss=0.2563, loss=0.2281, over 3266432.33 frames. utt_duration=1246 frames, utt_pad_proportion=0.05642, over 10498.90 utterances.], batch size: 44, lr: 1.18e-02, grad_scale: 8.0 2023-03-08 02:04:18,642 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35530.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:05:20,941 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-08 02:05:22,721 INFO [train2.py:809] (0/4) Epoch 9, batch 3700, loss[ctc_loss=0.1127, att_loss=0.2434, loss=0.2173, over 16124.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006279, over 42.00 utterances.], tot_loss[ctc_loss=0.1159, att_loss=0.2575, loss=0.2291, over 3280677.07 frames. utt_duration=1238 frames, utt_pad_proportion=0.05411, over 10609.06 utterances.], batch size: 42, lr: 1.18e-02, grad_scale: 8.0 2023-03-08 02:05:57,177 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35591.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:06:40,036 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.486e+02 3.166e+02 4.039e+02 7.024e+02, threshold=6.332e+02, percent-clipped=0.0 2023-03-08 02:06:43,372 INFO [train2.py:809] (0/4) Epoch 9, batch 3750, loss[ctc_loss=0.1506, att_loss=0.2841, loss=0.2574, over 17185.00 frames. utt_duration=695.7 frames, utt_pad_proportion=0.1271, over 99.00 utterances.], tot_loss[ctc_loss=0.1148, att_loss=0.2569, loss=0.2284, over 3286844.54 frames. utt_duration=1261 frames, utt_pad_proportion=0.04727, over 10438.69 utterances.], batch size: 99, lr: 1.18e-02, grad_scale: 16.0 2023-03-08 02:08:04,805 INFO [train2.py:809] (0/4) Epoch 9, batch 3800, loss[ctc_loss=0.1104, att_loss=0.2304, loss=0.2064, over 15503.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008594, over 36.00 utterances.], tot_loss[ctc_loss=0.114, att_loss=0.2563, loss=0.2279, over 3281108.46 frames. utt_duration=1259 frames, utt_pad_proportion=0.04952, over 10438.61 utterances.], batch size: 36, lr: 1.18e-02, grad_scale: 16.0 2023-03-08 02:08:56,046 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35701.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:09:23,353 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.475e+02 3.126e+02 3.794e+02 8.902e+02, threshold=6.253e+02, percent-clipped=4.0 2023-03-08 02:09:26,474 INFO [train2.py:809] (0/4) Epoch 9, batch 3850, loss[ctc_loss=0.09377, att_loss=0.2313, loss=0.2038, over 15649.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008043, over 37.00 utterances.], tot_loss[ctc_loss=0.1125, att_loss=0.255, loss=0.2265, over 3279401.19 frames. utt_duration=1275 frames, utt_pad_proportion=0.04694, over 10297.35 utterances.], batch size: 37, lr: 1.18e-02, grad_scale: 16.0 2023-03-08 02:09:39,274 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3543, 5.2269, 5.1645, 2.6915, 2.1021, 2.8060, 3.9376, 3.8488], device='cuda:0'), covar=tensor([0.0567, 0.0365, 0.0243, 0.3803, 0.6429, 0.2879, 0.1322, 0.2076], device='cuda:0'), in_proj_covar=tensor([0.0331, 0.0219, 0.0234, 0.0198, 0.0363, 0.0340, 0.0231, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.5429e-04, 8.3843e-05, 1.0180e-04, 9.0270e-05, 1.6017e-04, 1.4054e-04, 9.1561e-05, 1.5313e-04], device='cuda:0') 2023-03-08 02:09:42,215 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35730.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:10:10,698 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=35748.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:10:45,204 INFO [train2.py:809] (0/4) Epoch 9, batch 3900, loss[ctc_loss=0.1232, att_loss=0.2732, loss=0.2432, over 16954.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.0075, over 50.00 utterances.], tot_loss[ctc_loss=0.1131, att_loss=0.2554, loss=0.2269, over 3278730.82 frames. utt_duration=1264 frames, utt_pad_proportion=0.04865, over 10384.74 utterances.], batch size: 50, lr: 1.18e-02, grad_scale: 16.0 2023-03-08 02:10:47,081 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35771.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:11:26,108 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=35796.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:11:50,415 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=35811.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:12:00,555 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.615e+02 2.688e+02 3.135e+02 3.909e+02 9.741e+02, threshold=6.269e+02, percent-clipped=3.0 2023-03-08 02:12:03,697 INFO [train2.py:809] (0/4) Epoch 9, batch 3950, loss[ctc_loss=0.1295, att_loss=0.2638, loss=0.2369, over 17030.00 frames. utt_duration=689.4 frames, utt_pad_proportion=0.135, over 99.00 utterances.], tot_loss[ctc_loss=0.1135, att_loss=0.2558, loss=0.2273, over 3280665.81 frames. utt_duration=1248 frames, utt_pad_proportion=0.05064, over 10527.03 utterances.], batch size: 99, lr: 1.18e-02, grad_scale: 16.0 2023-03-08 02:12:22,912 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35832.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:12:56,114 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-9.pt 2023-03-08 02:13:22,877 INFO [train2.py:809] (0/4) Epoch 10, batch 0, loss[ctc_loss=0.1142, att_loss=0.2613, loss=0.2319, over 16616.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005797, over 47.00 utterances.], tot_loss[ctc_loss=0.1142, att_loss=0.2613, loss=0.2319, over 16616.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005797, over 47.00 utterances.], batch size: 47, lr: 1.12e-02, grad_scale: 16.0 2023-03-08 02:13:22,879 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 02:13:35,373 INFO [train2.py:843] (0/4) Epoch 10, validation: ctc_loss=0.0538, att_loss=0.2416, loss=0.204, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 02:13:35,374 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16076MB 2023-03-08 02:14:04,740 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=35872.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:14:25,896 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=35886.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:14:44,559 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2255, 4.7266, 4.7679, 4.8890, 2.7156, 4.9944, 2.9148, 2.0162], device='cuda:0'), covar=tensor([0.0314, 0.0146, 0.0550, 0.0125, 0.1792, 0.0111, 0.1402, 0.1666], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0108, 0.0253, 0.0108, 0.0221, 0.0101, 0.0224, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 02:14:54,634 INFO [train2.py:809] (0/4) Epoch 10, batch 50, loss[ctc_loss=0.1047, att_loss=0.2549, loss=0.2249, over 16409.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006966, over 44.00 utterances.], tot_loss[ctc_loss=0.1175, att_loss=0.2586, loss=0.2304, over 736488.32 frames. utt_duration=1118 frames, utt_pad_proportion=0.08759, over 2637.62 utterances.], batch size: 44, lr: 1.12e-02, grad_scale: 16.0 2023-03-08 02:15:18,098 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.853e+02 2.508e+02 2.998e+02 4.058e+02 7.580e+02, threshold=5.995e+02, percent-clipped=4.0 2023-03-08 02:16:05,785 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2394, 2.5218, 3.2832, 4.0834, 3.6442, 3.7143, 2.5143, 1.8276], device='cuda:0'), covar=tensor([0.0708, 0.2552, 0.0954, 0.0639, 0.0819, 0.0472, 0.1922, 0.2931], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0201, 0.0183, 0.0180, 0.0172, 0.0141, 0.0189, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 02:16:15,454 INFO [train2.py:809] (0/4) Epoch 10, batch 100, loss[ctc_loss=0.1346, att_loss=0.2645, loss=0.2385, over 17334.00 frames. utt_duration=1177 frames, utt_pad_proportion=0.02184, over 59.00 utterances.], tot_loss[ctc_loss=0.1159, att_loss=0.2576, loss=0.2292, over 1299953.53 frames. utt_duration=1118 frames, utt_pad_proportion=0.08704, over 4656.17 utterances.], batch size: 59, lr: 1.12e-02, grad_scale: 16.0 2023-03-08 02:17:30,970 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-36000.pt 2023-03-08 02:17:36,774 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36001.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:17:41,613 INFO [train2.py:809] (0/4) Epoch 10, batch 150, loss[ctc_loss=0.09326, att_loss=0.258, loss=0.225, over 17003.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.008143, over 51.00 utterances.], tot_loss[ctc_loss=0.1155, att_loss=0.2571, loss=0.2288, over 1738368.98 frames. utt_duration=1137 frames, utt_pad_proportion=0.08345, over 6125.06 utterances.], batch size: 51, lr: 1.12e-02, grad_scale: 16.0 2023-03-08 02:18:03,932 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.601e+02 3.072e+02 3.945e+02 7.612e+02, threshold=6.144e+02, percent-clipped=5.0 2023-03-08 02:18:22,403 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36030.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:18:50,008 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-03-08 02:18:52,118 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36049.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:19:00,381 INFO [train2.py:809] (0/4) Epoch 10, batch 200, loss[ctc_loss=0.08923, att_loss=0.2367, loss=0.2072, over 16144.00 frames. utt_duration=1539 frames, utt_pad_proportion=0.005171, over 42.00 utterances.], tot_loss[ctc_loss=0.1156, att_loss=0.2574, loss=0.2291, over 2079604.34 frames. utt_duration=1178 frames, utt_pad_proportion=0.07156, over 7071.01 utterances.], batch size: 42, lr: 1.12e-02, grad_scale: 16.0 2023-03-08 02:19:38,215 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36078.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:19:56,843 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5632, 2.6694, 3.3601, 4.4387, 4.2151, 3.8831, 2.9621, 1.9199], device='cuda:0'), covar=tensor([0.0621, 0.2423, 0.1115, 0.0514, 0.0545, 0.0446, 0.1532, 0.2712], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0201, 0.0184, 0.0183, 0.0171, 0.0141, 0.0188, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 02:20:11,266 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9082, 5.2706, 5.2049, 5.0588, 5.2459, 5.2521, 4.9962, 4.6875], device='cuda:0'), covar=tensor([0.1074, 0.0417, 0.0247, 0.0451, 0.0261, 0.0303, 0.0280, 0.0325], device='cuda:0'), in_proj_covar=tensor([0.0430, 0.0275, 0.0221, 0.0262, 0.0324, 0.0342, 0.0268, 0.0298], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 02:20:19,272 INFO [train2.py:809] (0/4) Epoch 10, batch 250, loss[ctc_loss=0.1526, att_loss=0.2808, loss=0.2552, over 14488.00 frames. utt_duration=398.5 frames, utt_pad_proportion=0.3058, over 146.00 utterances.], tot_loss[ctc_loss=0.1148, att_loss=0.256, loss=0.2278, over 2335912.75 frames. utt_duration=1204 frames, utt_pad_proportion=0.0679, over 7768.88 utterances.], batch size: 146, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:20:42,219 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.582e+02 2.693e+02 3.162e+02 4.102e+02 9.052e+02, threshold=6.324e+02, percent-clipped=7.0 2023-03-08 02:20:56,469 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36127.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:20:59,739 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5917, 2.8842, 3.6999, 3.0323, 3.6436, 4.6844, 4.4935, 3.5013], device='cuda:0'), covar=tensor([0.0393, 0.1606, 0.1014, 0.1198, 0.0947, 0.0626, 0.0471, 0.1103], device='cuda:0'), in_proj_covar=tensor([0.0224, 0.0220, 0.0236, 0.0200, 0.0232, 0.0280, 0.0208, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0003, 0.0003, 0.0002], device='cuda:0') 2023-03-08 02:21:39,368 INFO [train2.py:809] (0/4) Epoch 10, batch 300, loss[ctc_loss=0.1012, att_loss=0.2509, loss=0.2209, over 16273.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.00763, over 43.00 utterances.], tot_loss[ctc_loss=0.1123, att_loss=0.2545, loss=0.226, over 2547202.67 frames. utt_duration=1252 frames, utt_pad_proportion=0.05481, over 8149.81 utterances.], batch size: 43, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:21:50,372 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6739, 4.8903, 4.3889, 4.7966, 4.4382, 4.1616, 4.4578, 4.1981], device='cuda:0'), covar=tensor([0.1251, 0.1093, 0.0789, 0.0881, 0.1199, 0.1479, 0.2071, 0.2728], device='cuda:0'), in_proj_covar=tensor([0.0418, 0.0489, 0.0363, 0.0377, 0.0355, 0.0413, 0.0498, 0.0448], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 02:22:01,527 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36167.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:22:23,327 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36181.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:22:30,995 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36186.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:22:40,495 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1208, 4.1951, 4.0577, 4.2145, 4.4966, 4.2211, 4.1607, 2.3230], device='cuda:0'), covar=tensor([0.0307, 0.0359, 0.0308, 0.0156, 0.0972, 0.0246, 0.0277, 0.2082], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0123, 0.0128, 0.0132, 0.0317, 0.0121, 0.0115, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 02:23:01,514 INFO [train2.py:809] (0/4) Epoch 10, batch 350, loss[ctc_loss=0.09196, att_loss=0.2406, loss=0.2108, over 16165.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007087, over 41.00 utterances.], tot_loss[ctc_loss=0.1121, att_loss=0.2536, loss=0.2253, over 2702842.82 frames. utt_duration=1249 frames, utt_pad_proportion=0.05662, over 8665.03 utterances.], batch size: 41, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:23:24,551 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.621e+02 3.186e+02 4.132e+02 7.508e+02, threshold=6.371e+02, percent-clipped=3.0 2023-03-08 02:23:50,166 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36234.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:23:58,515 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2250, 4.5425, 4.4435, 4.6790, 2.3575, 4.6109, 2.3262, 2.2635], device='cuda:0'), covar=tensor([0.0270, 0.0135, 0.0793, 0.0152, 0.2467, 0.0190, 0.2167, 0.1829], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0109, 0.0255, 0.0110, 0.0224, 0.0105, 0.0228, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 02:24:03,256 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36242.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:24:24,485 INFO [train2.py:809] (0/4) Epoch 10, batch 400, loss[ctc_loss=0.1301, att_loss=0.2793, loss=0.2494, over 17430.00 frames. utt_duration=884 frames, utt_pad_proportion=0.07432, over 79.00 utterances.], tot_loss[ctc_loss=0.1131, att_loss=0.2551, loss=0.2267, over 2827135.54 frames. utt_duration=1183 frames, utt_pad_proportion=0.07371, over 9573.35 utterances.], batch size: 79, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:24:24,766 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5862, 5.0810, 4.7709, 5.0844, 5.1571, 4.7257, 3.3856, 5.0932], device='cuda:0'), covar=tensor([0.0110, 0.0102, 0.0109, 0.0082, 0.0113, 0.0108, 0.0654, 0.0174], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0068, 0.0079, 0.0050, 0.0054, 0.0064, 0.0086, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 02:24:27,142 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-08 02:25:35,568 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5549, 2.5349, 4.9641, 3.9261, 2.8363, 4.5332, 4.9473, 4.6519], device='cuda:0'), covar=tensor([0.0233, 0.1568, 0.0229, 0.0926, 0.2041, 0.0196, 0.0097, 0.0233], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0245, 0.0131, 0.0303, 0.0280, 0.0184, 0.0111, 0.0146], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0003, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-03-08 02:25:45,286 INFO [train2.py:809] (0/4) Epoch 10, batch 450, loss[ctc_loss=0.1203, att_loss=0.2499, loss=0.224, over 16176.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006412, over 41.00 utterances.], tot_loss[ctc_loss=0.112, att_loss=0.2546, loss=0.2261, over 2925092.01 frames. utt_duration=1203 frames, utt_pad_proportion=0.06837, over 9739.22 utterances.], batch size: 41, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:26:06,977 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.425e+02 2.409e+02 2.979e+02 3.960e+02 1.144e+03, threshold=5.958e+02, percent-clipped=2.0 2023-03-08 02:26:46,025 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7503, 2.3696, 1.9347, 1.2749, 2.7775, 2.1914, 1.5953, 2.7345], device='cuda:0'), covar=tensor([0.1023, 0.2642, 0.3732, 0.2589, 0.1109, 0.1157, 0.2950, 0.1285], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0082, 0.0090, 0.0076, 0.0073, 0.0068, 0.0082, 0.0063], device='cuda:0'), out_proj_covar=tensor([4.5232e-05, 5.4681e-05, 5.7971e-05, 4.8335e-05, 4.4467e-05, 4.6345e-05, 5.4129e-05, 4.3871e-05], device='cuda:0') 2023-03-08 02:27:03,694 INFO [train2.py:809] (0/4) Epoch 10, batch 500, loss[ctc_loss=0.09323, att_loss=0.2365, loss=0.2079, over 15963.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006004, over 41.00 utterances.], tot_loss[ctc_loss=0.1129, att_loss=0.2549, loss=0.2265, over 2998517.58 frames. utt_duration=1211 frames, utt_pad_proportion=0.067, over 9914.99 utterances.], batch size: 41, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:27:14,112 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7714, 6.0274, 5.5208, 5.8195, 5.6837, 5.2526, 5.4167, 5.2175], device='cuda:0'), covar=tensor([0.1104, 0.0771, 0.0703, 0.0716, 0.0745, 0.1285, 0.1925, 0.2193], device='cuda:0'), in_proj_covar=tensor([0.0420, 0.0481, 0.0361, 0.0376, 0.0353, 0.0409, 0.0497, 0.0445], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 02:27:51,244 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-08 02:28:22,568 INFO [train2.py:809] (0/4) Epoch 10, batch 550, loss[ctc_loss=0.09392, att_loss=0.2465, loss=0.216, over 16329.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006323, over 45.00 utterances.], tot_loss[ctc_loss=0.1126, att_loss=0.255, loss=0.2265, over 3068776.77 frames. utt_duration=1214 frames, utt_pad_proportion=0.06215, over 10121.07 utterances.], batch size: 45, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:28:45,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.595e+02 2.390e+02 3.015e+02 3.600e+02 8.483e+02, threshold=6.031e+02, percent-clipped=3.0 2023-03-08 02:28:59,584 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36427.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:29:41,950 INFO [train2.py:809] (0/4) Epoch 10, batch 600, loss[ctc_loss=0.105, att_loss=0.2438, loss=0.216, over 16020.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006506, over 40.00 utterances.], tot_loss[ctc_loss=0.1118, att_loss=0.2541, loss=0.2256, over 3116562.81 frames. utt_duration=1239 frames, utt_pad_proportion=0.05454, over 10074.63 utterances.], batch size: 40, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:30:03,547 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36467.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:30:15,894 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36475.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:30:56,102 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0373, 4.8400, 4.8272, 2.5011, 4.7266, 4.5258, 4.0625, 2.4214], device='cuda:0'), covar=tensor([0.0115, 0.0097, 0.0222, 0.1158, 0.0085, 0.0188, 0.0341, 0.1431], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0077, 0.0066, 0.0099, 0.0064, 0.0089, 0.0087, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 02:31:02,500 INFO [train2.py:809] (0/4) Epoch 10, batch 650, loss[ctc_loss=0.1555, att_loss=0.2978, loss=0.2693, over 16784.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005445, over 48.00 utterances.], tot_loss[ctc_loss=0.1119, att_loss=0.2547, loss=0.2261, over 3154829.87 frames. utt_duration=1250 frames, utt_pad_proportion=0.05142, over 10104.88 utterances.], batch size: 48, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:31:20,043 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36515.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:31:24,437 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.680e+02 2.550e+02 3.093e+02 3.802e+02 8.175e+02, threshold=6.187e+02, percent-clipped=1.0 2023-03-08 02:31:53,777 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36537.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:32:21,150 INFO [train2.py:809] (0/4) Epoch 10, batch 700, loss[ctc_loss=0.1305, att_loss=0.2672, loss=0.2399, over 16178.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006519, over 41.00 utterances.], tot_loss[ctc_loss=0.1124, att_loss=0.2555, loss=0.2269, over 3185649.59 frames. utt_duration=1272 frames, utt_pad_proportion=0.04512, over 10028.44 utterances.], batch size: 41, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:32:24,440 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0199, 5.3116, 4.7739, 5.4023, 4.7564, 5.0254, 5.4500, 5.2147], device='cuda:0'), covar=tensor([0.0454, 0.0316, 0.0814, 0.0220, 0.0370, 0.0200, 0.0217, 0.0162], device='cuda:0'), in_proj_covar=tensor([0.0311, 0.0245, 0.0307, 0.0233, 0.0246, 0.0193, 0.0227, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 02:32:24,636 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8005, 4.7260, 4.7274, 4.6951, 5.2653, 4.9728, 4.7355, 2.2171], device='cuda:0'), covar=tensor([0.0210, 0.0243, 0.0173, 0.0193, 0.0752, 0.0155, 0.0195, 0.2371], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0125, 0.0128, 0.0132, 0.0320, 0.0120, 0.0115, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 02:32:55,416 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1705, 5.2102, 5.1221, 2.3353, 1.9673, 2.8782, 3.1548, 3.6989], device='cuda:0'), covar=tensor([0.0598, 0.0215, 0.0211, 0.4498, 0.6028, 0.2515, 0.1824, 0.2008], device='cuda:0'), in_proj_covar=tensor([0.0333, 0.0217, 0.0232, 0.0199, 0.0359, 0.0339, 0.0232, 0.0353], device='cuda:0'), out_proj_covar=tensor([1.5424e-04, 8.2411e-05, 9.9978e-05, 9.0487e-05, 1.5834e-04, 1.3980e-04, 9.1802e-05, 1.5187e-04], device='cuda:0') 2023-03-08 02:33:01,526 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9364, 4.4007, 4.1899, 4.3541, 4.3681, 4.0763, 2.9799, 4.2417], device='cuda:0'), covar=tensor([0.0135, 0.0111, 0.0122, 0.0084, 0.0118, 0.0128, 0.0738, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0068, 0.0079, 0.0050, 0.0054, 0.0064, 0.0087, 0.0087], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 02:33:41,079 INFO [train2.py:809] (0/4) Epoch 10, batch 750, loss[ctc_loss=0.07495, att_loss=0.2308, loss=0.1997, over 16329.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006414, over 45.00 utterances.], tot_loss[ctc_loss=0.1109, att_loss=0.2546, loss=0.2259, over 3207429.52 frames. utt_duration=1276 frames, utt_pad_proportion=0.04345, over 10067.59 utterances.], batch size: 45, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:33:43,628 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36605.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:33:48,467 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36608.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:33:55,262 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36612.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:33:59,927 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8781, 4.7640, 4.7871, 4.6529, 5.2714, 5.0266, 4.8109, 2.2208], device='cuda:0'), covar=tensor([0.0152, 0.0274, 0.0184, 0.0231, 0.0986, 0.0119, 0.0206, 0.2255], device='cuda:0'), in_proj_covar=tensor([0.0128, 0.0126, 0.0130, 0.0133, 0.0324, 0.0121, 0.0117, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 02:34:04,092 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.731e+02 2.573e+02 3.133e+02 3.667e+02 8.255e+02, threshold=6.267e+02, percent-clipped=5.0 2023-03-08 02:35:01,774 INFO [train2.py:809] (0/4) Epoch 10, batch 800, loss[ctc_loss=0.1461, att_loss=0.281, loss=0.254, over 16696.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006037, over 46.00 utterances.], tot_loss[ctc_loss=0.1118, att_loss=0.2555, loss=0.2268, over 3234659.15 frames. utt_duration=1282 frames, utt_pad_proportion=0.03922, over 10106.92 utterances.], batch size: 46, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:35:21,812 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36666.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:35:26,441 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36669.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:35:32,474 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36673.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:36:07,757 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.18 vs. limit=5.0 2023-03-08 02:36:15,424 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36700.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:36:17,636 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 02:36:21,095 INFO [train2.py:809] (0/4) Epoch 10, batch 850, loss[ctc_loss=0.1101, att_loss=0.2619, loss=0.2316, over 17050.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008323, over 52.00 utterances.], tot_loss[ctc_loss=0.1121, att_loss=0.2557, loss=0.227, over 3242965.51 frames. utt_duration=1249 frames, utt_pad_proportion=0.04807, over 10396.56 utterances.], batch size: 52, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:36:43,991 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.977e+02 2.594e+02 3.302e+02 4.019e+02 1.127e+03, threshold=6.604e+02, percent-clipped=9.0 2023-03-08 02:36:55,523 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 02:37:34,269 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-03-08 02:37:41,456 INFO [train2.py:809] (0/4) Epoch 10, batch 900, loss[ctc_loss=0.1304, att_loss=0.2765, loss=0.2473, over 17496.00 frames. utt_duration=887.5 frames, utt_pad_proportion=0.0707, over 79.00 utterances.], tot_loss[ctc_loss=0.1124, att_loss=0.2558, loss=0.2271, over 3250201.01 frames. utt_duration=1245 frames, utt_pad_proportion=0.04874, over 10450.92 utterances.], batch size: 79, lr: 1.11e-02, grad_scale: 16.0 2023-03-08 02:37:53,185 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36761.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 02:38:04,138 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7463, 5.9717, 5.3039, 5.8297, 5.6631, 5.1686, 5.4024, 5.1888], device='cuda:0'), covar=tensor([0.1116, 0.0825, 0.0910, 0.0667, 0.0760, 0.1435, 0.2249, 0.2345], device='cuda:0'), in_proj_covar=tensor([0.0424, 0.0495, 0.0369, 0.0381, 0.0353, 0.0412, 0.0505, 0.0449], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 02:38:21,571 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0382, 3.7259, 3.0931, 3.3796, 3.9312, 3.5730, 2.7887, 4.2554], device='cuda:0'), covar=tensor([0.0901, 0.0445, 0.1126, 0.0702, 0.0613, 0.0686, 0.0908, 0.0395], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0173, 0.0201, 0.0170, 0.0222, 0.0206, 0.0179, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 02:39:01,057 INFO [train2.py:809] (0/4) Epoch 10, batch 950, loss[ctc_loss=0.1148, att_loss=0.2597, loss=0.2307, over 17402.00 frames. utt_duration=882.8 frames, utt_pad_proportion=0.07661, over 79.00 utterances.], tot_loss[ctc_loss=0.1108, att_loss=0.2545, loss=0.2258, over 3252023.70 frames. utt_duration=1280 frames, utt_pad_proportion=0.04265, over 10173.44 utterances.], batch size: 79, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:39:03,633 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36805.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 02:39:23,925 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.872e+02 2.661e+02 3.552e+02 4.499e+02 1.125e+03, threshold=7.103e+02, percent-clipped=5.0 2023-03-08 02:39:54,035 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=36837.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:39:58,602 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6844, 5.8478, 5.1319, 5.6838, 5.5174, 5.0646, 5.3006, 4.9892], device='cuda:0'), covar=tensor([0.1086, 0.0903, 0.0989, 0.0781, 0.0768, 0.1401, 0.2123, 0.2301], device='cuda:0'), in_proj_covar=tensor([0.0418, 0.0489, 0.0365, 0.0376, 0.0349, 0.0406, 0.0499, 0.0445], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 02:40:22,145 INFO [train2.py:809] (0/4) Epoch 10, batch 1000, loss[ctc_loss=0.1083, att_loss=0.2361, loss=0.2105, over 15959.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.005926, over 41.00 utterances.], tot_loss[ctc_loss=0.1108, att_loss=0.2546, loss=0.2258, over 3260686.99 frames. utt_duration=1284 frames, utt_pad_proportion=0.04134, over 10171.10 utterances.], batch size: 41, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:40:41,846 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36866.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:41:11,563 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=36885.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:41:11,937 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9839, 4.9926, 4.9025, 2.2685, 1.9060, 2.7577, 3.2501, 3.8181], device='cuda:0'), covar=tensor([0.0745, 0.0187, 0.0218, 0.3881, 0.5955, 0.2527, 0.1636, 0.1628], device='cuda:0'), in_proj_covar=tensor([0.0330, 0.0217, 0.0232, 0.0199, 0.0353, 0.0335, 0.0227, 0.0349], device='cuda:0'), out_proj_covar=tensor([1.5268e-04, 8.3101e-05, 1.0032e-04, 9.0022e-05, 1.5600e-04, 1.3783e-04, 8.9691e-05, 1.5019e-04], device='cuda:0') 2023-03-08 02:41:30,430 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6835, 5.9673, 5.3707, 5.8373, 5.6484, 5.3207, 5.3823, 5.1858], device='cuda:0'), covar=tensor([0.1221, 0.0881, 0.0898, 0.0612, 0.0658, 0.1404, 0.2365, 0.2185], device='cuda:0'), in_proj_covar=tensor([0.0414, 0.0484, 0.0362, 0.0371, 0.0346, 0.0401, 0.0492, 0.0438], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 02:41:43,594 INFO [train2.py:809] (0/4) Epoch 10, batch 1050, loss[ctc_loss=0.1298, att_loss=0.2675, loss=0.24, over 17131.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01439, over 56.00 utterances.], tot_loss[ctc_loss=0.1106, att_loss=0.255, loss=0.2262, over 3271051.52 frames. utt_duration=1268 frames, utt_pad_proportion=0.04455, over 10326.89 utterances.], batch size: 56, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:41:55,874 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36911.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:41:57,465 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9422, 1.6204, 2.3510, 2.3864, 3.2486, 2.1573, 1.5733, 2.2347], device='cuda:0'), covar=tensor([0.1002, 0.4245, 0.3802, 0.1443, 0.3031, 0.1465, 0.3478, 0.1807], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0084, 0.0090, 0.0075, 0.0074, 0.0070, 0.0083, 0.0065], device='cuda:0'), out_proj_covar=tensor([4.6149e-05, 5.5657e-05, 5.8095e-05, 4.8293e-05, 4.5284e-05, 4.7474e-05, 5.4689e-05, 4.4942e-05], device='cuda:0') 2023-03-08 02:42:00,549 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0643, 5.3936, 5.3954, 5.3468, 5.5127, 5.4723, 5.2517, 4.9591], device='cuda:0'), covar=tensor([0.1043, 0.0452, 0.0205, 0.0310, 0.0264, 0.0269, 0.0273, 0.0278], device='cuda:0'), in_proj_covar=tensor([0.0438, 0.0281, 0.0227, 0.0262, 0.0329, 0.0351, 0.0278, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 02:42:06,546 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.544e+02 2.981e+02 3.460e+02 1.238e+03, threshold=5.962e+02, percent-clipped=1.0 2023-03-08 02:42:40,958 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.98 vs. limit=2.0 2023-03-08 02:42:57,404 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-03-08 02:43:05,652 INFO [train2.py:809] (0/4) Epoch 10, batch 1100, loss[ctc_loss=0.09738, att_loss=0.2547, loss=0.2232, over 16875.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007185, over 49.00 utterances.], tot_loss[ctc_loss=0.1107, att_loss=0.2548, loss=0.226, over 3272354.94 frames. utt_duration=1264 frames, utt_pad_proportion=0.04722, over 10369.41 utterances.], batch size: 49, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:43:12,922 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=36958.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:43:17,515 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36961.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:43:22,066 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36964.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:43:28,262 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=36968.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:43:29,773 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7017, 5.1967, 5.0262, 5.0181, 5.1271, 4.9361, 3.2939, 4.9551], device='cuda:0'), covar=tensor([0.0106, 0.0119, 0.0099, 0.0096, 0.0093, 0.0088, 0.0782, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0070, 0.0082, 0.0051, 0.0055, 0.0065, 0.0089, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 02:43:34,583 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=36972.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:44:26,437 INFO [train2.py:809] (0/4) Epoch 10, batch 1150, loss[ctc_loss=0.1, att_loss=0.2313, loss=0.205, over 14531.00 frames. utt_duration=1818 frames, utt_pad_proportion=0.04515, over 32.00 utterances.], tot_loss[ctc_loss=0.1097, att_loss=0.2535, loss=0.2248, over 3274469.39 frames. utt_duration=1267 frames, utt_pad_proportion=0.04641, over 10352.97 utterances.], batch size: 32, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:44:26,788 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37004.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:44:33,859 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6074, 2.9225, 3.7149, 3.0904, 3.4533, 4.7181, 4.4365, 3.4872], device='cuda:0'), covar=tensor([0.0400, 0.1877, 0.1144, 0.1398, 0.1174, 0.0757, 0.0681, 0.1265], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0227, 0.0242, 0.0201, 0.0239, 0.0290, 0.0212, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 02:44:48,824 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.581e+02 2.521e+02 2.891e+02 3.394e+02 5.463e+02, threshold=5.783e+02, percent-clipped=0.0 2023-03-08 02:44:50,897 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37019.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:45:46,918 INFO [train2.py:809] (0/4) Epoch 10, batch 1200, loss[ctc_loss=0.0898, att_loss=0.2454, loss=0.2143, over 15943.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007264, over 41.00 utterances.], tot_loss[ctc_loss=0.1094, att_loss=0.2532, loss=0.2244, over 3268854.62 frames. utt_duration=1272 frames, utt_pad_proportion=0.04799, over 10288.97 utterances.], batch size: 41, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:45:50,263 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37056.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 02:46:04,748 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37065.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:46:05,578 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 02:46:43,635 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1421, 2.4241, 3.3946, 2.6609, 3.3319, 4.3570, 4.0189, 2.9448], device='cuda:0'), covar=tensor([0.0466, 0.1976, 0.1232, 0.1405, 0.1093, 0.0749, 0.0768, 0.1487], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0225, 0.0243, 0.0202, 0.0239, 0.0290, 0.0213, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 02:47:06,732 INFO [train2.py:809] (0/4) Epoch 10, batch 1250, loss[ctc_loss=0.0905, att_loss=0.2267, loss=0.1994, over 15493.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.009249, over 36.00 utterances.], tot_loss[ctc_loss=0.1095, att_loss=0.2532, loss=0.2245, over 3263100.54 frames. utt_duration=1280 frames, utt_pad_proportion=0.04729, over 10212.32 utterances.], batch size: 36, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:47:29,342 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.464e+02 2.541e+02 3.066e+02 3.675e+02 8.173e+02, threshold=6.132e+02, percent-clipped=8.0 2023-03-08 02:48:28,601 INFO [train2.py:809] (0/4) Epoch 10, batch 1300, loss[ctc_loss=0.1246, att_loss=0.2737, loss=0.2439, over 17405.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03081, over 63.00 utterances.], tot_loss[ctc_loss=0.1089, att_loss=0.2532, loss=0.2244, over 3272736.10 frames. utt_duration=1282 frames, utt_pad_proportion=0.04512, over 10225.66 utterances.], batch size: 63, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:48:40,429 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37161.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:48:58,028 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8649, 1.9276, 2.0377, 1.9958, 2.4885, 1.9057, 1.6699, 2.6812], device='cuda:0'), covar=tensor([0.0852, 0.2874, 0.3024, 0.1726, 0.1182, 0.1255, 0.2388, 0.0965], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0084, 0.0089, 0.0075, 0.0075, 0.0068, 0.0082, 0.0064], device='cuda:0'), out_proj_covar=tensor([4.6579e-05, 5.5614e-05, 5.8123e-05, 4.8465e-05, 4.5988e-05, 4.6901e-05, 5.4447e-05, 4.4798e-05], device='cuda:0') 2023-03-08 02:49:49,920 INFO [train2.py:809] (0/4) Epoch 10, batch 1350, loss[ctc_loss=0.1745, att_loss=0.2926, loss=0.269, over 14636.00 frames. utt_duration=402.5 frames, utt_pad_proportion=0.2999, over 146.00 utterances.], tot_loss[ctc_loss=0.1108, att_loss=0.2545, loss=0.2258, over 3271324.56 frames. utt_duration=1223 frames, utt_pad_proportion=0.05981, over 10714.05 utterances.], batch size: 146, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:50:12,584 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.658e+02 2.454e+02 3.108e+02 4.121e+02 7.457e+02, threshold=6.215e+02, percent-clipped=5.0 2023-03-08 02:51:04,521 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-03-08 02:51:11,265 INFO [train2.py:809] (0/4) Epoch 10, batch 1400, loss[ctc_loss=0.0999, att_loss=0.2503, loss=0.2202, over 16500.00 frames. utt_duration=1436 frames, utt_pad_proportion=0.005254, over 46.00 utterances.], tot_loss[ctc_loss=0.1102, att_loss=0.2544, loss=0.2256, over 3276495.34 frames. utt_duration=1238 frames, utt_pad_proportion=0.05542, over 10598.07 utterances.], batch size: 46, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:51:23,132 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37261.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:51:27,867 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37264.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:51:32,339 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37267.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:51:33,963 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:51:34,022 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37268.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:52:19,846 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0777, 4.4988, 4.2528, 4.7633, 2.3500, 4.6700, 2.3776, 1.5249], device='cuda:0'), covar=tensor([0.0270, 0.0179, 0.0845, 0.0148, 0.2126, 0.0135, 0.1892, 0.2070], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0106, 0.0252, 0.0109, 0.0218, 0.0102, 0.0226, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 02:52:31,798 INFO [train2.py:809] (0/4) Epoch 10, batch 1450, loss[ctc_loss=0.1067, att_loss=0.2565, loss=0.2266, over 16331.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006126, over 45.00 utterances.], tot_loss[ctc_loss=0.1102, att_loss=0.2541, loss=0.2253, over 3278912.96 frames. utt_duration=1250 frames, utt_pad_proportion=0.05201, over 10504.15 utterances.], batch size: 45, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:52:39,559 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37309.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:52:44,932 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37312.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:52:48,230 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37314.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:52:51,392 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37316.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:52:54,386 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.647e+02 2.516e+02 3.005e+02 3.782e+02 1.128e+03, threshold=6.010e+02, percent-clipped=1.0 2023-03-08 02:53:11,876 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37329.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:53:51,706 INFO [train2.py:809] (0/4) Epoch 10, batch 1500, loss[ctc_loss=0.1, att_loss=0.2601, loss=0.2281, over 16460.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.007005, over 46.00 utterances.], tot_loss[ctc_loss=0.1101, att_loss=0.2544, loss=0.2255, over 3273318.23 frames. utt_duration=1246 frames, utt_pad_proportion=0.05565, over 10524.25 utterances.], batch size: 46, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:53:55,186 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37356.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 02:54:02,004 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37360.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:54:08,242 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37364.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:54:48,351 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 02:55:12,294 INFO [train2.py:809] (0/4) Epoch 10, batch 1550, loss[ctc_loss=0.1041, att_loss=0.2664, loss=0.234, over 17359.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03372, over 63.00 utterances.], tot_loss[ctc_loss=0.1109, att_loss=0.255, loss=0.2262, over 3271706.17 frames. utt_duration=1238 frames, utt_pad_proportion=0.05717, over 10582.84 utterances.], batch size: 63, lr: 1.10e-02, grad_scale: 16.0 2023-03-08 02:55:12,375 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37404.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 02:55:34,597 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.660e+02 2.658e+02 3.084e+02 3.993e+02 7.629e+02, threshold=6.167e+02, percent-clipped=4.0 2023-03-08 02:55:45,943 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=37425.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 02:56:32,392 INFO [train2.py:809] (0/4) Epoch 10, batch 1600, loss[ctc_loss=0.1075, att_loss=0.2595, loss=0.2291, over 16897.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.006039, over 49.00 utterances.], tot_loss[ctc_loss=0.1102, att_loss=0.2553, loss=0.2263, over 3281187.63 frames. utt_duration=1229 frames, utt_pad_proportion=0.05721, over 10688.63 utterances.], batch size: 49, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 02:56:43,940 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37461.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 02:57:07,112 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0706, 5.0954, 5.0440, 2.7556, 4.8757, 4.5117, 4.3465, 2.3360], device='cuda:0'), covar=tensor([0.0131, 0.0078, 0.0154, 0.0988, 0.0078, 0.0185, 0.0294, 0.1521], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0079, 0.0069, 0.0101, 0.0067, 0.0090, 0.0089, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 02:57:53,646 INFO [train2.py:809] (0/4) Epoch 10, batch 1650, loss[ctc_loss=0.135, att_loss=0.2791, loss=0.2503, over 17143.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01363, over 56.00 utterances.], tot_loss[ctc_loss=0.1111, att_loss=0.2568, loss=0.2277, over 3296650.67 frames. utt_duration=1231 frames, utt_pad_proportion=0.05275, over 10729.28 utterances.], batch size: 56, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 02:58:01,289 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37509.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 02:58:14,778 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.538e+02 2.486e+02 2.891e+02 3.571e+02 1.016e+03, threshold=5.782e+02, percent-clipped=2.0 2023-03-08 02:59:12,769 INFO [train2.py:809] (0/4) Epoch 10, batch 1700, loss[ctc_loss=0.1491, att_loss=0.2793, loss=0.2533, over 14156.00 frames. utt_duration=389.3 frames, utt_pad_proportion=0.323, over 146.00 utterances.], tot_loss[ctc_loss=0.1107, att_loss=0.2562, loss=0.2271, over 3293755.50 frames. utt_duration=1241 frames, utt_pad_proportion=0.05093, over 10630.01 utterances.], batch size: 146, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 02:59:14,054 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-08 02:59:16,090 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0776, 5.3558, 4.8232, 5.4593, 4.8278, 5.0948, 5.5314, 5.3149], device='cuda:0'), covar=tensor([0.0417, 0.0252, 0.0801, 0.0182, 0.0369, 0.0174, 0.0161, 0.0139], device='cuda:0'), in_proj_covar=tensor([0.0313, 0.0246, 0.0308, 0.0238, 0.0251, 0.0195, 0.0230, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 02:59:33,295 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37567.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:00:33,883 INFO [train2.py:809] (0/4) Epoch 10, batch 1750, loss[ctc_loss=0.119, att_loss=0.2636, loss=0.2347, over 16956.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008093, over 50.00 utterances.], tot_loss[ctc_loss=0.1114, att_loss=0.2567, loss=0.2276, over 3293404.70 frames. utt_duration=1227 frames, utt_pad_proportion=0.0554, over 10746.19 utterances.], batch size: 50, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 03:00:49,692 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37614.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:00:51,062 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37615.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:00:55,646 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.432e+02 2.940e+02 3.870e+02 6.231e+02, threshold=5.879e+02, percent-clipped=2.0 2023-03-08 03:01:05,265 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37624.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:01:53,787 INFO [train2.py:809] (0/4) Epoch 10, batch 1800, loss[ctc_loss=0.08729, att_loss=0.2499, loss=0.2173, over 17414.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03297, over 63.00 utterances.], tot_loss[ctc_loss=0.11, att_loss=0.255, loss=0.226, over 3278410.82 frames. utt_duration=1207 frames, utt_pad_proportion=0.06397, over 10880.08 utterances.], batch size: 63, lr: 1.09e-02, grad_scale: 32.0 2023-03-08 03:02:03,003 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37660.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:02:05,722 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37662.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:03:07,959 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 03:03:13,561 INFO [train2.py:809] (0/4) Epoch 10, batch 1850, loss[ctc_loss=0.1005, att_loss=0.2367, loss=0.2095, over 15397.00 frames. utt_duration=1761 frames, utt_pad_proportion=0.008875, over 35.00 utterances.], tot_loss[ctc_loss=0.1098, att_loss=0.2542, loss=0.2253, over 3270912.01 frames. utt_duration=1210 frames, utt_pad_proportion=0.06566, over 10828.40 utterances.], batch size: 35, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 03:03:19,692 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37708.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:03:22,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-08 03:03:29,197 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1028, 4.9870, 4.9400, 2.3101, 2.0362, 2.8090, 2.7869, 3.8072], device='cuda:0'), covar=tensor([0.0655, 0.0178, 0.0189, 0.4818, 0.6069, 0.2670, 0.2332, 0.1694], device='cuda:0'), in_proj_covar=tensor([0.0333, 0.0217, 0.0234, 0.0200, 0.0353, 0.0334, 0.0229, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.5334e-04, 8.2783e-05, 1.0173e-04, 9.0986e-05, 1.5589e-04, 1.3679e-04, 9.0018e-05, 1.5143e-04], device='cuda:0') 2023-03-08 03:03:36,152 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.525e+02 2.414e+02 3.142e+02 4.365e+02 1.152e+03, threshold=6.284e+02, percent-clipped=6.0 2023-03-08 03:03:37,901 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=37720.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:04:33,113 INFO [train2.py:809] (0/4) Epoch 10, batch 1900, loss[ctc_loss=0.1898, att_loss=0.2997, loss=0.2777, over 13655.00 frames. utt_duration=378.1 frames, utt_pad_proportion=0.3424, over 145.00 utterances.], tot_loss[ctc_loss=0.1103, att_loss=0.2548, loss=0.2259, over 3268243.05 frames. utt_duration=1201 frames, utt_pad_proportion=0.06901, over 10900.91 utterances.], batch size: 145, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 03:05:53,429 INFO [train2.py:809] (0/4) Epoch 10, batch 1950, loss[ctc_loss=0.09679, att_loss=0.2358, loss=0.208, over 15931.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007669, over 41.00 utterances.], tot_loss[ctc_loss=0.11, att_loss=0.2544, loss=0.2256, over 3271297.67 frames. utt_duration=1219 frames, utt_pad_proportion=0.06455, over 10746.48 utterances.], batch size: 41, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 03:06:16,462 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.508e+02 3.059e+02 3.528e+02 6.330e+02, threshold=6.117e+02, percent-clipped=1.0 2023-03-08 03:07:12,869 INFO [train2.py:809] (0/4) Epoch 10, batch 2000, loss[ctc_loss=0.08094, att_loss=0.2357, loss=0.2047, over 16275.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007645, over 43.00 utterances.], tot_loss[ctc_loss=0.109, att_loss=0.2534, loss=0.2245, over 3268540.16 frames. utt_duration=1259 frames, utt_pad_proportion=0.05557, over 10400.83 utterances.], batch size: 43, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 03:07:43,147 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1701, 5.1767, 5.1108, 3.2286, 4.9401, 4.6709, 4.5478, 2.8856], device='cuda:0'), covar=tensor([0.0125, 0.0064, 0.0139, 0.0784, 0.0079, 0.0160, 0.0235, 0.1177], device='cuda:0'), in_proj_covar=tensor([0.0058, 0.0079, 0.0068, 0.0099, 0.0066, 0.0089, 0.0088, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 03:08:32,705 INFO [train2.py:809] (0/4) Epoch 10, batch 2050, loss[ctc_loss=0.1105, att_loss=0.2617, loss=0.2315, over 17081.00 frames. utt_duration=1222 frames, utt_pad_proportion=0.01704, over 56.00 utterances.], tot_loss[ctc_loss=0.109, att_loss=0.2544, loss=0.2253, over 3281918.56 frames. utt_duration=1264 frames, utt_pad_proportion=0.05051, over 10396.25 utterances.], batch size: 56, lr: 1.09e-02, grad_scale: 16.0 2023-03-08 03:08:34,475 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8403, 5.1059, 4.6062, 5.2576, 4.5356, 4.8601, 5.2667, 5.0517], device='cuda:0'), covar=tensor([0.0516, 0.0241, 0.0813, 0.0199, 0.0445, 0.0229, 0.0192, 0.0163], device='cuda:0'), in_proj_covar=tensor([0.0321, 0.0253, 0.0319, 0.0242, 0.0259, 0.0199, 0.0236, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 03:08:37,665 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5182, 2.6515, 3.5969, 2.9671, 3.3634, 4.6343, 4.2726, 3.2603], device='cuda:0'), covar=tensor([0.0387, 0.1852, 0.1142, 0.1308, 0.1205, 0.0665, 0.0716, 0.1303], device='cuda:0'), in_proj_covar=tensor([0.0229, 0.0228, 0.0243, 0.0207, 0.0238, 0.0294, 0.0215, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 03:08:39,260 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4660, 2.6161, 3.1552, 4.2704, 3.8622, 3.9067, 2.6507, 1.8401], device='cuda:0'), covar=tensor([0.0609, 0.2290, 0.1109, 0.0544, 0.0717, 0.0424, 0.1743, 0.2375], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0204, 0.0187, 0.0185, 0.0180, 0.0143, 0.0187, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 03:08:55,897 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 2.394e+02 2.856e+02 3.386e+02 7.118e+02, threshold=5.711e+02, percent-clipped=3.0 2023-03-08 03:09:04,088 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=37924.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:09:31,165 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2023-03-08 03:09:33,286 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-03-08 03:09:52,778 INFO [train2.py:809] (0/4) Epoch 10, batch 2100, loss[ctc_loss=0.09745, att_loss=0.2628, loss=0.2297, over 17427.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01275, over 57.00 utterances.], tot_loss[ctc_loss=0.11, att_loss=0.2552, loss=0.2262, over 3282965.48 frames. utt_duration=1238 frames, utt_pad_proportion=0.05741, over 10621.25 utterances.], batch size: 57, lr: 1.09e-02, grad_scale: 8.0 2023-03-08 03:10:21,670 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=37972.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:10:33,453 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=37979.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:11:06,245 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-38000.pt 2023-03-08 03:11:18,145 INFO [train2.py:809] (0/4) Epoch 10, batch 2150, loss[ctc_loss=0.09954, att_loss=0.2345, loss=0.2075, over 15990.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008256, over 40.00 utterances.], tot_loss[ctc_loss=0.1095, att_loss=0.2544, loss=0.2255, over 3277748.64 frames. utt_duration=1262 frames, utt_pad_proportion=0.05271, over 10398.62 utterances.], batch size: 40, lr: 1.09e-02, grad_scale: 8.0 2023-03-08 03:11:42,429 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.860e+02 3.437e+02 3.901e+02 7.219e+02, threshold=6.874e+02, percent-clipped=5.0 2023-03-08 03:11:42,742 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38020.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:12:13,860 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5948, 3.6972, 3.0264, 3.4211, 3.9299, 3.5201, 2.4375, 4.2843], device='cuda:0'), covar=tensor([0.1148, 0.0495, 0.1143, 0.0675, 0.0695, 0.0698, 0.1071, 0.0483], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0178, 0.0202, 0.0171, 0.0227, 0.0208, 0.0176, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 03:12:15,374 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38040.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:12:31,273 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5198, 2.6582, 3.0358, 4.3036, 3.8376, 4.0620, 2.7794, 1.7253], device='cuda:0'), covar=tensor([0.0551, 0.2271, 0.1150, 0.0587, 0.0741, 0.0413, 0.1615, 0.2695], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0207, 0.0187, 0.0187, 0.0181, 0.0145, 0.0188, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 03:12:37,086 INFO [train2.py:809] (0/4) Epoch 10, batch 2200, loss[ctc_loss=0.1103, att_loss=0.2445, loss=0.2177, over 16013.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007554, over 40.00 utterances.], tot_loss[ctc_loss=0.1093, att_loss=0.2543, loss=0.2253, over 3273766.90 frames. utt_duration=1278 frames, utt_pad_proportion=0.04836, over 10261.54 utterances.], batch size: 40, lr: 1.09e-02, grad_scale: 8.0 2023-03-08 03:12:58,884 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38068.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:13:56,128 INFO [train2.py:809] (0/4) Epoch 10, batch 2250, loss[ctc_loss=0.09619, att_loss=0.2328, loss=0.2054, over 16017.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006769, over 40.00 utterances.], tot_loss[ctc_loss=0.1104, att_loss=0.2553, loss=0.2263, over 3278840.85 frames. utt_duration=1244 frames, utt_pad_proportion=0.05496, over 10552.89 utterances.], batch size: 40, lr: 1.09e-02, grad_scale: 8.0 2023-03-08 03:14:21,119 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 2.484e+02 3.268e+02 3.999e+02 7.414e+02, threshold=6.535e+02, percent-clipped=1.0 2023-03-08 03:14:59,226 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7174, 5.9375, 5.3172, 5.7075, 5.5311, 5.1710, 5.3154, 5.1416], device='cuda:0'), covar=tensor([0.1094, 0.0919, 0.0930, 0.0714, 0.0777, 0.1399, 0.1991, 0.2113], device='cuda:0'), in_proj_covar=tensor([0.0416, 0.0481, 0.0363, 0.0369, 0.0343, 0.0405, 0.0488, 0.0446], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 03:15:14,534 INFO [train2.py:809] (0/4) Epoch 10, batch 2300, loss[ctc_loss=0.1046, att_loss=0.2292, loss=0.2043, over 15496.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.007948, over 36.00 utterances.], tot_loss[ctc_loss=0.11, att_loss=0.2544, loss=0.2255, over 3280624.07 frames. utt_duration=1265 frames, utt_pad_proportion=0.04881, over 10385.44 utterances.], batch size: 36, lr: 1.09e-02, grad_scale: 8.0 2023-03-08 03:15:19,424 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3401, 4.8127, 4.7121, 4.9169, 4.7686, 4.5239, 3.0065, 4.5963], device='cuda:0'), covar=tensor([0.0123, 0.0165, 0.0140, 0.0097, 0.0125, 0.0122, 0.0923, 0.0348], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0070, 0.0082, 0.0051, 0.0055, 0.0065, 0.0088, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 03:15:31,518 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0785, 5.3441, 5.6307, 5.5393, 5.4183, 5.9619, 5.2318, 6.0840], device='cuda:0'), covar=tensor([0.0672, 0.0675, 0.0703, 0.0934, 0.1956, 0.0945, 0.0615, 0.0657], device='cuda:0'), in_proj_covar=tensor([0.0678, 0.0415, 0.0484, 0.0537, 0.0714, 0.0472, 0.0392, 0.0475], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 03:16:33,796 INFO [train2.py:809] (0/4) Epoch 10, batch 2350, loss[ctc_loss=0.09195, att_loss=0.2422, loss=0.2122, over 16964.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007874, over 50.00 utterances.], tot_loss[ctc_loss=0.1104, att_loss=0.2546, loss=0.2257, over 3281524.73 frames. utt_duration=1242 frames, utt_pad_proportion=0.05421, over 10582.75 utterances.], batch size: 50, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:16:59,113 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.810e+02 2.404e+02 3.254e+02 4.140e+02 7.234e+02, threshold=6.508e+02, percent-clipped=3.0 2023-03-08 03:17:37,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6964, 3.9773, 3.8631, 3.9600, 3.9849, 3.7851, 3.0280, 3.9022], device='cuda:0'), covar=tensor([0.0124, 0.0123, 0.0132, 0.0083, 0.0092, 0.0119, 0.0612, 0.0203], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0071, 0.0084, 0.0051, 0.0056, 0.0067, 0.0089, 0.0089], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 03:17:53,623 INFO [train2.py:809] (0/4) Epoch 10, batch 2400, loss[ctc_loss=0.1142, att_loss=0.2602, loss=0.231, over 17312.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.01134, over 55.00 utterances.], tot_loss[ctc_loss=0.1093, att_loss=0.2537, loss=0.2248, over 3270700.58 frames. utt_duration=1262 frames, utt_pad_proportion=0.05155, over 10375.70 utterances.], batch size: 55, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:18:47,996 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5846, 4.6980, 4.5889, 4.6542, 5.0478, 4.8282, 4.6854, 2.2454], device='cuda:0'), covar=tensor([0.0199, 0.0234, 0.0209, 0.0198, 0.1090, 0.0141, 0.0170, 0.2528], device='cuda:0'), in_proj_covar=tensor([0.0122, 0.0124, 0.0129, 0.0133, 0.0320, 0.0118, 0.0113, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 03:19:14,658 INFO [train2.py:809] (0/4) Epoch 10, batch 2450, loss[ctc_loss=0.09223, att_loss=0.2428, loss=0.2127, over 16262.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.008239, over 43.00 utterances.], tot_loss[ctc_loss=0.1092, att_loss=0.2536, loss=0.2247, over 3268568.22 frames. utt_duration=1257 frames, utt_pad_proportion=0.05347, over 10413.99 utterances.], batch size: 43, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:19:40,304 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.470e+02 3.021e+02 3.879e+02 8.501e+02, threshold=6.042e+02, percent-clipped=3.0 2023-03-08 03:20:05,099 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38335.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:20:34,412 INFO [train2.py:809] (0/4) Epoch 10, batch 2500, loss[ctc_loss=0.1284, att_loss=0.2796, loss=0.2493, over 17040.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009594, over 52.00 utterances.], tot_loss[ctc_loss=0.109, att_loss=0.2528, loss=0.2241, over 3266510.20 frames. utt_duration=1269 frames, utt_pad_proportion=0.05109, over 10304.73 utterances.], batch size: 52, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:20:36,764 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-08 03:20:43,837 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38360.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:21:43,332 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-08 03:21:53,304 INFO [train2.py:809] (0/4) Epoch 10, batch 2550, loss[ctc_loss=0.0645, att_loss=0.2216, loss=0.1902, over 9739.00 frames. utt_duration=1857 frames, utt_pad_proportion=0.2552, over 21.00 utterances.], tot_loss[ctc_loss=0.1096, att_loss=0.2532, loss=0.2245, over 3258489.04 frames. utt_duration=1250 frames, utt_pad_proportion=0.05667, over 10439.74 utterances.], batch size: 21, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:22:18,520 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.579e+02 2.719e+02 3.198e+02 3.977e+02 1.146e+03, threshold=6.397e+02, percent-clipped=5.0 2023-03-08 03:22:21,690 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38421.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:23:12,911 INFO [train2.py:809] (0/4) Epoch 10, batch 2600, loss[ctc_loss=0.1317, att_loss=0.2766, loss=0.2476, over 17315.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02285, over 59.00 utterances.], tot_loss[ctc_loss=0.1104, att_loss=0.2538, loss=0.2251, over 3259005.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.06432, over 10654.11 utterances.], batch size: 59, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:23:19,815 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2023-03-08 03:24:31,644 INFO [train2.py:809] (0/4) Epoch 10, batch 2650, loss[ctc_loss=0.07463, att_loss=0.2251, loss=0.195, over 16185.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.005265, over 41.00 utterances.], tot_loss[ctc_loss=0.1103, att_loss=0.2542, loss=0.2255, over 3266215.35 frames. utt_duration=1224 frames, utt_pad_proportion=0.06198, over 10691.13 utterances.], batch size: 41, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:24:58,438 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.727e+02 3.206e+02 4.393e+02 9.975e+02, threshold=6.412e+02, percent-clipped=4.0 2023-03-08 03:25:51,259 INFO [train2.py:809] (0/4) Epoch 10, batch 2700, loss[ctc_loss=0.1104, att_loss=0.2717, loss=0.2395, over 17277.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01178, over 55.00 utterances.], tot_loss[ctc_loss=0.1107, att_loss=0.2549, loss=0.226, over 3272544.65 frames. utt_duration=1205 frames, utt_pad_proportion=0.06519, over 10874.47 utterances.], batch size: 55, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:25:54,746 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6857, 3.7986, 3.6749, 3.1157, 3.7371, 3.8739, 3.6943, 2.5044], device='cuda:0'), covar=tensor([0.1140, 0.1730, 0.2594, 0.6222, 0.1424, 0.2815, 0.0757, 0.8223], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0113, 0.0119, 0.0185, 0.0098, 0.0173, 0.0099, 0.0173], device='cuda:0'), out_proj_covar=tensor([8.9315e-05, 9.8815e-05, 1.0697e-04, 1.5045e-04, 9.0358e-05, 1.4314e-04, 8.7462e-05, 1.4008e-04], device='cuda:0') 2023-03-08 03:27:11,535 INFO [train2.py:809] (0/4) Epoch 10, batch 2750, loss[ctc_loss=0.1289, att_loss=0.2669, loss=0.2393, over 17357.00 frames. utt_duration=880.3 frames, utt_pad_proportion=0.07823, over 79.00 utterances.], tot_loss[ctc_loss=0.1109, att_loss=0.2554, loss=0.2265, over 3283288.16 frames. utt_duration=1214 frames, utt_pad_proportion=0.06093, over 10835.23 utterances.], batch size: 79, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:27:38,143 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.648e+02 3.075e+02 3.913e+02 9.612e+02, threshold=6.151e+02, percent-clipped=3.0 2023-03-08 03:28:01,947 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=38635.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:28:19,457 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4616, 2.4548, 5.0167, 3.8433, 3.1091, 4.3238, 4.7412, 4.6245], device='cuda:0'), covar=tensor([0.0266, 0.1917, 0.0136, 0.1160, 0.1950, 0.0251, 0.0106, 0.0230], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0251, 0.0134, 0.0314, 0.0283, 0.0189, 0.0117, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0003, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-03-08 03:28:31,006 INFO [train2.py:809] (0/4) Epoch 10, batch 2800, loss[ctc_loss=0.1088, att_loss=0.2612, loss=0.2307, over 17292.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01011, over 55.00 utterances.], tot_loss[ctc_loss=0.1107, att_loss=0.2555, loss=0.2265, over 3286110.54 frames. utt_duration=1228 frames, utt_pad_proportion=0.05685, over 10717.68 utterances.], batch size: 55, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:29:18,054 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=38683.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:29:51,020 INFO [train2.py:809] (0/4) Epoch 10, batch 2850, loss[ctc_loss=0.1185, att_loss=0.2708, loss=0.2403, over 16926.00 frames. utt_duration=685.4 frames, utt_pad_proportion=0.1412, over 99.00 utterances.], tot_loss[ctc_loss=0.1106, att_loss=0.2555, loss=0.2265, over 3286690.99 frames. utt_duration=1216 frames, utt_pad_proportion=0.05873, over 10822.17 utterances.], batch size: 99, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:30:11,787 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=38716.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:30:17,872 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.492e+02 2.597e+02 3.110e+02 4.075e+02 8.061e+02, threshold=6.220e+02, percent-clipped=4.0 2023-03-08 03:30:51,176 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1384, 4.2199, 4.0347, 4.0968, 4.6987, 4.3193, 4.2211, 2.2118], device='cuda:0'), covar=tensor([0.0300, 0.0518, 0.0439, 0.0282, 0.0987, 0.0268, 0.0339, 0.2249], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0127, 0.0133, 0.0135, 0.0327, 0.0120, 0.0117, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 03:30:57,130 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9691, 6.1140, 5.5219, 5.8802, 5.8292, 5.4340, 5.5924, 5.3749], device='cuda:0'), covar=tensor([0.0969, 0.0829, 0.0849, 0.0845, 0.0768, 0.1468, 0.2230, 0.2332], device='cuda:0'), in_proj_covar=tensor([0.0417, 0.0495, 0.0362, 0.0371, 0.0344, 0.0409, 0.0500, 0.0443], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 03:31:10,904 INFO [train2.py:809] (0/4) Epoch 10, batch 2900, loss[ctc_loss=0.07748, att_loss=0.2362, loss=0.2044, over 16277.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006782, over 43.00 utterances.], tot_loss[ctc_loss=0.1102, att_loss=0.2544, loss=0.2256, over 3279849.04 frames. utt_duration=1232 frames, utt_pad_proportion=0.05602, over 10662.18 utterances.], batch size: 43, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:31:52,377 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-03-08 03:32:30,744 INFO [train2.py:809] (0/4) Epoch 10, batch 2950, loss[ctc_loss=0.09652, att_loss=0.2455, loss=0.2157, over 16273.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007691, over 43.00 utterances.], tot_loss[ctc_loss=0.1108, att_loss=0.2545, loss=0.2258, over 3278280.38 frames. utt_duration=1230 frames, utt_pad_proportion=0.05692, over 10671.65 utterances.], batch size: 43, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:32:57,340 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.796e+02 2.729e+02 3.322e+02 4.124e+02 1.749e+03, threshold=6.643e+02, percent-clipped=6.0 2023-03-08 03:33:02,355 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=38823.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:33:32,143 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1421, 5.1945, 5.0328, 2.7417, 5.0396, 4.6705, 4.3384, 2.8454], device='cuda:0'), covar=tensor([0.0139, 0.0079, 0.0236, 0.1083, 0.0076, 0.0157, 0.0294, 0.1244], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0079, 0.0071, 0.0100, 0.0067, 0.0090, 0.0089, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 03:33:41,431 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8290, 1.5609, 1.9654, 2.3259, 2.5684, 1.6255, 1.6846, 2.8203], device='cuda:0'), covar=tensor([0.0779, 0.3861, 0.3222, 0.1227, 0.1318, 0.2068, 0.2721, 0.0893], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0085, 0.0088, 0.0072, 0.0072, 0.0068, 0.0081, 0.0061], device='cuda:0'), out_proj_covar=tensor([4.6019e-05, 5.7031e-05, 5.8403e-05, 4.8017e-05, 4.5863e-05, 4.7189e-05, 5.4566e-05, 4.3294e-05], device='cuda:0') 2023-03-08 03:33:50,298 INFO [train2.py:809] (0/4) Epoch 10, batch 3000, loss[ctc_loss=0.1079, att_loss=0.2564, loss=0.2267, over 16114.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006941, over 42.00 utterances.], tot_loss[ctc_loss=0.1113, att_loss=0.2552, loss=0.2264, over 3283755.13 frames. utt_duration=1221 frames, utt_pad_proportion=0.05807, over 10775.09 utterances.], batch size: 42, lr: 1.08e-02, grad_scale: 8.0 2023-03-08 03:33:50,300 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 03:34:06,507 INFO [train2.py:843] (0/4) Epoch 10, validation: ctc_loss=0.0536, att_loss=0.2406, loss=0.2032, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 03:34:06,508 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 03:34:30,287 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9945, 5.0086, 4.9570, 2.1099, 2.0113, 2.4060, 2.9883, 3.8544], device='cuda:0'), covar=tensor([0.0762, 0.0225, 0.0214, 0.4887, 0.6031, 0.3226, 0.2280, 0.1883], device='cuda:0'), in_proj_covar=tensor([0.0332, 0.0215, 0.0232, 0.0202, 0.0352, 0.0336, 0.0228, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.5224e-04, 8.1902e-05, 9.9740e-05, 9.0978e-05, 1.5480e-04, 1.3720e-04, 9.0181e-05, 1.5115e-04], device='cuda:0') 2023-03-08 03:34:54,864 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=38884.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:35:22,983 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-03-08 03:35:26,805 INFO [train2.py:809] (0/4) Epoch 10, batch 3050, loss[ctc_loss=0.09193, att_loss=0.2397, loss=0.2102, over 16179.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006837, over 41.00 utterances.], tot_loss[ctc_loss=0.1106, att_loss=0.2545, loss=0.2257, over 3273960.97 frames. utt_duration=1207 frames, utt_pad_proportion=0.06528, over 10867.55 utterances.], batch size: 41, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:35:53,206 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.391e+02 2.901e+02 3.646e+02 9.557e+02, threshold=5.803e+02, percent-clipped=2.0 2023-03-08 03:36:27,902 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0320, 6.2132, 5.5658, 6.0468, 5.9568, 5.4329, 5.6456, 5.4305], device='cuda:0'), covar=tensor([0.1301, 0.0838, 0.0867, 0.0718, 0.0673, 0.1342, 0.2196, 0.2308], device='cuda:0'), in_proj_covar=tensor([0.0421, 0.0490, 0.0362, 0.0375, 0.0349, 0.0410, 0.0503, 0.0446], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 03:36:47,155 INFO [train2.py:809] (0/4) Epoch 10, batch 3100, loss[ctc_loss=0.1184, att_loss=0.2634, loss=0.2344, over 16879.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007649, over 49.00 utterances.], tot_loss[ctc_loss=0.1094, att_loss=0.2536, loss=0.2248, over 3268843.66 frames. utt_duration=1225 frames, utt_pad_proportion=0.06204, over 10684.39 utterances.], batch size: 49, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:38:08,095 INFO [train2.py:809] (0/4) Epoch 10, batch 3150, loss[ctc_loss=0.1463, att_loss=0.2788, loss=0.2523, over 17298.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01149, over 55.00 utterances.], tot_loss[ctc_loss=0.1089, att_loss=0.2536, loss=0.2247, over 3276904.12 frames. utt_duration=1247 frames, utt_pad_proportion=0.05537, over 10524.72 utterances.], batch size: 55, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:38:28,464 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39016.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:38:34,151 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.681e+02 2.629e+02 3.270e+02 4.032e+02 9.981e+02, threshold=6.541e+02, percent-clipped=6.0 2023-03-08 03:39:27,538 INFO [train2.py:809] (0/4) Epoch 10, batch 3200, loss[ctc_loss=0.0919, att_loss=0.2381, loss=0.2089, over 16175.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006581, over 41.00 utterances.], tot_loss[ctc_loss=0.1078, att_loss=0.2531, loss=0.224, over 3275191.90 frames. utt_duration=1247 frames, utt_pad_proportion=0.05476, over 10519.80 utterances.], batch size: 41, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:39:44,534 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39064.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:40:00,722 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-03-08 03:40:23,773 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0290, 5.3119, 5.6005, 5.4543, 5.4117, 6.0018, 5.1259, 6.0759], device='cuda:0'), covar=tensor([0.0663, 0.0741, 0.0707, 0.1073, 0.1765, 0.0746, 0.0604, 0.0589], device='cuda:0'), in_proj_covar=tensor([0.0682, 0.0417, 0.0480, 0.0544, 0.0724, 0.0475, 0.0388, 0.0476], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 03:40:48,184 INFO [train2.py:809] (0/4) Epoch 10, batch 3250, loss[ctc_loss=0.1197, att_loss=0.2654, loss=0.2362, over 17047.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009978, over 53.00 utterances.], tot_loss[ctc_loss=0.1086, att_loss=0.2539, loss=0.2248, over 3283034.02 frames. utt_duration=1219 frames, utt_pad_proportion=0.05812, over 10789.06 utterances.], batch size: 53, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:41:13,727 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.722e+02 2.584e+02 3.081e+02 4.035e+02 1.100e+03, threshold=6.162e+02, percent-clipped=4.0 2023-03-08 03:41:51,135 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-03-08 03:42:08,483 INFO [train2.py:809] (0/4) Epoch 10, batch 3300, loss[ctc_loss=0.09495, att_loss=0.2567, loss=0.2244, over 17013.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008312, over 51.00 utterances.], tot_loss[ctc_loss=0.1078, att_loss=0.2533, loss=0.2242, over 3285836.22 frames. utt_duration=1252 frames, utt_pad_proportion=0.04982, over 10510.95 utterances.], batch size: 51, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:42:19,014 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0953, 5.2221, 5.0085, 2.7106, 4.9967, 4.6496, 4.1504, 2.5190], device='cuda:0'), covar=tensor([0.0103, 0.0067, 0.0209, 0.1144, 0.0074, 0.0168, 0.0371, 0.1627], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0080, 0.0071, 0.0101, 0.0067, 0.0092, 0.0090, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 03:42:37,175 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.96 vs. limit=5.0 2023-03-08 03:42:48,928 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39179.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:42:58,364 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39185.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:43:29,378 INFO [train2.py:809] (0/4) Epoch 10, batch 3350, loss[ctc_loss=0.1466, att_loss=0.2887, loss=0.2603, over 17125.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01467, over 56.00 utterances.], tot_loss[ctc_loss=0.1083, att_loss=0.2535, loss=0.2245, over 3288178.27 frames. utt_duration=1258 frames, utt_pad_proportion=0.04806, over 10465.66 utterances.], batch size: 56, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:43:38,137 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6491, 2.8186, 3.7021, 3.0094, 3.4588, 4.7073, 4.3151, 3.4766], device='cuda:0'), covar=tensor([0.0358, 0.1922, 0.1156, 0.1437, 0.1223, 0.0752, 0.0719, 0.1300], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0228, 0.0243, 0.0206, 0.0240, 0.0298, 0.0215, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 03:43:41,970 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.96 vs. limit=2.0 2023-03-08 03:43:54,947 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.724e+02 2.443e+02 2.910e+02 3.809e+02 9.404e+02, threshold=5.819e+02, percent-clipped=2.0 2023-03-08 03:44:29,188 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2347, 5.1499, 5.0914, 2.4859, 2.0906, 2.9349, 3.1719, 3.9609], device='cuda:0'), covar=tensor([0.0604, 0.0255, 0.0193, 0.4149, 0.5854, 0.2447, 0.2067, 0.1766], device='cuda:0'), in_proj_covar=tensor([0.0329, 0.0213, 0.0230, 0.0202, 0.0346, 0.0335, 0.0225, 0.0349], device='cuda:0'), out_proj_covar=tensor([1.5031e-04, 8.0527e-05, 9.8815e-05, 9.0585e-05, 1.5241e-04, 1.3646e-04, 8.8877e-05, 1.4901e-04], device='cuda:0') 2023-03-08 03:44:36,994 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39246.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:44:49,734 INFO [train2.py:809] (0/4) Epoch 10, batch 3400, loss[ctc_loss=0.1149, att_loss=0.2689, loss=0.2381, over 16977.00 frames. utt_duration=1360 frames, utt_pad_proportion=0.006764, over 50.00 utterances.], tot_loss[ctc_loss=0.1088, att_loss=0.2535, loss=0.2246, over 3277527.67 frames. utt_duration=1252 frames, utt_pad_proportion=0.05095, over 10485.17 utterances.], batch size: 50, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:45:37,862 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39284.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:46:11,197 INFO [train2.py:809] (0/4) Epoch 10, batch 3450, loss[ctc_loss=0.1223, att_loss=0.2688, loss=0.2395, over 16689.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006364, over 46.00 utterances.], tot_loss[ctc_loss=0.1084, att_loss=0.2536, loss=0.2246, over 3283083.70 frames. utt_duration=1264 frames, utt_pad_proportion=0.04777, over 10404.13 utterances.], batch size: 46, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:46:28,638 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0185, 4.2821, 4.0263, 4.2416, 2.5234, 4.2817, 2.3933, 1.8227], device='cuda:0'), covar=tensor([0.0357, 0.0162, 0.0735, 0.0191, 0.1913, 0.0158, 0.1703, 0.1782], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0106, 0.0252, 0.0107, 0.0219, 0.0104, 0.0222, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 03:46:36,007 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.709e+02 2.326e+02 2.969e+02 3.835e+02 9.960e+02, threshold=5.938e+02, percent-clipped=3.0 2023-03-08 03:46:47,258 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5266, 5.0781, 4.8236, 5.0413, 5.1606, 4.7536, 3.6808, 4.9542], device='cuda:0'), covar=tensor([0.0102, 0.0111, 0.0094, 0.0072, 0.0067, 0.0103, 0.0572, 0.0160], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0069, 0.0083, 0.0052, 0.0056, 0.0066, 0.0088, 0.0088], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 03:46:58,364 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7377, 3.7082, 3.1101, 3.3458, 3.8224, 3.4855, 2.6399, 4.2454], device='cuda:0'), covar=tensor([0.1167, 0.0467, 0.1099, 0.0698, 0.0725, 0.0708, 0.0970, 0.0476], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0178, 0.0202, 0.0171, 0.0226, 0.0207, 0.0178, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 03:47:16,331 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39345.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:47:31,574 INFO [train2.py:809] (0/4) Epoch 10, batch 3500, loss[ctc_loss=0.09074, att_loss=0.2318, loss=0.2036, over 15489.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.009616, over 36.00 utterances.], tot_loss[ctc_loss=0.1083, att_loss=0.2534, loss=0.2243, over 3285173.52 frames. utt_duration=1267 frames, utt_pad_proportion=0.0462, over 10387.49 utterances.], batch size: 36, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:47:48,553 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.04 vs. limit=2.0 2023-03-08 03:48:52,042 INFO [train2.py:809] (0/4) Epoch 10, batch 3550, loss[ctc_loss=0.1002, att_loss=0.2359, loss=0.2088, over 15794.00 frames. utt_duration=1664 frames, utt_pad_proportion=0.007317, over 38.00 utterances.], tot_loss[ctc_loss=0.1079, att_loss=0.2531, loss=0.2241, over 3286666.77 frames. utt_duration=1276 frames, utt_pad_proportion=0.04424, over 10312.22 utterances.], batch size: 38, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:49:17,422 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.387e+02 2.370e+02 2.861e+02 3.451e+02 5.590e+02, threshold=5.722e+02, percent-clipped=0.0 2023-03-08 03:50:11,854 INFO [train2.py:809] (0/4) Epoch 10, batch 3600, loss[ctc_loss=0.1121, att_loss=0.2672, loss=0.2361, over 17015.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.007429, over 51.00 utterances.], tot_loss[ctc_loss=0.1084, att_loss=0.2536, loss=0.2245, over 3292723.62 frames. utt_duration=1277 frames, utt_pad_proportion=0.04274, over 10322.13 utterances.], batch size: 51, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:50:27,108 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39463.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:50:44,306 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6332, 2.7256, 3.0664, 4.4491, 4.0506, 3.9684, 2.9999, 2.0479], device='cuda:0'), covar=tensor([0.0494, 0.2209, 0.1147, 0.0454, 0.0645, 0.0378, 0.1444, 0.2507], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0205, 0.0183, 0.0187, 0.0182, 0.0148, 0.0191, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 03:50:52,025 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39479.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:51:17,781 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2423, 4.0722, 3.3674, 3.9146, 4.1370, 3.7658, 3.0576, 4.5280], device='cuda:0'), covar=tensor([0.0898, 0.0386, 0.0896, 0.0430, 0.0591, 0.0572, 0.0843, 0.0465], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0180, 0.0205, 0.0173, 0.0229, 0.0208, 0.0180, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 03:51:32,917 INFO [train2.py:809] (0/4) Epoch 10, batch 3650, loss[ctc_loss=0.09789, att_loss=0.2514, loss=0.2207, over 17312.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02373, over 59.00 utterances.], tot_loss[ctc_loss=0.1092, att_loss=0.2542, loss=0.2252, over 3299030.06 frames. utt_duration=1231 frames, utt_pad_proportion=0.05057, over 10735.75 utterances.], batch size: 59, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:51:33,424 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4456, 4.4014, 4.2627, 4.2944, 4.7194, 4.3414, 4.2860, 2.1462], device='cuda:0'), covar=tensor([0.0200, 0.0231, 0.0302, 0.0230, 0.0982, 0.0231, 0.0239, 0.2225], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0125, 0.0132, 0.0134, 0.0325, 0.0121, 0.0117, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 03:51:57,847 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.423e+02 2.407e+02 3.049e+02 3.792e+02 1.207e+03, threshold=6.099e+02, percent-clipped=6.0 2023-03-08 03:52:04,563 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39524.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:52:09,222 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39527.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:52:31,740 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39541.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:52:53,643 INFO [train2.py:809] (0/4) Epoch 10, batch 3700, loss[ctc_loss=0.1123, att_loss=0.2526, loss=0.2245, over 16322.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006566, over 45.00 utterances.], tot_loss[ctc_loss=0.1083, att_loss=0.2539, loss=0.2248, over 3294823.03 frames. utt_duration=1230 frames, utt_pad_proportion=0.05171, over 10728.44 utterances.], batch size: 45, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:53:03,146 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8100, 3.7442, 3.0773, 3.4293, 3.8714, 3.5343, 2.5995, 4.2627], device='cuda:0'), covar=tensor([0.1002, 0.0425, 0.1019, 0.0526, 0.0575, 0.0583, 0.0970, 0.0441], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0177, 0.0203, 0.0172, 0.0227, 0.0207, 0.0180, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 03:54:12,930 INFO [train2.py:809] (0/4) Epoch 10, batch 3750, loss[ctc_loss=0.1085, att_loss=0.2509, loss=0.2224, over 16553.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005642, over 45.00 utterances.], tot_loss[ctc_loss=0.1083, att_loss=0.2537, loss=0.2246, over 3294811.81 frames. utt_duration=1253 frames, utt_pad_proportion=0.04618, over 10531.71 utterances.], batch size: 45, lr: 1.07e-02, grad_scale: 8.0 2023-03-08 03:54:38,432 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.452e+02 2.346e+02 2.722e+02 3.425e+02 7.090e+02, threshold=5.444e+02, percent-clipped=1.0 2023-03-08 03:55:08,046 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39639.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 03:55:09,563 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39640.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:55:33,232 INFO [train2.py:809] (0/4) Epoch 10, batch 3800, loss[ctc_loss=0.0985, att_loss=0.2353, loss=0.208, over 16012.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007094, over 40.00 utterances.], tot_loss[ctc_loss=0.1072, att_loss=0.2523, loss=0.2232, over 3290502.45 frames. utt_duration=1275 frames, utt_pad_proportion=0.04328, over 10336.93 utterances.], batch size: 40, lr: 1.06e-02, grad_scale: 8.0 2023-03-08 03:56:46,642 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39700.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 03:56:53,081 INFO [train2.py:809] (0/4) Epoch 10, batch 3850, loss[ctc_loss=0.1285, att_loss=0.2699, loss=0.2417, over 17302.00 frames. utt_duration=877.3 frames, utt_pad_proportion=0.08035, over 79.00 utterances.], tot_loss[ctc_loss=0.1068, att_loss=0.2519, loss=0.2228, over 3277364.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.04983, over 10396.57 utterances.], batch size: 79, lr: 1.06e-02, grad_scale: 8.0 2023-03-08 03:57:18,319 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.719e+02 2.275e+02 2.627e+02 3.517e+02 7.871e+02, threshold=5.254e+02, percent-clipped=5.0 2023-03-08 03:57:49,020 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5009, 2.2146, 5.0107, 3.9860, 2.9364, 4.3720, 4.8012, 4.5685], device='cuda:0'), covar=tensor([0.0257, 0.1843, 0.0146, 0.0901, 0.1903, 0.0254, 0.0096, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0248, 0.0134, 0.0308, 0.0278, 0.0189, 0.0116, 0.0151], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0003, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-03-08 03:58:02,964 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2545, 5.2293, 5.1058, 2.6475, 2.0049, 2.7843, 3.1313, 3.8724], device='cuda:0'), covar=tensor([0.0591, 0.0228, 0.0231, 0.3990, 0.5872, 0.2765, 0.2091, 0.1859], device='cuda:0'), in_proj_covar=tensor([0.0333, 0.0217, 0.0231, 0.0207, 0.0352, 0.0338, 0.0229, 0.0355], device='cuda:0'), out_proj_covar=tensor([1.5183e-04, 8.1667e-05, 9.9315e-05, 9.2970e-05, 1.5433e-04, 1.3769e-04, 9.0799e-05, 1.5074e-04], device='cuda:0') 2023-03-08 03:58:10,237 INFO [train2.py:809] (0/4) Epoch 10, batch 3900, loss[ctc_loss=0.0762, att_loss=0.2216, loss=0.1925, over 15769.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008589, over 38.00 utterances.], tot_loss[ctc_loss=0.106, att_loss=0.2511, loss=0.2221, over 3274508.11 frames. utt_duration=1278 frames, utt_pad_proportion=0.04784, over 10264.59 utterances.], batch size: 38, lr: 1.06e-02, grad_scale: 8.0 2023-03-08 03:58:23,689 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-08 03:58:30,638 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39767.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:58:31,461 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 03:58:56,922 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-03-08 03:59:26,945 INFO [train2.py:809] (0/4) Epoch 10, batch 3950, loss[ctc_loss=0.106, att_loss=0.2497, loss=0.221, over 16165.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.00768, over 41.00 utterances.], tot_loss[ctc_loss=0.1068, att_loss=0.2518, loss=0.2228, over 3275210.31 frames. utt_duration=1268 frames, utt_pad_proportion=0.04996, over 10343.67 utterances.], batch size: 41, lr: 1.06e-02, grad_scale: 8.0 2023-03-08 03:59:44,064 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=39815.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:59:50,034 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39819.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 03:59:51,183 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 2.608e+02 3.275e+02 3.896e+02 7.798e+02, threshold=6.551e+02, percent-clipped=9.0 2023-03-08 04:00:04,086 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39828.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:00:18,483 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-10.pt 2023-03-08 04:00:45,403 INFO [train2.py:809] (0/4) Epoch 11, batch 0, loss[ctc_loss=0.1027, att_loss=0.2463, loss=0.2176, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.008085, over 44.00 utterances.], tot_loss[ctc_loss=0.1027, att_loss=0.2463, loss=0.2176, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.008085, over 44.00 utterances.], batch size: 44, lr: 1.01e-02, grad_scale: 8.0 2023-03-08 04:00:45,405 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 04:00:57,590 INFO [train2.py:843] (0/4) Epoch 11, validation: ctc_loss=0.05063, att_loss=0.2383, loss=0.2008, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 04:00:57,591 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 04:01:02,321 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39841.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:01:58,642 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=39876.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:02:18,273 INFO [train2.py:809] (0/4) Epoch 11, batch 50, loss[ctc_loss=0.0866, att_loss=0.2384, loss=0.208, over 16470.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006403, over 46.00 utterances.], tot_loss[ctc_loss=0.1028, att_loss=0.2504, loss=0.2209, over 737765.25 frames. utt_duration=1388 frames, utt_pad_proportion=0.01852, over 2128.46 utterances.], batch size: 46, lr: 1.01e-02, grad_scale: 8.0 2023-03-08 04:02:19,877 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39889.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:02:55,386 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9939, 4.9949, 4.9000, 4.8197, 5.4188, 5.0552, 4.8539, 2.5187], device='cuda:0'), covar=tensor([0.0178, 0.0189, 0.0187, 0.0246, 0.0752, 0.0167, 0.0206, 0.2119], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0125, 0.0133, 0.0135, 0.0325, 0.0121, 0.0118, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 04:03:08,052 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.626e+02 3.075e+02 3.847e+02 1.291e+03, threshold=6.150e+02, percent-clipped=4.0 2023-03-08 04:03:37,056 INFO [train2.py:809] (0/4) Epoch 11, batch 100, loss[ctc_loss=0.1355, att_loss=0.2722, loss=0.2448, over 17121.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01493, over 56.00 utterances.], tot_loss[ctc_loss=0.1031, att_loss=0.2486, loss=0.2195, over 1294236.02 frames. utt_duration=1355 frames, utt_pad_proportion=0.03084, over 3824.36 utterances.], batch size: 56, lr: 1.01e-02, grad_scale: 8.0 2023-03-08 04:03:40,233 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=39940.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:03:51,336 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5756, 2.6967, 4.9581, 4.0346, 3.0841, 4.5211, 5.0696, 4.7168], device='cuda:0'), covar=tensor([0.0280, 0.1640, 0.0276, 0.1006, 0.1921, 0.0221, 0.0100, 0.0232], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0246, 0.0134, 0.0307, 0.0277, 0.0188, 0.0115, 0.0149], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-03-08 04:04:15,722 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0283, 6.2014, 5.5339, 5.9496, 5.8764, 5.5204, 5.7201, 5.4154], device='cuda:0'), covar=tensor([0.1042, 0.0886, 0.0891, 0.0803, 0.0628, 0.1385, 0.2009, 0.2611], device='cuda:0'), in_proj_covar=tensor([0.0426, 0.0500, 0.0365, 0.0382, 0.0356, 0.0409, 0.0503, 0.0450], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 04:04:55,826 INFO [train2.py:809] (0/4) Epoch 11, batch 150, loss[ctc_loss=0.09206, att_loss=0.2305, loss=0.2028, over 15879.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009393, over 39.00 utterances.], tot_loss[ctc_loss=0.1044, att_loss=0.2498, loss=0.2207, over 1733285.48 frames. utt_duration=1319 frames, utt_pad_proportion=0.03801, over 5263.60 utterances.], batch size: 39, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:04:55,936 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=39988.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:05:07,695 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=39995.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 04:05:15,674 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-40000.pt 2023-03-08 04:05:51,853 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.648e+02 2.498e+02 3.139e+02 4.301e+02 8.550e+02, threshold=6.278e+02, percent-clipped=5.0 2023-03-08 04:06:20,488 INFO [train2.py:809] (0/4) Epoch 11, batch 200, loss[ctc_loss=0.1262, att_loss=0.2708, loss=0.2419, over 17228.00 frames. utt_duration=1169 frames, utt_pad_proportion=0.02872, over 59.00 utterances.], tot_loss[ctc_loss=0.1048, att_loss=0.2495, loss=0.2206, over 2070958.06 frames. utt_duration=1285 frames, utt_pad_proportion=0.04867, over 6453.27 utterances.], batch size: 59, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:07:39,777 INFO [train2.py:809] (0/4) Epoch 11, batch 250, loss[ctc_loss=0.09373, att_loss=0.2498, loss=0.2186, over 16764.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.00691, over 48.00 utterances.], tot_loss[ctc_loss=0.1052, att_loss=0.2505, loss=0.2214, over 2341108.31 frames. utt_duration=1278 frames, utt_pad_proportion=0.04799, over 7337.09 utterances.], batch size: 48, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:08:01,539 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3287, 2.7583, 3.5399, 2.6239, 3.4504, 4.4797, 4.3175, 3.1852], device='cuda:0'), covar=tensor([0.0363, 0.1753, 0.1099, 0.1592, 0.1012, 0.0753, 0.0529, 0.1302], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0226, 0.0243, 0.0205, 0.0238, 0.0298, 0.0214, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:08:12,429 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3420, 2.5093, 2.9015, 4.2608, 3.9079, 3.8653, 2.7672, 1.8359], device='cuda:0'), covar=tensor([0.0616, 0.2489, 0.1180, 0.0606, 0.0597, 0.0448, 0.1528, 0.2602], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0207, 0.0187, 0.0188, 0.0180, 0.0147, 0.0190, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 04:08:28,137 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40119.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:08:29,482 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.494e+02 2.990e+02 3.866e+02 8.648e+02, threshold=5.979e+02, percent-clipped=3.0 2023-03-08 04:08:34,911 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40123.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:08:59,715 INFO [train2.py:809] (0/4) Epoch 11, batch 300, loss[ctc_loss=0.11, att_loss=0.2653, loss=0.2342, over 17274.00 frames. utt_duration=876.2 frames, utt_pad_proportion=0.08154, over 79.00 utterances.], tot_loss[ctc_loss=0.1043, att_loss=0.2505, loss=0.2213, over 2551524.25 frames. utt_duration=1308 frames, utt_pad_proportion=0.03975, over 7808.99 utterances.], batch size: 79, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:09:08,826 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-08 04:09:16,997 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 04:09:45,530 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40167.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:09:52,632 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40171.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:10:16,047 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6748, 4.4900, 4.3515, 4.4424, 4.9708, 4.6674, 4.3205, 2.3873], device='cuda:0'), covar=tensor([0.0213, 0.0287, 0.0346, 0.0232, 0.1087, 0.0183, 0.0375, 0.2228], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0129, 0.0137, 0.0138, 0.0333, 0.0123, 0.0123, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 04:10:20,964 INFO [train2.py:809] (0/4) Epoch 11, batch 350, loss[ctc_loss=0.1551, att_loss=0.279, loss=0.2542, over 13871.00 frames. utt_duration=384.2 frames, utt_pad_proportion=0.333, over 145.00 utterances.], tot_loss[ctc_loss=0.1039, att_loss=0.2503, loss=0.221, over 2714412.16 frames. utt_duration=1305 frames, utt_pad_proportion=0.03973, over 8326.72 utterances.], batch size: 145, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:10:31,912 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40195.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:10:40,488 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-03-08 04:11:04,793 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3114, 4.5394, 4.5650, 4.5506, 4.6280, 4.6183, 4.3348, 4.2191], device='cuda:0'), covar=tensor([0.1193, 0.0706, 0.0275, 0.0462, 0.0336, 0.0320, 0.0340, 0.0409], device='cuda:0'), in_proj_covar=tensor([0.0443, 0.0284, 0.0239, 0.0271, 0.0333, 0.0354, 0.0279, 0.0316], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 04:11:07,988 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40218.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:11:11,923 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.687e+02 2.384e+02 2.929e+02 3.608e+02 1.665e+03, threshold=5.857e+02, percent-clipped=2.0 2023-03-08 04:11:33,787 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.80 vs. limit=2.0 2023-03-08 04:11:41,141 INFO [train2.py:809] (0/4) Epoch 11, batch 400, loss[ctc_loss=0.1355, att_loss=0.2703, loss=0.2433, over 17026.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.008408, over 51.00 utterances.], tot_loss[ctc_loss=0.1042, att_loss=0.2501, loss=0.2209, over 2835479.96 frames. utt_duration=1291 frames, utt_pad_proportion=0.04442, over 8797.15 utterances.], batch size: 51, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:12:09,430 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40256.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:12:31,355 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.0647, 3.2030, 3.2342, 2.7964, 3.2935, 3.2734, 3.2387, 2.1161], device='cuda:0'), covar=tensor([0.1270, 0.2257, 0.4707, 0.8535, 0.2155, 0.3438, 0.1055, 0.9504], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0114, 0.0123, 0.0186, 0.0095, 0.0174, 0.0099, 0.0167], device='cuda:0'), out_proj_covar=tensor([8.8984e-05, 9.9919e-05, 1.1065e-04, 1.5138e-04, 8.8697e-05, 1.4440e-04, 8.7556e-05, 1.3711e-04], device='cuda:0') 2023-03-08 04:12:36,149 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40272.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:12:47,321 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40279.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:13:01,945 INFO [train2.py:809] (0/4) Epoch 11, batch 450, loss[ctc_loss=0.0917, att_loss=0.221, loss=0.1951, over 15360.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01183, over 35.00 utterances.], tot_loss[ctc_loss=0.1045, att_loss=0.2505, loss=0.2213, over 2934987.39 frames. utt_duration=1268 frames, utt_pad_proportion=0.05012, over 9268.11 utterances.], batch size: 35, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:13:12,950 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40295.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 04:13:52,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.652e+02 2.417e+02 2.903e+02 3.339e+02 6.124e+02, threshold=5.806e+02, percent-clipped=1.0 2023-03-08 04:14:08,020 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40330.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:14:13,257 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40333.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:14:21,406 INFO [train2.py:809] (0/4) Epoch 11, batch 500, loss[ctc_loss=0.1044, att_loss=0.2595, loss=0.2285, over 17313.00 frames. utt_duration=1005 frames, utt_pad_proportion=0.05189, over 69.00 utterances.], tot_loss[ctc_loss=0.1049, att_loss=0.251, loss=0.2218, over 3017252.94 frames. utt_duration=1282 frames, utt_pad_proportion=0.04437, over 9422.24 utterances.], batch size: 69, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:14:29,172 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40343.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 04:15:41,804 INFO [train2.py:809] (0/4) Epoch 11, batch 550, loss[ctc_loss=0.1139, att_loss=0.2669, loss=0.2363, over 16776.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006155, over 48.00 utterances.], tot_loss[ctc_loss=0.1057, att_loss=0.2515, loss=0.2223, over 3072516.62 frames. utt_duration=1263 frames, utt_pad_proportion=0.05006, over 9741.30 utterances.], batch size: 48, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:15:46,903 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40391.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:16:11,170 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 04:16:22,934 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1741, 4.1504, 4.1860, 4.8924, 2.4691, 4.3633, 2.5191, 2.1568], device='cuda:0'), covar=tensor([0.0291, 0.0244, 0.0764, 0.0084, 0.2066, 0.0169, 0.1703, 0.1752], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0107, 0.0252, 0.0105, 0.0219, 0.0107, 0.0223, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 04:16:33,186 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.316e+02 2.892e+02 3.547e+02 1.294e+03, threshold=5.784e+02, percent-clipped=4.0 2023-03-08 04:16:38,002 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40423.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:17:02,141 INFO [train2.py:809] (0/4) Epoch 11, batch 600, loss[ctc_loss=0.08561, att_loss=0.2211, loss=0.194, over 15874.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01005, over 39.00 utterances.], tot_loss[ctc_loss=0.105, att_loss=0.2511, loss=0.2219, over 3124480.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.0483, over 9926.21 utterances.], batch size: 39, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:17:28,441 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6779, 3.7416, 3.2224, 3.4742, 3.8820, 3.5192, 2.8844, 4.2712], device='cuda:0'), covar=tensor([0.1137, 0.0488, 0.0969, 0.0653, 0.0636, 0.0669, 0.0931, 0.0426], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0180, 0.0204, 0.0173, 0.0232, 0.0209, 0.0181, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 04:17:54,378 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:17:54,586 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40471.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:18:22,137 INFO [train2.py:809] (0/4) Epoch 11, batch 650, loss[ctc_loss=0.1054, att_loss=0.2395, loss=0.2126, over 16112.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.005687, over 42.00 utterances.], tot_loss[ctc_loss=0.1056, att_loss=0.2515, loss=0.2223, over 3156459.15 frames. utt_duration=1245 frames, utt_pad_proportion=0.05335, over 10152.32 utterances.], batch size: 42, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:18:28,772 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40492.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:19:11,085 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40519.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:19:12,491 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.622e+02 2.380e+02 3.105e+02 3.848e+02 6.391e+02, threshold=6.210e+02, percent-clipped=2.0 2023-03-08 04:19:41,590 INFO [train2.py:809] (0/4) Epoch 11, batch 700, loss[ctc_loss=0.09918, att_loss=0.2359, loss=0.2085, over 16188.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.006377, over 41.00 utterances.], tot_loss[ctc_loss=0.1063, att_loss=0.2524, loss=0.2232, over 3186679.64 frames. utt_duration=1226 frames, utt_pad_proportion=0.05762, over 10408.48 utterances.], batch size: 41, lr: 1.01e-02, grad_scale: 16.0 2023-03-08 04:19:53,827 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9483, 6.1759, 5.5374, 5.9666, 5.8046, 5.4417, 5.5981, 5.4188], device='cuda:0'), covar=tensor([0.1083, 0.0949, 0.0854, 0.0771, 0.0831, 0.1423, 0.2376, 0.2590], device='cuda:0'), in_proj_covar=tensor([0.0436, 0.0496, 0.0369, 0.0385, 0.0358, 0.0414, 0.0512, 0.0455], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 04:20:01,432 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40551.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:20:04,666 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40553.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:20:35,059 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4814, 1.9203, 2.3623, 1.8774, 3.1049, 2.3256, 1.9428, 2.4610], device='cuda:0'), covar=tensor([0.0730, 0.4687, 0.3263, 0.2740, 0.2434, 0.1809, 0.3306, 0.1531], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0090, 0.0091, 0.0077, 0.0077, 0.0072, 0.0088, 0.0063], device='cuda:0'), out_proj_covar=tensor([4.9115e-05, 6.0388e-05, 6.1373e-05, 5.1989e-05, 4.9210e-05, 5.0321e-05, 5.8741e-05, 4.5583e-05], device='cuda:0') 2023-03-08 04:20:37,996 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40574.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:21:00,394 INFO [train2.py:809] (0/4) Epoch 11, batch 750, loss[ctc_loss=0.1676, att_loss=0.3002, loss=0.2737, over 17384.00 frames. utt_duration=1009 frames, utt_pad_proportion=0.04879, over 69.00 utterances.], tot_loss[ctc_loss=0.1068, att_loss=0.2524, loss=0.2233, over 3211091.55 frames. utt_duration=1231 frames, utt_pad_proportion=0.05513, over 10447.21 utterances.], batch size: 69, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:21:01,260 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-08 04:21:24,324 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7290, 3.7263, 2.9815, 3.3240, 3.8625, 3.4535, 2.4759, 4.2740], device='cuda:0'), covar=tensor([0.1080, 0.0434, 0.1064, 0.0670, 0.0620, 0.0672, 0.1099, 0.0473], device='cuda:0'), in_proj_covar=tensor([0.0184, 0.0181, 0.0203, 0.0174, 0.0233, 0.0211, 0.0182, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 04:21:51,405 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 2.391e+02 2.736e+02 3.441e+02 6.786e+02, threshold=5.472e+02, percent-clipped=2.0 2023-03-08 04:22:03,984 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40628.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:22:20,790 INFO [train2.py:809] (0/4) Epoch 11, batch 800, loss[ctc_loss=0.07262, att_loss=0.2252, loss=0.1947, over 15889.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008926, over 39.00 utterances.], tot_loss[ctc_loss=0.106, att_loss=0.2523, loss=0.223, over 3222170.81 frames. utt_duration=1242 frames, utt_pad_proportion=0.05446, over 10385.65 utterances.], batch size: 39, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:22:41,717 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40651.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:22:48,051 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2460, 5.2332, 5.0424, 1.8901, 1.9296, 2.6723, 3.0211, 3.7698], device='cuda:0'), covar=tensor([0.0623, 0.0241, 0.0223, 0.5133, 0.6933, 0.3069, 0.2088, 0.1999], device='cuda:0'), in_proj_covar=tensor([0.0333, 0.0217, 0.0234, 0.0207, 0.0352, 0.0338, 0.0231, 0.0355], device='cuda:0'), out_proj_covar=tensor([1.5169e-04, 8.1405e-05, 1.0027e-04, 9.3028e-05, 1.5461e-04, 1.3755e-04, 9.1265e-05, 1.5108e-04], device='cuda:0') 2023-03-08 04:23:38,219 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40686.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:23:41,160 INFO [train2.py:809] (0/4) Epoch 11, batch 850, loss[ctc_loss=0.1916, att_loss=0.2973, loss=0.2762, over 14303.00 frames. utt_duration=380.4 frames, utt_pad_proportion=0.3182, over 151.00 utterances.], tot_loss[ctc_loss=0.106, att_loss=0.2523, loss=0.2231, over 3232411.32 frames. utt_duration=1242 frames, utt_pad_proportion=0.05558, over 10418.79 utterances.], batch size: 151, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:23:49,361 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4560, 2.5929, 2.6197, 2.4310, 2.4808, 2.5385, 2.4704, 1.8521], device='cuda:0'), covar=tensor([0.0920, 0.1921, 0.2590, 0.5523, 0.2171, 0.3746, 0.2013, 0.6494], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0117, 0.0122, 0.0188, 0.0099, 0.0177, 0.0102, 0.0166], device='cuda:0'), out_proj_covar=tensor([8.9601e-05, 1.0287e-04, 1.1014e-04, 1.5374e-04, 9.2047e-05, 1.4629e-04, 9.0135e-05, 1.3686e-04], device='cuda:0') 2023-03-08 04:24:18,985 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40712.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:24:31,775 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 2.347e+02 2.849e+02 3.646e+02 6.696e+02, threshold=5.699e+02, percent-clipped=1.0 2023-03-08 04:25:00,584 INFO [train2.py:809] (0/4) Epoch 11, batch 900, loss[ctc_loss=0.08382, att_loss=0.2366, loss=0.2061, over 15992.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008916, over 40.00 utterances.], tot_loss[ctc_loss=0.1046, att_loss=0.251, loss=0.2217, over 3235393.05 frames. utt_duration=1261 frames, utt_pad_proportion=0.05387, over 10277.66 utterances.], batch size: 40, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:25:06,154 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.20 vs. limit=5.0 2023-03-08 04:25:14,879 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9451, 5.1877, 5.4441, 5.3280, 5.2939, 5.8482, 5.1679, 6.0289], device='cuda:0'), covar=tensor([0.0672, 0.0750, 0.0816, 0.1159, 0.1982, 0.0848, 0.0653, 0.0542], device='cuda:0'), in_proj_covar=tensor([0.0698, 0.0426, 0.0489, 0.0552, 0.0735, 0.0492, 0.0398, 0.0483], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:25:54,879 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2558, 2.2826, 3.4087, 2.6449, 3.2183, 4.3935, 4.2911, 2.8208], device='cuda:0'), covar=tensor([0.0529, 0.2500, 0.1154, 0.1812, 0.1122, 0.0879, 0.0594, 0.1847], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0222, 0.0241, 0.0203, 0.0237, 0.0298, 0.0218, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:26:21,117 INFO [train2.py:809] (0/4) Epoch 11, batch 950, loss[ctc_loss=0.08706, att_loss=0.2262, loss=0.1983, over 15358.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.0121, over 35.00 utterances.], tot_loss[ctc_loss=0.1055, att_loss=0.2516, loss=0.2224, over 3238682.88 frames. utt_duration=1227 frames, utt_pad_proportion=0.06309, over 10570.14 utterances.], batch size: 35, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:27:12,368 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.301e+02 2.865e+02 3.345e+02 1.136e+03, threshold=5.731e+02, percent-clipped=4.0 2023-03-08 04:27:41,336 INFO [train2.py:809] (0/4) Epoch 11, batch 1000, loss[ctc_loss=0.09498, att_loss=0.2627, loss=0.2291, over 17062.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009253, over 53.00 utterances.], tot_loss[ctc_loss=0.1051, att_loss=0.2513, loss=0.2221, over 3238268.92 frames. utt_duration=1237 frames, utt_pad_proportion=0.06047, over 10481.03 utterances.], batch size: 53, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:27:56,969 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=40848.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:28:01,794 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40851.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:28:23,493 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.5686, 5.8215, 5.1003, 5.6235, 5.4586, 4.9380, 5.1189, 5.0169], device='cuda:0'), covar=tensor([0.1237, 0.0881, 0.0905, 0.0677, 0.0819, 0.1465, 0.2244, 0.2060], device='cuda:0'), in_proj_covar=tensor([0.0433, 0.0497, 0.0365, 0.0377, 0.0353, 0.0415, 0.0511, 0.0452], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 04:28:25,232 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40865.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:28:39,147 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40874.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:28:47,872 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8354, 5.0845, 4.5809, 5.2540, 4.5342, 4.9706, 5.2632, 5.0450], device='cuda:0'), covar=tensor([0.0505, 0.0287, 0.0926, 0.0240, 0.0441, 0.0259, 0.0208, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0334, 0.0257, 0.0321, 0.0253, 0.0265, 0.0203, 0.0242, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 04:29:01,313 INFO [train2.py:809] (0/4) Epoch 11, batch 1050, loss[ctc_loss=0.09199, att_loss=0.2292, loss=0.2017, over 15995.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007517, over 40.00 utterances.], tot_loss[ctc_loss=0.1053, att_loss=0.2514, loss=0.2222, over 3248729.96 frames. utt_duration=1246 frames, utt_pad_proportion=0.05806, over 10440.69 utterances.], batch size: 40, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:29:18,880 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40899.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:29:20,763 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40900.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:29:52,388 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.885e+02 2.443e+02 2.925e+02 3.809e+02 9.259e+02, threshold=5.849e+02, percent-clipped=3.0 2023-03-08 04:29:56,094 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40922.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:30:03,296 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40926.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:30:06,347 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40928.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:30:22,024 INFO [train2.py:809] (0/4) Epoch 11, batch 1100, loss[ctc_loss=0.07225, att_loss=0.2212, loss=0.1914, over 15862.00 frames. utt_duration=1628 frames, utt_pad_proportion=0.01075, over 39.00 utterances.], tot_loss[ctc_loss=0.1051, att_loss=0.2516, loss=0.2223, over 3257780.03 frames. utt_duration=1256 frames, utt_pad_proportion=0.05369, over 10386.98 utterances.], batch size: 39, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:30:45,163 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4463, 4.4047, 4.3961, 4.5069, 4.9745, 4.4875, 4.3603, 2.4308], device='cuda:0'), covar=tensor([0.0243, 0.0296, 0.0254, 0.0180, 0.0740, 0.0197, 0.0283, 0.1871], device='cuda:0'), in_proj_covar=tensor([0.0123, 0.0127, 0.0132, 0.0134, 0.0318, 0.0120, 0.0121, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 04:30:59,216 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=40961.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 04:31:02,882 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-03-08 04:31:23,532 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=40976.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:31:34,516 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9934, 5.0163, 4.7946, 2.9573, 4.6826, 4.5713, 4.3296, 2.4628], device='cuda:0'), covar=tensor([0.0099, 0.0075, 0.0228, 0.0886, 0.0079, 0.0185, 0.0268, 0.1448], device='cuda:0'), in_proj_covar=tensor([0.0059, 0.0081, 0.0073, 0.0100, 0.0067, 0.0091, 0.0089, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 04:31:39,185 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=40986.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:31:42,181 INFO [train2.py:809] (0/4) Epoch 11, batch 1150, loss[ctc_loss=0.11, att_loss=0.2735, loss=0.2408, over 17066.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008084, over 52.00 utterances.], tot_loss[ctc_loss=0.1059, att_loss=0.2524, loss=0.2231, over 3251978.79 frames. utt_duration=1257 frames, utt_pad_proportion=0.0544, over 10359.77 utterances.], batch size: 52, lr: 1.00e-02, grad_scale: 16.0 2023-03-08 04:31:56,055 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=40997.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:32:12,874 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41007.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:32:34,182 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.552e+02 3.086e+02 3.791e+02 8.496e+02, threshold=6.172e+02, percent-clipped=3.0 2023-03-08 04:32:43,653 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-03-08 04:32:56,296 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41034.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:33:02,226 INFO [train2.py:809] (0/4) Epoch 11, batch 1200, loss[ctc_loss=0.1045, att_loss=0.2628, loss=0.2311, over 17359.00 frames. utt_duration=1264 frames, utt_pad_proportion=0.007821, over 55.00 utterances.], tot_loss[ctc_loss=0.1069, att_loss=0.2525, loss=0.2234, over 3251766.90 frames. utt_duration=1237 frames, utt_pad_proportion=0.06112, over 10531.04 utterances.], batch size: 55, lr: 9.99e-03, grad_scale: 16.0 2023-03-08 04:33:08,962 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3042, 4.2098, 4.1728, 4.3039, 4.6867, 4.3496, 4.2742, 2.1815], device='cuda:0'), covar=tensor([0.0260, 0.0394, 0.0349, 0.0194, 0.1009, 0.0205, 0.0285, 0.2110], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0129, 0.0134, 0.0135, 0.0325, 0.0121, 0.0122, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 04:33:13,634 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41045.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:33:34,413 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41058.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:34:22,826 INFO [train2.py:809] (0/4) Epoch 11, batch 1250, loss[ctc_loss=0.07304, att_loss=0.2274, loss=0.1965, over 16283.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007097, over 43.00 utterances.], tot_loss[ctc_loss=0.1063, att_loss=0.2523, loss=0.2231, over 3258087.61 frames. utt_duration=1240 frames, utt_pad_proportion=0.05979, over 10525.21 utterances.], batch size: 43, lr: 9.99e-03, grad_scale: 16.0 2023-03-08 04:34:31,603 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.78 vs. limit=2.0 2023-03-08 04:34:45,092 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41102.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:34:51,804 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41106.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:34:57,099 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5692, 2.6305, 3.5442, 4.3218, 3.9371, 3.9603, 2.8241, 2.4122], device='cuda:0'), covar=tensor([0.0610, 0.2687, 0.0937, 0.0674, 0.0706, 0.0378, 0.1818, 0.2309], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0207, 0.0184, 0.0188, 0.0183, 0.0144, 0.0192, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 04:35:14,659 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.546e+02 2.278e+02 2.660e+02 3.579e+02 5.724e+02, threshold=5.320e+02, percent-clipped=1.0 2023-03-08 04:35:42,812 INFO [train2.py:809] (0/4) Epoch 11, batch 1300, loss[ctc_loss=0.1286, att_loss=0.2535, loss=0.2286, over 15635.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009573, over 37.00 utterances.], tot_loss[ctc_loss=0.1067, att_loss=0.2528, loss=0.2236, over 3264235.99 frames. utt_duration=1254 frames, utt_pad_proportion=0.05329, over 10421.51 utterances.], batch size: 37, lr: 9.98e-03, grad_scale: 16.0 2023-03-08 04:35:53,424 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 04:35:58,587 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41148.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:36:23,569 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41163.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:37:03,124 INFO [train2.py:809] (0/4) Epoch 11, batch 1350, loss[ctc_loss=0.09783, att_loss=0.2279, loss=0.2019, over 14478.00 frames. utt_duration=1811 frames, utt_pad_proportion=0.0383, over 32.00 utterances.], tot_loss[ctc_loss=0.1053, att_loss=0.2509, loss=0.2218, over 3266300.21 frames. utt_duration=1288 frames, utt_pad_proportion=0.0456, over 10156.44 utterances.], batch size: 32, lr: 9.98e-03, grad_scale: 16.0 2023-03-08 04:37:15,569 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41196.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:37:55,022 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.568e+02 2.465e+02 2.945e+02 3.733e+02 9.191e+02, threshold=5.889e+02, percent-clipped=5.0 2023-03-08 04:37:56,864 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41221.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:38:23,862 INFO [train2.py:809] (0/4) Epoch 11, batch 1400, loss[ctc_loss=0.1192, att_loss=0.2665, loss=0.237, over 17267.00 frames. utt_duration=1172 frames, utt_pad_proportion=0.02459, over 59.00 utterances.], tot_loss[ctc_loss=0.105, att_loss=0.251, loss=0.2218, over 3268801.19 frames. utt_duration=1273 frames, utt_pad_proportion=0.04877, over 10283.51 utterances.], batch size: 59, lr: 9.97e-03, grad_scale: 16.0 2023-03-08 04:38:44,241 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8306, 6.0608, 5.5644, 5.9047, 5.8083, 5.4326, 5.4375, 5.2530], device='cuda:0'), covar=tensor([0.1307, 0.0950, 0.0828, 0.0772, 0.0779, 0.1357, 0.2512, 0.2613], device='cuda:0'), in_proj_covar=tensor([0.0441, 0.0511, 0.0371, 0.0386, 0.0359, 0.0422, 0.0525, 0.0464], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 04:38:52,717 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41256.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 04:39:44,637 INFO [train2.py:809] (0/4) Epoch 11, batch 1450, loss[ctc_loss=0.1088, att_loss=0.2658, loss=0.2344, over 17315.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02389, over 59.00 utterances.], tot_loss[ctc_loss=0.1063, att_loss=0.2523, loss=0.2231, over 3276864.29 frames. utt_duration=1234 frames, utt_pad_proportion=0.05588, over 10631.29 utterances.], batch size: 59, lr: 9.96e-03, grad_scale: 16.0 2023-03-08 04:40:15,965 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41307.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:40:37,477 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.706e+02 2.447e+02 2.872e+02 3.322e+02 6.594e+02, threshold=5.743e+02, percent-clipped=2.0 2023-03-08 04:41:05,351 INFO [train2.py:809] (0/4) Epoch 11, batch 1500, loss[ctc_loss=0.06494, att_loss=0.2187, loss=0.188, over 15495.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.008423, over 36.00 utterances.], tot_loss[ctc_loss=0.1057, att_loss=0.2525, loss=0.2231, over 3280974.36 frames. utt_duration=1214 frames, utt_pad_proportion=0.05995, over 10825.10 utterances.], batch size: 36, lr: 9.96e-03, grad_scale: 16.0 2023-03-08 04:41:11,152 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-03-08 04:41:26,747 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2717, 2.5491, 3.5930, 2.7040, 3.4226, 4.5547, 4.2356, 3.0476], device='cuda:0'), covar=tensor([0.0460, 0.2089, 0.1057, 0.1650, 0.1152, 0.0617, 0.0630, 0.1494], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0224, 0.0239, 0.0204, 0.0238, 0.0295, 0.0214, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:41:30,357 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41353.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:41:33,378 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41355.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:42:25,613 INFO [train2.py:809] (0/4) Epoch 11, batch 1550, loss[ctc_loss=0.1027, att_loss=0.2408, loss=0.2132, over 15956.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006113, over 41.00 utterances.], tot_loss[ctc_loss=0.1049, att_loss=0.2517, loss=0.2223, over 3276509.00 frames. utt_duration=1233 frames, utt_pad_proportion=0.05599, over 10644.63 utterances.], batch size: 41, lr: 9.95e-03, grad_scale: 16.0 2023-03-08 04:42:37,929 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 04:42:47,304 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41401.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:43:11,482 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-03-08 04:43:19,828 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.785e+02 2.586e+02 2.982e+02 3.761e+02 9.691e+02, threshold=5.963e+02, percent-clipped=6.0 2023-03-08 04:43:45,719 INFO [train2.py:809] (0/4) Epoch 11, batch 1600, loss[ctc_loss=0.101, att_loss=0.2576, loss=0.2263, over 16320.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.00687, over 45.00 utterances.], tot_loss[ctc_loss=0.1061, att_loss=0.2529, loss=0.2236, over 3283667.13 frames. utt_duration=1230 frames, utt_pad_proportion=0.05521, over 10689.94 utterances.], batch size: 45, lr: 9.95e-03, grad_scale: 8.0 2023-03-08 04:44:00,691 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7420, 4.8146, 4.5127, 4.6565, 5.3057, 4.8605, 4.6704, 2.4652], device='cuda:0'), covar=tensor([0.0213, 0.0227, 0.0346, 0.0270, 0.0851, 0.0174, 0.0312, 0.2088], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0132, 0.0139, 0.0139, 0.0330, 0.0122, 0.0123, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 04:44:05,357 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5764, 4.5517, 4.3844, 4.5438, 5.0944, 4.6910, 4.4866, 2.2676], device='cuda:0'), covar=tensor([0.0209, 0.0279, 0.0346, 0.0194, 0.0894, 0.0186, 0.0306, 0.2078], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0132, 0.0139, 0.0139, 0.0330, 0.0122, 0.0124, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 04:44:17,907 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=41458.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:45:05,418 INFO [train2.py:809] (0/4) Epoch 11, batch 1650, loss[ctc_loss=0.08991, att_loss=0.2239, loss=0.1971, over 15381.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01056, over 35.00 utterances.], tot_loss[ctc_loss=0.1064, att_loss=0.2531, loss=0.2238, over 3285130.53 frames. utt_duration=1232 frames, utt_pad_proportion=0.05375, over 10675.28 utterances.], batch size: 35, lr: 9.94e-03, grad_scale: 8.0 2023-03-08 04:45:58,326 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.532e+02 2.866e+02 3.682e+02 7.453e+02, threshold=5.731e+02, percent-clipped=3.0 2023-03-08 04:45:58,607 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41521.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:46:25,295 INFO [train2.py:809] (0/4) Epoch 11, batch 1700, loss[ctc_loss=0.07564, att_loss=0.2446, loss=0.2108, over 16478.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.005216, over 46.00 utterances.], tot_loss[ctc_loss=0.1065, att_loss=0.2533, loss=0.2239, over 3282388.77 frames. utt_duration=1216 frames, utt_pad_proportion=0.05915, over 10808.23 utterances.], batch size: 46, lr: 9.93e-03, grad_scale: 8.0 2023-03-08 04:46:37,407 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0375, 5.0566, 4.9387, 2.3565, 2.0342, 2.8528, 2.7472, 3.9746], device='cuda:0'), covar=tensor([0.0692, 0.0182, 0.0187, 0.4198, 0.6047, 0.2608, 0.2388, 0.1554], device='cuda:0'), in_proj_covar=tensor([0.0334, 0.0220, 0.0236, 0.0212, 0.0351, 0.0337, 0.0230, 0.0356], device='cuda:0'), out_proj_covar=tensor([1.5103e-04, 8.2339e-05, 1.0119e-04, 9.4830e-05, 1.5383e-04, 1.3691e-04, 9.1074e-05, 1.5066e-04], device='cuda:0') 2023-03-08 04:46:55,402 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41556.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 04:47:08,788 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4855, 2.7324, 3.7542, 2.7522, 3.5599, 4.7922, 4.4953, 3.1919], device='cuda:0'), covar=tensor([0.0406, 0.1865, 0.0979, 0.1616, 0.0994, 0.0547, 0.0431, 0.1449], device='cuda:0'), in_proj_covar=tensor([0.0227, 0.0225, 0.0240, 0.0206, 0.0240, 0.0298, 0.0213, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:47:16,309 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41569.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:47:46,032 INFO [train2.py:809] (0/4) Epoch 11, batch 1750, loss[ctc_loss=0.09786, att_loss=0.2678, loss=0.2338, over 17335.00 frames. utt_duration=1177 frames, utt_pad_proportion=0.02185, over 59.00 utterances.], tot_loss[ctc_loss=0.1063, att_loss=0.2535, loss=0.2241, over 3290237.73 frames. utt_duration=1221 frames, utt_pad_proportion=0.05446, over 10793.81 utterances.], batch size: 59, lr: 9.93e-03, grad_scale: 8.0 2023-03-08 04:48:12,437 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41604.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:48:39,824 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.573e+02 2.288e+02 2.927e+02 3.480e+02 6.791e+02, threshold=5.853e+02, percent-clipped=3.0 2023-03-08 04:49:06,234 INFO [train2.py:809] (0/4) Epoch 11, batch 1800, loss[ctc_loss=0.122, att_loss=0.2573, loss=0.2303, over 17432.00 frames. utt_duration=884.1 frames, utt_pad_proportion=0.07232, over 79.00 utterances.], tot_loss[ctc_loss=0.1053, att_loss=0.2525, loss=0.2231, over 3288897.89 frames. utt_duration=1242 frames, utt_pad_proportion=0.04909, over 10603.80 utterances.], batch size: 79, lr: 9.92e-03, grad_scale: 8.0 2023-03-08 04:49:31,685 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41653.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:49:33,264 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41654.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:50:10,487 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5424, 2.8124, 3.7412, 2.8515, 3.6113, 4.7163, 4.4778, 3.4110], device='cuda:0'), covar=tensor([0.0427, 0.1815, 0.0930, 0.1498, 0.0959, 0.0677, 0.0484, 0.1255], device='cuda:0'), in_proj_covar=tensor([0.0231, 0.0226, 0.0242, 0.0206, 0.0241, 0.0301, 0.0214, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:50:27,076 INFO [train2.py:809] (0/4) Epoch 11, batch 1850, loss[ctc_loss=0.1097, att_loss=0.2651, loss=0.234, over 16635.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004796, over 47.00 utterances.], tot_loss[ctc_loss=0.1037, att_loss=0.2509, loss=0.2215, over 3280515.77 frames. utt_duration=1274 frames, utt_pad_proportion=0.04398, over 10314.87 utterances.], batch size: 47, lr: 9.92e-03, grad_scale: 8.0 2023-03-08 04:50:48,878 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:50:49,077 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41701.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:51:12,094 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41715.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 04:51:20,774 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.644e+02 2.415e+02 2.840e+02 3.452e+02 8.289e+02, threshold=5.680e+02, percent-clipped=4.0 2023-03-08 04:51:46,880 INFO [train2.py:809] (0/4) Epoch 11, batch 1900, loss[ctc_loss=0.1456, att_loss=0.2736, loss=0.248, over 14216.00 frames. utt_duration=393.6 frames, utt_pad_proportion=0.3155, over 145.00 utterances.], tot_loss[ctc_loss=0.103, att_loss=0.2507, loss=0.2212, over 3281999.82 frames. utt_duration=1265 frames, utt_pad_proportion=0.04684, over 10386.19 utterances.], batch size: 145, lr: 9.91e-03, grad_scale: 8.0 2023-03-08 04:52:04,771 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41749.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:52:19,241 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=41758.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:52:26,696 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7944, 5.1318, 5.1250, 5.0406, 5.2051, 5.1422, 4.8439, 4.6428], device='cuda:0'), covar=tensor([0.1098, 0.0470, 0.0249, 0.0453, 0.0282, 0.0326, 0.0335, 0.0364], device='cuda:0'), in_proj_covar=tensor([0.0451, 0.0282, 0.0238, 0.0278, 0.0336, 0.0354, 0.0284, 0.0316], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 04:53:07,400 INFO [train2.py:809] (0/4) Epoch 11, batch 1950, loss[ctc_loss=0.08319, att_loss=0.2457, loss=0.2132, over 16268.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.008011, over 43.00 utterances.], tot_loss[ctc_loss=0.102, att_loss=0.2499, loss=0.2203, over 3277685.50 frames. utt_duration=1291 frames, utt_pad_proportion=0.04345, over 10166.17 utterances.], batch size: 43, lr: 9.91e-03, grad_scale: 8.0 2023-03-08 04:53:14,445 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4560, 2.6842, 3.8179, 2.9646, 3.4607, 4.7200, 4.6151, 3.0539], device='cuda:0'), covar=tensor([0.0554, 0.2283, 0.0877, 0.1644, 0.1111, 0.0834, 0.0435, 0.1905], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0225, 0.0242, 0.0204, 0.0242, 0.0299, 0.0212, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:53:23,945 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0027, 5.0522, 4.9644, 2.3705, 1.9238, 2.5814, 2.9358, 3.8103], device='cuda:0'), covar=tensor([0.0665, 0.0195, 0.0197, 0.4085, 0.6129, 0.2933, 0.2397, 0.1792], device='cuda:0'), in_proj_covar=tensor([0.0332, 0.0221, 0.0236, 0.0209, 0.0350, 0.0336, 0.0229, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.5006e-04, 8.2692e-05, 1.0111e-04, 9.3880e-05, 1.5322e-04, 1.3649e-04, 9.0572e-05, 1.5006e-04], device='cuda:0') 2023-03-08 04:53:36,829 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=41806.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:53:39,321 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0396, 5.3805, 5.3221, 5.3120, 5.4550, 5.3637, 5.0618, 4.8942], device='cuda:0'), covar=tensor([0.1086, 0.0467, 0.0268, 0.0429, 0.0261, 0.0293, 0.0344, 0.0336], device='cuda:0'), in_proj_covar=tensor([0.0455, 0.0286, 0.0241, 0.0280, 0.0339, 0.0358, 0.0288, 0.0318], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 04:54:00,604 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.550e+02 3.116e+02 4.012e+02 6.949e+02, threshold=6.232e+02, percent-clipped=6.0 2023-03-08 04:54:27,442 INFO [train2.py:809] (0/4) Epoch 11, batch 2000, loss[ctc_loss=0.1133, att_loss=0.2683, loss=0.2373, over 16760.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006204, over 48.00 utterances.], tot_loss[ctc_loss=0.1021, att_loss=0.2506, loss=0.2209, over 3278331.94 frames. utt_duration=1285 frames, utt_pad_proportion=0.0423, over 10217.39 utterances.], batch size: 48, lr: 9.90e-03, grad_scale: 8.0 2023-03-08 04:55:03,511 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9578, 5.0119, 4.8273, 2.5768, 4.7092, 4.6528, 4.0343, 2.0581], device='cuda:0'), covar=tensor([0.0155, 0.0109, 0.0251, 0.1413, 0.0134, 0.0192, 0.0513, 0.2423], device='cuda:0'), in_proj_covar=tensor([0.0060, 0.0082, 0.0075, 0.0103, 0.0069, 0.0094, 0.0092, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 04:55:29,701 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3275, 2.1816, 2.4142, 1.9203, 2.4965, 2.3607, 2.1186, 2.2789], device='cuda:0'), covar=tensor([0.1006, 0.4683, 0.4282, 0.2866, 0.1658, 0.1850, 0.3310, 0.1513], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0085, 0.0090, 0.0075, 0.0076, 0.0070, 0.0084, 0.0061], device='cuda:0'), out_proj_covar=tensor([4.9051e-05, 5.8660e-05, 6.0821e-05, 5.1077e-05, 4.9320e-05, 4.9219e-05, 5.7054e-05, 4.4103e-05], device='cuda:0') 2023-03-08 04:55:47,168 INFO [train2.py:809] (0/4) Epoch 11, batch 2050, loss[ctc_loss=0.0895, att_loss=0.2341, loss=0.2052, over 15888.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008988, over 39.00 utterances.], tot_loss[ctc_loss=0.1037, att_loss=0.2514, loss=0.2219, over 3278280.98 frames. utt_duration=1284 frames, utt_pad_proportion=0.04218, over 10228.30 utterances.], batch size: 39, lr: 9.89e-03, grad_scale: 8.0 2023-03-08 04:55:56,813 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6515, 3.9616, 3.8501, 3.9251, 3.9294, 3.7981, 3.0977, 3.8837], device='cuda:0'), covar=tensor([0.0132, 0.0120, 0.0139, 0.0080, 0.0089, 0.0112, 0.0578, 0.0211], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0073, 0.0088, 0.0053, 0.0059, 0.0070, 0.0092, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 04:56:17,846 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41906.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:56:37,756 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7360, 3.7239, 3.0349, 3.3151, 3.9216, 3.5416, 2.7240, 4.2266], device='cuda:0'), covar=tensor([0.1049, 0.0453, 0.1014, 0.0642, 0.0622, 0.0629, 0.0888, 0.0427], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0182, 0.0202, 0.0174, 0.0235, 0.0215, 0.0185, 0.0253], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 04:56:40,437 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.467e+02 2.492e+02 2.943e+02 3.583e+02 1.498e+03, threshold=5.886e+02, percent-clipped=5.0 2023-03-08 04:57:07,332 INFO [train2.py:809] (0/4) Epoch 11, batch 2100, loss[ctc_loss=0.098, att_loss=0.2314, loss=0.2047, over 14566.00 frames. utt_duration=1822 frames, utt_pad_proportion=0.03739, over 32.00 utterances.], tot_loss[ctc_loss=0.1043, att_loss=0.2517, loss=0.2222, over 3281604.93 frames. utt_duration=1262 frames, utt_pad_proportion=0.04667, over 10411.31 utterances.], batch size: 32, lr: 9.89e-03, grad_scale: 8.0 2023-03-08 04:57:23,931 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=41948.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 04:57:34,614 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9585, 2.0304, 1.9509, 1.6380, 2.2097, 2.1863, 2.0531, 2.7217], device='cuda:0'), covar=tensor([0.1139, 0.4035, 0.4567, 0.2620, 0.1669, 0.1656, 0.2704, 0.0891], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0086, 0.0089, 0.0076, 0.0075, 0.0069, 0.0084, 0.0061], device='cuda:0'), out_proj_covar=tensor([4.9129e-05, 5.8757e-05, 6.0686e-05, 5.1483e-05, 4.9160e-05, 4.8879e-05, 5.7329e-05, 4.4181e-05], device='cuda:0') 2023-03-08 04:57:54,578 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=41967.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 04:58:27,971 INFO [train2.py:809] (0/4) Epoch 11, batch 2150, loss[ctc_loss=0.07983, att_loss=0.2234, loss=0.1947, over 15363.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01114, over 35.00 utterances.], tot_loss[ctc_loss=0.1039, att_loss=0.2515, loss=0.222, over 3282538.08 frames. utt_duration=1264 frames, utt_pad_proportion=0.04783, over 10401.48 utterances.], batch size: 35, lr: 9.88e-03, grad_scale: 8.0 2023-03-08 04:58:48,382 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-42000.pt 2023-03-08 04:59:06,969 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42009.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 04:59:08,244 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42010.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 04:59:25,317 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.469e+02 2.816e+02 3.466e+02 1.172e+03, threshold=5.632e+02, percent-clipped=3.0 2023-03-08 04:59:26,323 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-08 04:59:52,146 INFO [train2.py:809] (0/4) Epoch 11, batch 2200, loss[ctc_loss=0.1312, att_loss=0.2471, loss=0.2239, over 15753.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009641, over 38.00 utterances.], tot_loss[ctc_loss=0.1031, att_loss=0.2511, loss=0.2215, over 3281249.24 frames. utt_duration=1266 frames, utt_pad_proportion=0.04724, over 10377.14 utterances.], batch size: 38, lr: 9.88e-03, grad_scale: 8.0 2023-03-08 05:00:29,746 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3629, 4.3380, 4.2924, 4.4116, 4.9092, 4.5890, 4.4698, 2.0741], device='cuda:0'), covar=tensor([0.0250, 0.0349, 0.0340, 0.0191, 0.0990, 0.0190, 0.0232, 0.2397], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0130, 0.0137, 0.0137, 0.0328, 0.0120, 0.0119, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 05:01:09,212 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1477, 2.3452, 3.0339, 4.0080, 3.7345, 3.8039, 2.6937, 2.0725], device='cuda:0'), covar=tensor([0.0743, 0.2361, 0.1088, 0.0585, 0.0758, 0.0373, 0.1463, 0.2327], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0200, 0.0182, 0.0185, 0.0184, 0.0147, 0.0190, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 05:01:12,042 INFO [train2.py:809] (0/4) Epoch 11, batch 2250, loss[ctc_loss=0.09948, att_loss=0.2495, loss=0.2195, over 16352.00 frames. utt_duration=1455 frames, utt_pad_proportion=0.004925, over 45.00 utterances.], tot_loss[ctc_loss=0.1022, att_loss=0.2507, loss=0.221, over 3283654.11 frames. utt_duration=1283 frames, utt_pad_proportion=0.04184, over 10248.83 utterances.], batch size: 45, lr: 9.87e-03, grad_scale: 8.0 2023-03-08 05:02:04,982 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.270e+02 2.801e+02 3.546e+02 6.959e+02, threshold=5.601e+02, percent-clipped=2.0 2023-03-08 05:02:32,245 INFO [train2.py:809] (0/4) Epoch 11, batch 2300, loss[ctc_loss=0.1076, att_loss=0.2591, loss=0.2288, over 16121.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006526, over 42.00 utterances.], tot_loss[ctc_loss=0.1017, att_loss=0.2501, loss=0.2204, over 3279522.67 frames. utt_duration=1265 frames, utt_pad_proportion=0.0472, over 10382.23 utterances.], batch size: 42, lr: 9.86e-03, grad_scale: 8.0 2023-03-08 05:03:03,955 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-03-08 05:03:52,293 INFO [train2.py:809] (0/4) Epoch 11, batch 2350, loss[ctc_loss=0.1309, att_loss=0.2795, loss=0.2498, over 17316.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02383, over 59.00 utterances.], tot_loss[ctc_loss=0.1017, att_loss=0.2503, loss=0.2206, over 3280582.51 frames. utt_duration=1282 frames, utt_pad_proportion=0.04268, over 10249.01 utterances.], batch size: 59, lr: 9.86e-03, grad_scale: 8.0 2023-03-08 05:04:45,693 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.909e+02 2.492e+02 2.954e+02 3.764e+02 8.189e+02, threshold=5.908e+02, percent-clipped=4.0 2023-03-08 05:05:13,513 INFO [train2.py:809] (0/4) Epoch 11, batch 2400, loss[ctc_loss=0.1054, att_loss=0.2566, loss=0.2264, over 17380.00 frames. utt_duration=1009 frames, utt_pad_proportion=0.04634, over 69.00 utterances.], tot_loss[ctc_loss=0.1022, att_loss=0.2509, loss=0.2212, over 3286270.64 frames. utt_duration=1276 frames, utt_pad_proportion=0.04272, over 10314.50 utterances.], batch size: 69, lr: 9.85e-03, grad_scale: 8.0 2023-03-08 05:05:23,886 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-03-08 05:05:52,515 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42262.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:06:34,468 INFO [train2.py:809] (0/4) Epoch 11, batch 2450, loss[ctc_loss=0.1381, att_loss=0.2693, loss=0.2431, over 16339.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005685, over 45.00 utterances.], tot_loss[ctc_loss=0.1026, att_loss=0.2509, loss=0.2212, over 3283725.05 frames. utt_duration=1263 frames, utt_pad_proportion=0.04677, over 10412.52 utterances.], batch size: 45, lr: 9.85e-03, grad_scale: 8.0 2023-03-08 05:06:59,952 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42304.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 05:07:09,507 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42310.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 05:07:26,991 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.597e+02 2.356e+02 2.915e+02 3.557e+02 7.393e+02, threshold=5.830e+02, percent-clipped=4.0 2023-03-08 05:07:54,611 INFO [train2.py:809] (0/4) Epoch 11, batch 2500, loss[ctc_loss=0.1256, att_loss=0.2544, loss=0.2286, over 16466.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006915, over 46.00 utterances.], tot_loss[ctc_loss=0.1034, att_loss=0.2513, loss=0.2217, over 3287625.80 frames. utt_duration=1254 frames, utt_pad_proportion=0.04878, over 10498.62 utterances.], batch size: 46, lr: 9.84e-03, grad_scale: 8.0 2023-03-08 05:08:26,082 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42358.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:08:35,806 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5418, 4.5309, 4.5466, 4.4513, 5.2438, 4.7597, 4.5603, 2.1518], device='cuda:0'), covar=tensor([0.0224, 0.0269, 0.0242, 0.0232, 0.0931, 0.0185, 0.0265, 0.2270], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0130, 0.0138, 0.0140, 0.0330, 0.0120, 0.0121, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 05:08:59,973 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42379.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:09:14,575 INFO [train2.py:809] (0/4) Epoch 11, batch 2550, loss[ctc_loss=0.1218, att_loss=0.259, loss=0.2316, over 16393.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.006072, over 44.00 utterances.], tot_loss[ctc_loss=0.1039, att_loss=0.2514, loss=0.2219, over 3289372.25 frames. utt_duration=1216 frames, utt_pad_proportion=0.05804, over 10830.71 utterances.], batch size: 44, lr: 9.84e-03, grad_scale: 8.0 2023-03-08 05:10:06,487 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.376e+02 2.981e+02 3.759e+02 6.397e+02, threshold=5.961e+02, percent-clipped=1.0 2023-03-08 05:10:34,787 INFO [train2.py:809] (0/4) Epoch 11, batch 2600, loss[ctc_loss=0.1166, att_loss=0.2707, loss=0.2398, over 16618.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005678, over 47.00 utterances.], tot_loss[ctc_loss=0.1035, att_loss=0.2514, loss=0.2218, over 3292718.74 frames. utt_duration=1242 frames, utt_pad_proportion=0.05214, over 10618.46 utterances.], batch size: 47, lr: 9.83e-03, grad_scale: 8.0 2023-03-08 05:10:38,259 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42440.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:10:53,587 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0678, 5.4140, 4.7446, 5.2318, 4.9708, 4.6326, 4.7758, 4.6600], device='cuda:0'), covar=tensor([0.1255, 0.0847, 0.0910, 0.0785, 0.0849, 0.1490, 0.2407, 0.2285], device='cuda:0'), in_proj_covar=tensor([0.0449, 0.0507, 0.0377, 0.0384, 0.0367, 0.0424, 0.0533, 0.0468], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 05:11:55,321 INFO [train2.py:809] (0/4) Epoch 11, batch 2650, loss[ctc_loss=0.0962, att_loss=0.2319, loss=0.2047, over 16125.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.00631, over 42.00 utterances.], tot_loss[ctc_loss=0.1029, att_loss=0.2504, loss=0.2209, over 3280422.68 frames. utt_duration=1235 frames, utt_pad_proportion=0.05731, over 10634.38 utterances.], batch size: 42, lr: 9.82e-03, grad_scale: 8.0 2023-03-08 05:12:23,764 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42506.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:12:47,504 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.508e+02 2.235e+02 2.761e+02 3.286e+02 8.509e+02, threshold=5.523e+02, percent-clipped=3.0 2023-03-08 05:13:15,973 INFO [train2.py:809] (0/4) Epoch 11, batch 2700, loss[ctc_loss=0.08732, att_loss=0.2453, loss=0.2137, over 16971.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007319, over 50.00 utterances.], tot_loss[ctc_loss=0.104, att_loss=0.251, loss=0.2216, over 3284024.12 frames. utt_duration=1229 frames, utt_pad_proportion=0.05769, over 10704.03 utterances.], batch size: 50, lr: 9.82e-03, grad_scale: 8.0 2023-03-08 05:13:19,584 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2622, 4.1375, 4.3113, 4.2742, 5.0459, 4.4079, 4.2920, 2.1330], device='cuda:0'), covar=tensor([0.0252, 0.0488, 0.0304, 0.0203, 0.0758, 0.0223, 0.0301, 0.2201], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0132, 0.0139, 0.0142, 0.0331, 0.0122, 0.0122, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 05:13:53,895 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42562.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:14:02,072 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42567.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:14:37,195 INFO [train2.py:809] (0/4) Epoch 11, batch 2750, loss[ctc_loss=0.09414, att_loss=0.2343, loss=0.2062, over 15638.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009399, over 37.00 utterances.], tot_loss[ctc_loss=0.1026, att_loss=0.2496, loss=0.2202, over 3284886.42 frames. utt_duration=1235 frames, utt_pad_proportion=0.05505, over 10652.64 utterances.], batch size: 37, lr: 9.81e-03, grad_scale: 8.0 2023-03-08 05:14:47,112 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.12 vs. limit=5.0 2023-03-08 05:15:02,399 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=42604.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 05:15:12,232 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42610.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:15:26,524 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-03-08 05:15:30,220 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.426e+02 2.955e+02 3.528e+02 6.434e+02, threshold=5.910e+02, percent-clipped=2.0 2023-03-08 05:15:57,768 INFO [train2.py:809] (0/4) Epoch 11, batch 2800, loss[ctc_loss=0.1055, att_loss=0.2549, loss=0.225, over 16689.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005773, over 46.00 utterances.], tot_loss[ctc_loss=0.102, att_loss=0.2494, loss=0.2199, over 3281803.57 frames. utt_duration=1248 frames, utt_pad_proportion=0.05206, over 10528.55 utterances.], batch size: 46, lr: 9.81e-03, grad_scale: 8.0 2023-03-08 05:16:15,457 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42649.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:16:19,836 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=42652.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 05:16:56,091 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2228, 4.8659, 4.4877, 5.0148, 2.4243, 4.5246, 2.4696, 1.7749], device='cuda:0'), covar=tensor([0.0269, 0.0108, 0.0696, 0.0093, 0.2165, 0.0156, 0.1769, 0.1948], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0111, 0.0256, 0.0107, 0.0218, 0.0108, 0.0221, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 05:17:17,561 INFO [train2.py:809] (0/4) Epoch 11, batch 2850, loss[ctc_loss=0.1155, att_loss=0.2722, loss=0.2408, over 17403.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03276, over 63.00 utterances.], tot_loss[ctc_loss=0.1027, att_loss=0.2497, loss=0.2203, over 3274730.67 frames. utt_duration=1238 frames, utt_pad_proportion=0.05491, over 10591.11 utterances.], batch size: 63, lr: 9.80e-03, grad_scale: 8.0 2023-03-08 05:17:48,560 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-03-08 05:17:52,612 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42710.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:18:10,594 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.499e+02 2.468e+02 2.915e+02 3.973e+02 1.035e+03, threshold=5.829e+02, percent-clipped=4.0 2023-03-08 05:18:13,468 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 05:18:33,529 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42735.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:18:38,144 INFO [train2.py:809] (0/4) Epoch 11, batch 2900, loss[ctc_loss=0.1013, att_loss=0.2669, loss=0.2338, over 17016.00 frames. utt_duration=1286 frames, utt_pad_proportion=0.01188, over 53.00 utterances.], tot_loss[ctc_loss=0.1024, att_loss=0.2499, loss=0.2204, over 3272977.78 frames. utt_duration=1228 frames, utt_pad_proportion=0.05757, over 10671.92 utterances.], batch size: 53, lr: 9.80e-03, grad_scale: 8.0 2023-03-08 05:19:58,447 INFO [train2.py:809] (0/4) Epoch 11, batch 2950, loss[ctc_loss=0.0961, att_loss=0.23, loss=0.2032, over 16026.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006134, over 40.00 utterances.], tot_loss[ctc_loss=0.1013, att_loss=0.2497, loss=0.22, over 3278327.17 frames. utt_duration=1249 frames, utt_pad_proportion=0.05253, over 10513.60 utterances.], batch size: 40, lr: 9.79e-03, grad_scale: 8.0 2023-03-08 05:20:52,199 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.678e+02 2.273e+02 2.678e+02 3.532e+02 7.141e+02, threshold=5.356e+02, percent-clipped=1.0 2023-03-08 05:20:56,988 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7286, 6.0404, 5.3160, 5.7389, 5.6368, 5.3333, 5.4237, 5.1824], device='cuda:0'), covar=tensor([0.1364, 0.0786, 0.0856, 0.0771, 0.0938, 0.1430, 0.2299, 0.2509], device='cuda:0'), in_proj_covar=tensor([0.0445, 0.0504, 0.0377, 0.0390, 0.0365, 0.0424, 0.0526, 0.0465], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 05:21:19,017 INFO [train2.py:809] (0/4) Epoch 11, batch 3000, loss[ctc_loss=0.1022, att_loss=0.2541, loss=0.2237, over 17473.00 frames. utt_duration=1014 frames, utt_pad_proportion=0.04293, over 69.00 utterances.], tot_loss[ctc_loss=0.1013, att_loss=0.2502, loss=0.2204, over 3286272.48 frames. utt_duration=1261 frames, utt_pad_proportion=0.04739, over 10435.95 utterances.], batch size: 69, lr: 9.78e-03, grad_scale: 8.0 2023-03-08 05:21:19,019 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 05:21:32,742 INFO [train2.py:843] (0/4) Epoch 11, validation: ctc_loss=0.04985, att_loss=0.2382, loss=0.2006, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 05:21:32,743 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 05:22:05,852 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42859.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:22:10,438 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=42862.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:22:37,879 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=42879.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:22:52,095 INFO [train2.py:809] (0/4) Epoch 11, batch 3050, loss[ctc_loss=0.1116, att_loss=0.2652, loss=0.2345, over 16870.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007449, over 49.00 utterances.], tot_loss[ctc_loss=0.1024, att_loss=0.2514, loss=0.2216, over 3295490.23 frames. utt_duration=1257 frames, utt_pad_proportion=0.04583, over 10498.30 utterances.], batch size: 49, lr: 9.78e-03, grad_scale: 8.0 2023-03-08 05:22:58,415 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1008, 5.3762, 5.6180, 5.4809, 5.5583, 6.0935, 5.2873, 6.1205], device='cuda:0'), covar=tensor([0.0694, 0.0576, 0.0700, 0.1054, 0.1744, 0.0725, 0.0456, 0.0735], device='cuda:0'), in_proj_covar=tensor([0.0724, 0.0432, 0.0495, 0.0572, 0.0755, 0.0504, 0.0407, 0.0497], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 05:23:27,270 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2556, 3.7395, 3.0420, 3.3365, 3.9029, 3.5665, 3.0187, 4.3318], device='cuda:0'), covar=tensor([0.0870, 0.0559, 0.1255, 0.0759, 0.0726, 0.0768, 0.0944, 0.0537], device='cuda:0'), in_proj_covar=tensor([0.0186, 0.0184, 0.0204, 0.0174, 0.0238, 0.0213, 0.0185, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 05:23:43,389 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42920.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:23:44,613 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.578e+02 2.391e+02 3.019e+02 3.669e+02 9.404e+02, threshold=6.038e+02, percent-clipped=6.0 2023-03-08 05:24:11,644 INFO [train2.py:809] (0/4) Epoch 11, batch 3100, loss[ctc_loss=0.115, att_loss=0.2588, loss=0.2301, over 17401.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03272, over 63.00 utterances.], tot_loss[ctc_loss=0.1022, att_loss=0.2511, loss=0.2213, over 3298400.09 frames. utt_duration=1282 frames, utt_pad_proportion=0.03987, over 10303.47 utterances.], batch size: 63, lr: 9.77e-03, grad_scale: 8.0 2023-03-08 05:24:15,757 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=42940.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:25:32,674 INFO [train2.py:809] (0/4) Epoch 11, batch 3150, loss[ctc_loss=0.09116, att_loss=0.2205, loss=0.1946, over 14502.00 frames. utt_duration=1814 frames, utt_pad_proportion=0.03616, over 32.00 utterances.], tot_loss[ctc_loss=0.1031, att_loss=0.2506, loss=0.2211, over 3278694.40 frames. utt_duration=1249 frames, utt_pad_proportion=0.05379, over 10510.68 utterances.], batch size: 32, lr: 9.77e-03, grad_scale: 8.0 2023-03-08 05:26:00,846 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43005.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:26:26,642 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.340e+02 2.878e+02 3.301e+02 6.036e+02, threshold=5.756e+02, percent-clipped=0.0 2023-03-08 05:26:49,543 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43035.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:26:53,985 INFO [train2.py:809] (0/4) Epoch 11, batch 3200, loss[ctc_loss=0.06089, att_loss=0.215, loss=0.1842, over 14479.00 frames. utt_duration=1811 frames, utt_pad_proportion=0.04467, over 32.00 utterances.], tot_loss[ctc_loss=0.1021, att_loss=0.2497, loss=0.2202, over 3272524.01 frames. utt_duration=1260 frames, utt_pad_proportion=0.05359, over 10402.27 utterances.], batch size: 32, lr: 9.76e-03, grad_scale: 8.0 2023-03-08 05:27:09,172 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5745, 2.3397, 4.9304, 3.7696, 2.7059, 4.2077, 4.8353, 4.5469], device='cuda:0'), covar=tensor([0.0224, 0.1988, 0.0141, 0.1118, 0.2258, 0.0268, 0.0108, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0247, 0.0138, 0.0310, 0.0278, 0.0190, 0.0122, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 05:28:06,589 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43083.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:28:13,492 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9806, 6.2677, 5.6146, 5.9975, 5.8470, 5.4948, 5.6315, 5.4366], device='cuda:0'), covar=tensor([0.1308, 0.0856, 0.0811, 0.0811, 0.0726, 0.1251, 0.2425, 0.2501], device='cuda:0'), in_proj_covar=tensor([0.0447, 0.0506, 0.0377, 0.0384, 0.0364, 0.0421, 0.0523, 0.0462], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 05:28:14,933 INFO [train2.py:809] (0/4) Epoch 11, batch 3250, loss[ctc_loss=0.08109, att_loss=0.2253, loss=0.1965, over 15511.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.008051, over 36.00 utterances.], tot_loss[ctc_loss=0.1018, att_loss=0.2493, loss=0.2198, over 3273328.91 frames. utt_duration=1267 frames, utt_pad_proportion=0.05235, over 10346.01 utterances.], batch size: 36, lr: 9.76e-03, grad_scale: 8.0 2023-03-08 05:29:07,588 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.575e+02 2.382e+02 2.901e+02 3.870e+02 8.262e+02, threshold=5.801e+02, percent-clipped=4.0 2023-03-08 05:29:35,237 INFO [train2.py:809] (0/4) Epoch 11, batch 3300, loss[ctc_loss=0.1424, att_loss=0.2749, loss=0.2484, over 16961.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007772, over 50.00 utterances.], tot_loss[ctc_loss=0.1024, att_loss=0.25, loss=0.2205, over 3278514.78 frames. utt_duration=1245 frames, utt_pad_proportion=0.05649, over 10545.12 utterances.], batch size: 50, lr: 9.75e-03, grad_scale: 8.0 2023-03-08 05:29:43,209 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8729, 4.8854, 4.7526, 2.7349, 4.7276, 4.5475, 4.1096, 2.3522], device='cuda:0'), covar=tensor([0.0191, 0.0105, 0.0256, 0.1442, 0.0110, 0.0200, 0.0485, 0.2171], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0084, 0.0077, 0.0106, 0.0071, 0.0096, 0.0094, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 05:30:09,711 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7930, 2.4257, 4.0015, 3.4978, 2.9189, 3.6977, 3.8890, 3.7872], device='cuda:0'), covar=tensor([0.0250, 0.1449, 0.0117, 0.0859, 0.1455, 0.0283, 0.0121, 0.0252], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0245, 0.0137, 0.0307, 0.0275, 0.0188, 0.0121, 0.0157], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 05:30:12,614 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43162.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:30:55,499 INFO [train2.py:809] (0/4) Epoch 11, batch 3350, loss[ctc_loss=0.1174, att_loss=0.2748, loss=0.2433, over 17170.00 frames. utt_duration=870.9 frames, utt_pad_proportion=0.08616, over 79.00 utterances.], tot_loss[ctc_loss=0.1028, att_loss=0.2503, loss=0.2208, over 3272866.91 frames. utt_duration=1239 frames, utt_pad_proportion=0.05791, over 10575.40 utterances.], batch size: 79, lr: 9.75e-03, grad_scale: 8.0 2023-03-08 05:31:30,909 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43210.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:31:39,582 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43215.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:31:49,462 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.559e+02 2.322e+02 2.901e+02 3.775e+02 1.145e+03, threshold=5.802e+02, percent-clipped=9.0 2023-03-08 05:31:49,672 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6488, 5.9949, 5.2843, 5.7126, 5.6186, 5.2507, 5.3213, 5.2260], device='cuda:0'), covar=tensor([0.1328, 0.0933, 0.0891, 0.0763, 0.0844, 0.1313, 0.2609, 0.2283], device='cuda:0'), in_proj_covar=tensor([0.0449, 0.0509, 0.0381, 0.0388, 0.0367, 0.0423, 0.0531, 0.0466], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 05:32:12,671 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43235.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:32:17,048 INFO [train2.py:809] (0/4) Epoch 11, batch 3400, loss[ctc_loss=0.08686, att_loss=0.2283, loss=0.2, over 15892.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.00877, over 39.00 utterances.], tot_loss[ctc_loss=0.1021, att_loss=0.2502, loss=0.2206, over 3277259.90 frames. utt_duration=1250 frames, utt_pad_proportion=0.05431, over 10496.44 utterances.], batch size: 39, lr: 9.74e-03, grad_scale: 8.0 2023-03-08 05:33:37,402 INFO [train2.py:809] (0/4) Epoch 11, batch 3450, loss[ctc_loss=0.122, att_loss=0.2665, loss=0.2376, over 14551.00 frames. utt_duration=400.2 frames, utt_pad_proportion=0.3027, over 146.00 utterances.], tot_loss[ctc_loss=0.1033, att_loss=0.2507, loss=0.2212, over 3277018.50 frames. utt_duration=1206 frames, utt_pad_proportion=0.06409, over 10886.82 utterances.], batch size: 146, lr: 9.73e-03, grad_scale: 8.0 2023-03-08 05:34:03,772 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43305.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:34:29,498 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.748e+02 2.188e+02 2.793e+02 3.558e+02 5.760e+02, threshold=5.586e+02, percent-clipped=0.0 2023-03-08 05:34:57,078 INFO [train2.py:809] (0/4) Epoch 11, batch 3500, loss[ctc_loss=0.09294, att_loss=0.2348, loss=0.2064, over 15489.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.0096, over 36.00 utterances.], tot_loss[ctc_loss=0.1025, att_loss=0.2501, loss=0.2206, over 3278078.38 frames. utt_duration=1235 frames, utt_pad_proportion=0.05636, over 10626.02 utterances.], batch size: 36, lr: 9.73e-03, grad_scale: 8.0 2023-03-08 05:35:09,850 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0526, 5.4044, 4.8808, 5.5072, 4.8385, 5.0793, 5.5806, 5.2449], device='cuda:0'), covar=tensor([0.0515, 0.0229, 0.0771, 0.0200, 0.0384, 0.0220, 0.0207, 0.0188], device='cuda:0'), in_proj_covar=tensor([0.0335, 0.0260, 0.0316, 0.0255, 0.0265, 0.0203, 0.0247, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 05:35:20,536 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43353.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:36:17,029 INFO [train2.py:809] (0/4) Epoch 11, batch 3550, loss[ctc_loss=0.1217, att_loss=0.2749, loss=0.2442, over 17119.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01493, over 56.00 utterances.], tot_loss[ctc_loss=0.1018, att_loss=0.249, loss=0.2196, over 3269474.24 frames. utt_duration=1233 frames, utt_pad_proportion=0.05861, over 10616.36 utterances.], batch size: 56, lr: 9.72e-03, grad_scale: 4.0 2023-03-08 05:37:11,205 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.765e+02 2.366e+02 2.941e+02 3.430e+02 6.532e+02, threshold=5.882e+02, percent-clipped=4.0 2023-03-08 05:37:37,563 INFO [train2.py:809] (0/4) Epoch 11, batch 3600, loss[ctc_loss=0.1303, att_loss=0.2711, loss=0.2429, over 17045.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01012, over 53.00 utterances.], tot_loss[ctc_loss=0.101, att_loss=0.2485, loss=0.219, over 3265231.34 frames. utt_duration=1257 frames, utt_pad_proportion=0.05486, over 10400.77 utterances.], batch size: 53, lr: 9.72e-03, grad_scale: 8.0 2023-03-08 05:37:37,833 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0163, 5.0833, 4.9041, 2.6255, 4.8523, 4.6477, 4.3317, 2.3727], device='cuda:0'), covar=tensor([0.0201, 0.0086, 0.0234, 0.1368, 0.0088, 0.0202, 0.0342, 0.1873], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0085, 0.0077, 0.0106, 0.0071, 0.0096, 0.0094, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 05:38:49,377 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5170, 4.7748, 4.7907, 4.7400, 4.8456, 4.7825, 4.4652, 4.3049], device='cuda:0'), covar=tensor([0.1036, 0.0627, 0.0292, 0.0470, 0.0294, 0.0329, 0.0355, 0.0370], device='cuda:0'), in_proj_covar=tensor([0.0459, 0.0292, 0.0251, 0.0288, 0.0347, 0.0363, 0.0289, 0.0324], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 05:38:58,671 INFO [train2.py:809] (0/4) Epoch 11, batch 3650, loss[ctc_loss=0.07832, att_loss=0.2445, loss=0.2112, over 16382.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.00809, over 44.00 utterances.], tot_loss[ctc_loss=0.1014, att_loss=0.2494, loss=0.2198, over 3265843.25 frames. utt_duration=1236 frames, utt_pad_proportion=0.06048, over 10581.76 utterances.], batch size: 44, lr: 9.71e-03, grad_scale: 8.0 2023-03-08 05:39:07,307 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 05:39:20,210 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1636, 5.0130, 5.0658, 3.3278, 4.8548, 4.7427, 4.5312, 2.9207], device='cuda:0'), covar=tensor([0.0101, 0.0097, 0.0162, 0.0827, 0.0093, 0.0163, 0.0278, 0.1226], device='cuda:0'), in_proj_covar=tensor([0.0062, 0.0085, 0.0077, 0.0106, 0.0071, 0.0096, 0.0095, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 05:39:41,229 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43514.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:39:42,795 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43515.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:39:53,255 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.496e+02 2.334e+02 2.934e+02 3.698e+02 7.886e+02, threshold=5.869e+02, percent-clipped=6.0 2023-03-08 05:40:14,902 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=43535.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:40:19,099 INFO [train2.py:809] (0/4) Epoch 11, batch 3700, loss[ctc_loss=0.09206, att_loss=0.2581, loss=0.2249, over 16624.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005394, over 47.00 utterances.], tot_loss[ctc_loss=0.1007, att_loss=0.2488, loss=0.2192, over 3269389.93 frames. utt_duration=1256 frames, utt_pad_proportion=0.05521, over 10424.69 utterances.], batch size: 47, lr: 9.71e-03, grad_scale: 8.0 2023-03-08 05:40:59,608 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43563.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:41:19,047 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43575.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:41:30,961 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=43583.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:41:38,577 INFO [train2.py:809] (0/4) Epoch 11, batch 3750, loss[ctc_loss=0.09056, att_loss=0.2374, loss=0.208, over 15980.00 frames. utt_duration=1561 frames, utt_pad_proportion=0.005301, over 41.00 utterances.], tot_loss[ctc_loss=0.1015, att_loss=0.2493, loss=0.2197, over 3261997.99 frames. utt_duration=1234 frames, utt_pad_proportion=0.06211, over 10582.98 utterances.], batch size: 41, lr: 9.70e-03, grad_scale: 8.0 2023-03-08 05:42:32,237 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.570e+02 2.236e+02 2.829e+02 3.523e+02 7.340e+02, threshold=5.658e+02, percent-clipped=1.0 2023-03-08 05:42:58,803 INFO [train2.py:809] (0/4) Epoch 11, batch 3800, loss[ctc_loss=0.1162, att_loss=0.2573, loss=0.229, over 16417.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006785, over 44.00 utterances.], tot_loss[ctc_loss=0.1013, att_loss=0.2491, loss=0.2195, over 3267357.45 frames. utt_duration=1247 frames, utt_pad_proportion=0.05647, over 10492.69 utterances.], batch size: 44, lr: 9.70e-03, grad_scale: 8.0 2023-03-08 05:44:18,443 INFO [train2.py:809] (0/4) Epoch 11, batch 3850, loss[ctc_loss=0.07179, att_loss=0.2166, loss=0.1876, over 15886.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009284, over 39.00 utterances.], tot_loss[ctc_loss=0.1017, att_loss=0.2495, loss=0.22, over 3265008.73 frames. utt_duration=1240 frames, utt_pad_proportion=0.05893, over 10543.34 utterances.], batch size: 39, lr: 9.69e-03, grad_scale: 8.0 2023-03-08 05:44:39,952 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.2452, 5.5274, 5.7467, 5.6076, 5.7117, 6.1579, 5.1871, 6.2601], device='cuda:0'), covar=tensor([0.0650, 0.0629, 0.0769, 0.1161, 0.1872, 0.0892, 0.0552, 0.0694], device='cuda:0'), in_proj_covar=tensor([0.0740, 0.0441, 0.0510, 0.0575, 0.0763, 0.0516, 0.0412, 0.0498], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 05:45:10,391 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 2.441e+02 3.005e+02 4.132e+02 9.940e+02, threshold=6.009e+02, percent-clipped=5.0 2023-03-08 05:45:35,982 INFO [train2.py:809] (0/4) Epoch 11, batch 3900, loss[ctc_loss=0.09244, att_loss=0.2415, loss=0.2117, over 15934.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007222, over 41.00 utterances.], tot_loss[ctc_loss=0.1026, att_loss=0.2501, loss=0.2206, over 3256710.78 frames. utt_duration=1209 frames, utt_pad_proportion=0.06956, over 10789.34 utterances.], batch size: 41, lr: 9.69e-03, grad_scale: 8.0 2023-03-08 05:45:40,130 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.18 vs. limit=5.0 2023-03-08 05:46:00,964 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8968, 1.8411, 2.0492, 2.3108, 2.8123, 2.3343, 2.0132, 2.7311], device='cuda:0'), covar=tensor([0.1367, 0.4929, 0.3538, 0.3733, 0.1383, 0.2109, 0.4182, 0.1703], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0087, 0.0091, 0.0077, 0.0075, 0.0070, 0.0088, 0.0063], device='cuda:0'), out_proj_covar=tensor([5.2867e-05, 6.0815e-05, 6.2549e-05, 5.2675e-05, 4.9454e-05, 5.0335e-05, 6.0101e-05, 4.6394e-05], device='cuda:0') 2023-03-08 05:46:13,897 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.93 vs. limit=5.0 2023-03-08 05:46:44,184 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1970, 5.4598, 5.7537, 5.6685, 5.6576, 6.1175, 5.3010, 6.1879], device='cuda:0'), covar=tensor([0.0524, 0.0615, 0.0641, 0.0940, 0.1496, 0.0772, 0.0463, 0.0579], device='cuda:0'), in_proj_covar=tensor([0.0735, 0.0436, 0.0504, 0.0569, 0.0754, 0.0508, 0.0406, 0.0491], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 05:46:51,850 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43787.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:46:53,040 INFO [train2.py:809] (0/4) Epoch 11, batch 3950, loss[ctc_loss=0.1104, att_loss=0.2428, loss=0.2163, over 15631.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009152, over 37.00 utterances.], tot_loss[ctc_loss=0.1022, att_loss=0.2501, loss=0.2205, over 3265266.69 frames. utt_duration=1232 frames, utt_pad_proportion=0.06253, over 10613.46 utterances.], batch size: 37, lr: 9.68e-03, grad_scale: 8.0 2023-03-08 05:47:16,360 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8272, 6.0701, 5.5172, 5.7684, 5.7797, 5.3655, 5.4298, 5.2453], device='cuda:0'), covar=tensor([0.1135, 0.0850, 0.0760, 0.0695, 0.0730, 0.1363, 0.2209, 0.2114], device='cuda:0'), in_proj_covar=tensor([0.0445, 0.0507, 0.0379, 0.0382, 0.0362, 0.0423, 0.0519, 0.0455], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 05:47:27,455 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4777, 2.4031, 4.8872, 3.9028, 3.0354, 4.1914, 4.6283, 4.6199], device='cuda:0'), covar=tensor([0.0202, 0.1927, 0.0115, 0.0998, 0.1821, 0.0240, 0.0120, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0241, 0.0134, 0.0302, 0.0269, 0.0183, 0.0119, 0.0153], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 05:47:45,786 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-11.pt 2023-03-08 05:48:13,297 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.409e+02 2.914e+02 3.960e+02 1.049e+03, threshold=5.829e+02, percent-clipped=5.0 2023-03-08 05:48:13,345 INFO [train2.py:809] (0/4) Epoch 12, batch 0, loss[ctc_loss=0.08376, att_loss=0.2457, loss=0.2133, over 16617.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005887, over 47.00 utterances.], tot_loss[ctc_loss=0.08376, att_loss=0.2457, loss=0.2133, over 16617.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005887, over 47.00 utterances.], batch size: 47, lr: 9.27e-03, grad_scale: 8.0 2023-03-08 05:48:13,346 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 05:48:20,665 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1523, 4.1684, 4.0285, 2.3222, 3.9543, 4.0220, 3.5877, 2.4404], device='cuda:0'), covar=tensor([0.0152, 0.0130, 0.0242, 0.1180, 0.0139, 0.0215, 0.0375, 0.1476], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0086, 0.0078, 0.0107, 0.0072, 0.0097, 0.0095, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 05:48:25,650 INFO [train2.py:843] (0/4) Epoch 12, validation: ctc_loss=0.04785, att_loss=0.2384, loss=0.2003, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 05:48:25,650 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 05:48:54,244 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 05:49:09,120 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43848.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:49:12,396 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4229, 3.0346, 2.5302, 2.7710, 3.1358, 3.0306, 2.3836, 3.0123], device='cuda:0'), covar=tensor([0.1000, 0.0425, 0.0888, 0.0668, 0.0653, 0.0541, 0.0842, 0.0503], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0182, 0.0202, 0.0173, 0.0236, 0.0208, 0.0181, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 05:49:45,199 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=43870.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:49:47,998 INFO [train2.py:809] (0/4) Epoch 12, batch 50, loss[ctc_loss=0.08011, att_loss=0.2111, loss=0.1849, over 14205.00 frames. utt_duration=1834 frames, utt_pad_proportion=0.04256, over 31.00 utterances.], tot_loss[ctc_loss=0.1022, att_loss=0.2509, loss=0.2211, over 741185.31 frames. utt_duration=1105 frames, utt_pad_proportion=0.08622, over 2686.23 utterances.], batch size: 31, lr: 9.26e-03, grad_scale: 8.0 2023-03-08 05:50:48,012 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43909.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:50:52,797 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7292, 3.7127, 3.0468, 3.2957, 3.8594, 3.4037, 2.6746, 4.2412], device='cuda:0'), covar=tensor([0.1068, 0.0429, 0.1029, 0.0665, 0.0597, 0.0661, 0.0851, 0.0420], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0182, 0.0201, 0.0172, 0.0236, 0.0207, 0.0181, 0.0247], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 05:51:09,635 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.511e+02 2.447e+02 2.836e+02 3.349e+02 4.947e+02, threshold=5.672e+02, percent-clipped=0.0 2023-03-08 05:51:09,678 INFO [train2.py:809] (0/4) Epoch 12, batch 100, loss[ctc_loss=0.09075, att_loss=0.2274, loss=0.2001, over 15768.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008154, over 38.00 utterances.], tot_loss[ctc_loss=0.1005, att_loss=0.2495, loss=0.2197, over 1300602.03 frames. utt_duration=1152 frames, utt_pad_proportion=0.07518, over 4523.62 utterances.], batch size: 38, lr: 9.26e-03, grad_scale: 8.0 2023-03-08 05:52:09,298 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=43959.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:52:28,366 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=43970.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:52:31,099 INFO [train2.py:809] (0/4) Epoch 12, batch 150, loss[ctc_loss=0.07217, att_loss=0.2119, loss=0.1839, over 15992.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007471, over 40.00 utterances.], tot_loss[ctc_loss=0.0979, att_loss=0.2475, loss=0.2176, over 1739283.01 frames. utt_duration=1242 frames, utt_pad_proportion=0.0543, over 5607.56 utterances.], batch size: 40, lr: 9.25e-03, grad_scale: 8.0 2023-03-08 05:52:53,021 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6286, 1.9214, 2.0912, 2.1286, 3.1875, 2.1275, 1.9876, 2.5664], device='cuda:0'), covar=tensor([0.1353, 0.3913, 0.3183, 0.2100, 0.0813, 0.1894, 0.3050, 0.1345], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0087, 0.0092, 0.0078, 0.0075, 0.0070, 0.0089, 0.0064], device='cuda:0'), out_proj_covar=tensor([5.3590e-05, 6.0963e-05, 6.3301e-05, 5.3387e-05, 4.9932e-05, 5.0715e-05, 6.0868e-05, 4.6947e-05], device='cuda:0') 2023-03-08 05:53:16,465 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-44000.pt 2023-03-08 05:53:52,662 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44020.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:53:55,403 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.396e+02 2.379e+02 2.873e+02 3.449e+02 9.028e+02, threshold=5.746e+02, percent-clipped=3.0 2023-03-08 05:53:55,446 INFO [train2.py:809] (0/4) Epoch 12, batch 200, loss[ctc_loss=0.1239, att_loss=0.2677, loss=0.239, over 16811.00 frames. utt_duration=687.5 frames, utt_pad_proportion=0.1384, over 98.00 utterances.], tot_loss[ctc_loss=0.09833, att_loss=0.248, loss=0.2181, over 2080927.41 frames. utt_duration=1234 frames, utt_pad_proportion=0.05601, over 6752.07 utterances.], batch size: 98, lr: 9.25e-03, grad_scale: 8.0 2023-03-08 05:54:20,732 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8525, 3.8005, 3.6128, 3.0094, 3.6077, 3.7775, 3.6438, 2.6967], device='cuda:0'), covar=tensor([0.1157, 0.1928, 0.2672, 0.8253, 1.1003, 0.8239, 0.1074, 0.7281], device='cuda:0'), in_proj_covar=tensor([0.0102, 0.0123, 0.0136, 0.0204, 0.0105, 0.0187, 0.0107, 0.0176], device='cuda:0'), out_proj_covar=tensor([9.8501e-05, 1.0916e-04, 1.2193e-04, 1.6674e-04, 9.8410e-05, 1.5541e-04, 9.5477e-05, 1.4611e-04], device='cuda:0') 2023-03-08 05:55:14,810 INFO [train2.py:809] (0/4) Epoch 12, batch 250, loss[ctc_loss=0.07585, att_loss=0.2104, loss=0.1835, over 15869.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01017, over 39.00 utterances.], tot_loss[ctc_loss=0.09836, att_loss=0.2479, loss=0.218, over 2349383.25 frames. utt_duration=1228 frames, utt_pad_proportion=0.05646, over 7663.16 utterances.], batch size: 39, lr: 9.24e-03, grad_scale: 8.0 2023-03-08 05:56:24,152 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-03-08 05:56:34,282 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3063, 2.5191, 3.1466, 4.3399, 3.9733, 3.8276, 2.6498, 2.1616], device='cuda:0'), covar=tensor([0.0772, 0.2426, 0.1104, 0.0567, 0.0631, 0.0440, 0.1755, 0.2407], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0207, 0.0187, 0.0196, 0.0189, 0.0153, 0.0197, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 05:56:35,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.521e+02 2.301e+02 2.863e+02 3.737e+02 8.031e+02, threshold=5.726e+02, percent-clipped=4.0 2023-03-08 05:56:35,481 INFO [train2.py:809] (0/4) Epoch 12, batch 300, loss[ctc_loss=0.1159, att_loss=0.2655, loss=0.2356, over 17003.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.009469, over 51.00 utterances.], tot_loss[ctc_loss=0.09818, att_loss=0.2477, loss=0.2178, over 2554938.52 frames. utt_duration=1251 frames, utt_pad_proportion=0.05156, over 8182.11 utterances.], batch size: 51, lr: 9.24e-03, grad_scale: 8.0 2023-03-08 05:56:40,541 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7845, 5.1952, 4.6602, 5.3165, 4.6164, 4.9277, 5.3411, 5.1122], device='cuda:0'), covar=tensor([0.0599, 0.0270, 0.0825, 0.0237, 0.0416, 0.0204, 0.0212, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0347, 0.0268, 0.0325, 0.0264, 0.0275, 0.0210, 0.0256, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0005, 0.0006, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 05:56:47,131 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6640, 2.2666, 5.0751, 3.9615, 2.9712, 4.3775, 4.9435, 4.7130], device='cuda:0'), covar=tensor([0.0189, 0.1902, 0.0124, 0.0948, 0.1879, 0.0212, 0.0087, 0.0194], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0241, 0.0135, 0.0302, 0.0269, 0.0181, 0.0121, 0.0154], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 05:57:10,211 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44143.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:57:53,908 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44170.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:57:56,720 INFO [train2.py:809] (0/4) Epoch 12, batch 350, loss[ctc_loss=0.07759, att_loss=0.2208, loss=0.1921, over 15767.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008981, over 38.00 utterances.], tot_loss[ctc_loss=0.09754, att_loss=0.2476, loss=0.2176, over 2722365.38 frames. utt_duration=1274 frames, utt_pad_proportion=0.04414, over 8557.95 utterances.], batch size: 38, lr: 9.23e-03, grad_scale: 8.0 2023-03-08 05:57:57,628 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2023-03-08 05:59:10,315 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44218.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 05:59:17,105 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.261e+02 2.765e+02 3.364e+02 1.191e+03, threshold=5.531e+02, percent-clipped=3.0 2023-03-08 05:59:17,147 INFO [train2.py:809] (0/4) Epoch 12, batch 400, loss[ctc_loss=0.06859, att_loss=0.2111, loss=0.1826, over 15514.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.007939, over 36.00 utterances.], tot_loss[ctc_loss=0.09959, att_loss=0.2487, loss=0.2189, over 2842626.92 frames. utt_duration=1227 frames, utt_pad_proportion=0.05868, over 9280.34 utterances.], batch size: 36, lr: 9.23e-03, grad_scale: 8.0 2023-03-08 05:59:52,984 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-08 06:00:24,786 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44265.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:00:25,468 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-03-08 06:00:36,681 INFO [train2.py:809] (0/4) Epoch 12, batch 450, loss[ctc_loss=0.1011, att_loss=0.2328, loss=0.2065, over 15777.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008819, over 38.00 utterances.], tot_loss[ctc_loss=0.09882, att_loss=0.2481, loss=0.2182, over 2944432.63 frames. utt_duration=1256 frames, utt_pad_proportion=0.05017, over 9391.54 utterances.], batch size: 38, lr: 9.22e-03, grad_scale: 8.0 2023-03-08 06:01:28,121 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-08 06:01:44,257 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44315.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:01:56,075 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.632e+02 2.333e+02 2.897e+02 3.791e+02 1.244e+03, threshold=5.793e+02, percent-clipped=7.0 2023-03-08 06:01:56,121 INFO [train2.py:809] (0/4) Epoch 12, batch 500, loss[ctc_loss=0.08492, att_loss=0.2422, loss=0.2108, over 16560.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005132, over 45.00 utterances.], tot_loss[ctc_loss=0.09976, att_loss=0.2486, loss=0.2188, over 3020500.87 frames. utt_duration=1245 frames, utt_pad_proportion=0.05179, over 9712.90 utterances.], batch size: 45, lr: 9.22e-03, grad_scale: 8.0 2023-03-08 06:03:03,243 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-08 06:03:15,763 INFO [train2.py:809] (0/4) Epoch 12, batch 550, loss[ctc_loss=0.08485, att_loss=0.2606, loss=0.2255, over 16882.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006803, over 49.00 utterances.], tot_loss[ctc_loss=0.1007, att_loss=0.249, loss=0.2194, over 3071888.57 frames. utt_duration=1227 frames, utt_pad_proportion=0.05792, over 10026.69 utterances.], batch size: 49, lr: 9.21e-03, grad_scale: 8.0 2023-03-08 06:03:21,510 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 06:04:34,992 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.625e+02 2.230e+02 2.908e+02 3.570e+02 8.469e+02, threshold=5.817e+02, percent-clipped=3.0 2023-03-08 06:04:35,037 INFO [train2.py:809] (0/4) Epoch 12, batch 600, loss[ctc_loss=0.09313, att_loss=0.2629, loss=0.2289, over 17050.00 frames. utt_duration=1339 frames, utt_pad_proportion=0.006143, over 51.00 utterances.], tot_loss[ctc_loss=0.1004, att_loss=0.2489, loss=0.2192, over 3116810.58 frames. utt_duration=1219 frames, utt_pad_proportion=0.05996, over 10240.17 utterances.], batch size: 51, lr: 9.21e-03, grad_scale: 8.0 2023-03-08 06:05:08,606 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44443.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:05:53,593 INFO [train2.py:809] (0/4) Epoch 12, batch 650, loss[ctc_loss=0.09554, att_loss=0.2502, loss=0.2193, over 16481.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006579, over 46.00 utterances.], tot_loss[ctc_loss=0.101, att_loss=0.2492, loss=0.2195, over 3149146.29 frames. utt_duration=1237 frames, utt_pad_proportion=0.05729, over 10192.22 utterances.], batch size: 46, lr: 9.20e-03, grad_scale: 8.0 2023-03-08 06:06:12,152 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 06:06:23,431 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44491.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:07:12,354 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.481e+02 2.522e+02 3.198e+02 3.829e+02 8.235e+02, threshold=6.395e+02, percent-clipped=2.0 2023-03-08 06:07:12,400 INFO [train2.py:809] (0/4) Epoch 12, batch 700, loss[ctc_loss=0.1028, att_loss=0.2617, loss=0.2299, over 17374.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.0337, over 63.00 utterances.], tot_loss[ctc_loss=0.1015, att_loss=0.2501, loss=0.2204, over 3182169.95 frames. utt_duration=1231 frames, utt_pad_proportion=0.05686, over 10351.19 utterances.], batch size: 63, lr: 9.20e-03, grad_scale: 8.0 2023-03-08 06:07:20,657 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44527.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:07:57,854 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=44550.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:08:21,777 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44565.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:08:32,490 INFO [train2.py:809] (0/4) Epoch 12, batch 750, loss[ctc_loss=0.117, att_loss=0.2672, loss=0.2371, over 17296.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01232, over 55.00 utterances.], tot_loss[ctc_loss=0.1012, att_loss=0.2503, loss=0.2205, over 3214225.41 frames. utt_duration=1227 frames, utt_pad_proportion=0.0533, over 10491.17 utterances.], batch size: 55, lr: 9.19e-03, grad_scale: 8.0 2023-03-08 06:08:50,786 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9078, 3.7531, 3.1731, 3.4995, 4.0001, 3.7026, 2.8973, 4.2815], device='cuda:0'), covar=tensor([0.1007, 0.0447, 0.0971, 0.0641, 0.0566, 0.0531, 0.0819, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0186, 0.0206, 0.0177, 0.0241, 0.0213, 0.0184, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 06:08:58,575 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44588.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:09:34,893 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=44611.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:09:38,988 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44613.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:09:42,262 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=44615.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:09:52,664 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.593e+02 2.480e+02 2.940e+02 3.562e+02 6.927e+02, threshold=5.880e+02, percent-clipped=2.0 2023-03-08 06:09:52,707 INFO [train2.py:809] (0/4) Epoch 12, batch 800, loss[ctc_loss=0.09424, att_loss=0.2387, loss=0.2098, over 16279.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007158, over 43.00 utterances.], tot_loss[ctc_loss=0.1011, att_loss=0.2502, loss=0.2203, over 3229936.57 frames. utt_duration=1223 frames, utt_pad_proportion=0.05436, over 10573.11 utterances.], batch size: 43, lr: 9.19e-03, grad_scale: 8.0 2023-03-08 06:10:57,796 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=44663.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:11:12,232 INFO [train2.py:809] (0/4) Epoch 12, batch 850, loss[ctc_loss=0.0847, att_loss=0.2671, loss=0.2306, over 16767.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006466, over 48.00 utterances.], tot_loss[ctc_loss=0.09999, att_loss=0.2496, loss=0.2197, over 3243292.33 frames. utt_duration=1246 frames, utt_pad_proportion=0.04965, over 10422.54 utterances.], batch size: 48, lr: 9.18e-03, grad_scale: 8.0 2023-03-08 06:12:30,574 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.599e+02 2.371e+02 2.719e+02 3.452e+02 7.248e+02, threshold=5.437e+02, percent-clipped=1.0 2023-03-08 06:12:30,617 INFO [train2.py:809] (0/4) Epoch 12, batch 900, loss[ctc_loss=0.1174, att_loss=0.2726, loss=0.2416, over 17045.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009876, over 53.00 utterances.], tot_loss[ctc_loss=0.09959, att_loss=0.2491, loss=0.2192, over 3254547.25 frames. utt_duration=1268 frames, utt_pad_proportion=0.04423, over 10281.28 utterances.], batch size: 53, lr: 9.18e-03, grad_scale: 8.0 2023-03-08 06:13:36,854 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2023-03-08 06:13:49,827 INFO [train2.py:809] (0/4) Epoch 12, batch 950, loss[ctc_loss=0.1244, att_loss=0.277, loss=0.2465, over 17325.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02317, over 59.00 utterances.], tot_loss[ctc_loss=0.1001, att_loss=0.2496, loss=0.2197, over 3257698.59 frames. utt_duration=1230 frames, utt_pad_proportion=0.05542, over 10602.79 utterances.], batch size: 59, lr: 9.17e-03, grad_scale: 8.0 2023-03-08 06:13:50,030 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0046, 5.3634, 5.5422, 5.4360, 5.4565, 5.9678, 5.2303, 6.0927], device='cuda:0'), covar=tensor([0.0690, 0.0653, 0.0744, 0.1265, 0.1845, 0.0940, 0.0602, 0.0602], device='cuda:0'), in_proj_covar=tensor([0.0729, 0.0429, 0.0499, 0.0565, 0.0752, 0.0515, 0.0405, 0.0490], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:14:46,459 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-03-08 06:14:47,548 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7092, 1.7711, 1.8202, 2.1132, 2.2799, 2.5485, 1.5763, 2.6084], device='cuda:0'), covar=tensor([0.1526, 0.4148, 0.3575, 0.1349, 0.2387, 0.1284, 0.2471, 0.1608], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0085, 0.0091, 0.0075, 0.0077, 0.0067, 0.0090, 0.0063], device='cuda:0'), out_proj_covar=tensor([5.2595e-05, 5.9975e-05, 6.2862e-05, 5.1801e-05, 5.0640e-05, 4.9411e-05, 6.1240e-05, 4.6486e-05], device='cuda:0') 2023-03-08 06:15:09,799 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.480e+02 3.005e+02 3.455e+02 6.503e+02, threshold=6.009e+02, percent-clipped=2.0 2023-03-08 06:15:09,842 INFO [train2.py:809] (0/4) Epoch 12, batch 1000, loss[ctc_loss=0.07117, att_loss=0.2283, loss=0.1968, over 16123.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006556, over 42.00 utterances.], tot_loss[ctc_loss=0.1003, att_loss=0.2498, loss=0.2199, over 3257256.86 frames. utt_duration=1207 frames, utt_pad_proportion=0.06446, over 10804.94 utterances.], batch size: 42, lr: 9.17e-03, grad_scale: 8.0 2023-03-08 06:16:28,101 INFO [train2.py:809] (0/4) Epoch 12, batch 1050, loss[ctc_loss=0.0859, att_loss=0.2154, loss=0.1895, over 15635.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009621, over 37.00 utterances.], tot_loss[ctc_loss=0.1009, att_loss=0.2499, loss=0.2201, over 3264089.35 frames. utt_duration=1221 frames, utt_pad_proportion=0.06017, over 10706.53 utterances.], batch size: 37, lr: 9.16e-03, grad_scale: 8.0 2023-03-08 06:16:39,948 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8374, 5.2116, 5.3717, 5.1774, 5.3195, 5.7882, 5.0275, 5.8841], device='cuda:0'), covar=tensor([0.0588, 0.0601, 0.0668, 0.1084, 0.1524, 0.0776, 0.0670, 0.0628], device='cuda:0'), in_proj_covar=tensor([0.0732, 0.0430, 0.0501, 0.0566, 0.0754, 0.0517, 0.0410, 0.0493], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:16:46,084 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44883.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:17:22,292 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=44906.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:17:43,632 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-08 06:17:47,211 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.611e+02 2.358e+02 3.023e+02 3.796e+02 1.133e+03, threshold=6.045e+02, percent-clipped=5.0 2023-03-08 06:17:47,254 INFO [train2.py:809] (0/4) Epoch 12, batch 1100, loss[ctc_loss=0.1064, att_loss=0.2437, loss=0.2162, over 15866.00 frames. utt_duration=1628 frames, utt_pad_proportion=0.01064, over 39.00 utterances.], tot_loss[ctc_loss=0.1006, att_loss=0.2486, loss=0.219, over 3255851.95 frames. utt_duration=1235 frames, utt_pad_proportion=0.06023, over 10561.16 utterances.], batch size: 39, lr: 9.16e-03, grad_scale: 8.0 2023-03-08 06:18:04,689 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.28 vs. limit=5.0 2023-03-08 06:19:06,840 INFO [train2.py:809] (0/4) Epoch 12, batch 1150, loss[ctc_loss=0.1062, att_loss=0.2641, loss=0.2326, over 16958.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007339, over 50.00 utterances.], tot_loss[ctc_loss=0.09979, att_loss=0.2484, loss=0.2187, over 3260056.35 frames. utt_duration=1240 frames, utt_pad_proportion=0.0591, over 10526.70 utterances.], batch size: 50, lr: 9.15e-03, grad_scale: 8.0 2023-03-08 06:20:06,696 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-08 06:20:27,105 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.801e+02 2.531e+02 3.116e+02 3.796e+02 1.058e+03, threshold=6.232e+02, percent-clipped=4.0 2023-03-08 06:20:27,147 INFO [train2.py:809] (0/4) Epoch 12, batch 1200, loss[ctc_loss=0.1674, att_loss=0.2938, loss=0.2685, over 13482.00 frames. utt_duration=370.8 frames, utt_pad_proportion=0.3541, over 146.00 utterances.], tot_loss[ctc_loss=0.09962, att_loss=0.2486, loss=0.2188, over 3257752.30 frames. utt_duration=1232 frames, utt_pad_proportion=0.06094, over 10589.90 utterances.], batch size: 146, lr: 9.15e-03, grad_scale: 8.0 2023-03-08 06:20:56,512 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4878, 5.0355, 4.6882, 4.9577, 5.0087, 4.7112, 3.5492, 4.9039], device='cuda:0'), covar=tensor([0.0110, 0.0092, 0.0128, 0.0065, 0.0074, 0.0106, 0.0622, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0070, 0.0086, 0.0052, 0.0058, 0.0069, 0.0090, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:21:06,269 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-08 06:21:11,678 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9769, 4.1462, 3.8022, 4.1694, 3.8129, 3.7373, 4.2271, 4.1030], device='cuda:0'), covar=tensor([0.0574, 0.0283, 0.0811, 0.0334, 0.0487, 0.0787, 0.0244, 0.0220], device='cuda:0'), in_proj_covar=tensor([0.0336, 0.0259, 0.0316, 0.0256, 0.0267, 0.0205, 0.0248, 0.0239], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 06:21:22,283 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45056.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:21:47,503 INFO [train2.py:809] (0/4) Epoch 12, batch 1250, loss[ctc_loss=0.07293, att_loss=0.2267, loss=0.1959, over 16170.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007512, over 41.00 utterances.], tot_loss[ctc_loss=0.09928, att_loss=0.2478, loss=0.2181, over 3259458.60 frames. utt_duration=1231 frames, utt_pad_proportion=0.06106, over 10603.05 utterances.], batch size: 41, lr: 9.14e-03, grad_scale: 8.0 2023-03-08 06:22:02,248 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45081.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:22:58,395 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45117.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 06:23:05,868 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.644e+02 2.248e+02 2.874e+02 3.449e+02 8.124e+02, threshold=5.748e+02, percent-clipped=2.0 2023-03-08 06:23:05,913 INFO [train2.py:809] (0/4) Epoch 12, batch 1300, loss[ctc_loss=0.128, att_loss=0.2673, loss=0.2394, over 16950.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008415, over 50.00 utterances.], tot_loss[ctc_loss=0.09939, att_loss=0.2483, loss=0.2185, over 3272055.34 frames. utt_duration=1255 frames, utt_pad_proportion=0.05236, over 10437.63 utterances.], batch size: 50, lr: 9.14e-03, grad_scale: 8.0 2023-03-08 06:23:37,480 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45142.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:23:41,113 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.01 vs. limit=2.0 2023-03-08 06:23:51,081 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9305, 2.5180, 3.3709, 2.6788, 3.3629, 4.2781, 4.0565, 2.9475], device='cuda:0'), covar=tensor([0.0499, 0.1937, 0.1058, 0.1369, 0.0954, 0.0700, 0.0580, 0.1344], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0231, 0.0251, 0.0209, 0.0241, 0.0310, 0.0223, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:24:09,811 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-03-08 06:24:24,488 INFO [train2.py:809] (0/4) Epoch 12, batch 1350, loss[ctc_loss=0.0948, att_loss=0.2305, loss=0.2034, over 16002.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007099, over 40.00 utterances.], tot_loss[ctc_loss=0.0988, att_loss=0.2473, loss=0.2176, over 3260310.04 frames. utt_duration=1271 frames, utt_pad_proportion=0.05071, over 10273.22 utterances.], batch size: 40, lr: 9.13e-03, grad_scale: 8.0 2023-03-08 06:24:42,130 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45183.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:24:53,058 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9022, 4.8685, 4.7925, 2.8734, 4.5189, 4.4429, 3.9763, 2.7909], device='cuda:0'), covar=tensor([0.0124, 0.0085, 0.0206, 0.1029, 0.0110, 0.0209, 0.0387, 0.1341], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0085, 0.0079, 0.0105, 0.0072, 0.0097, 0.0095, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 06:25:18,413 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45206.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:25:37,647 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-03-08 06:25:43,526 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.529e+02 2.386e+02 2.782e+02 3.300e+02 6.204e+02, threshold=5.564e+02, percent-clipped=3.0 2023-03-08 06:25:43,570 INFO [train2.py:809] (0/4) Epoch 12, batch 1400, loss[ctc_loss=0.1114, att_loss=0.2605, loss=0.2307, over 16835.00 frames. utt_duration=681.7 frames, utt_pad_proportion=0.1415, over 99.00 utterances.], tot_loss[ctc_loss=0.0984, att_loss=0.2468, loss=0.2171, over 3264069.01 frames. utt_duration=1281 frames, utt_pad_proportion=0.04771, over 10201.96 utterances.], batch size: 99, lr: 9.13e-03, grad_scale: 8.0 2023-03-08 06:25:58,450 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45231.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:26:34,482 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45254.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:27:04,205 INFO [train2.py:809] (0/4) Epoch 12, batch 1450, loss[ctc_loss=0.09225, att_loss=0.2562, loss=0.2234, over 17311.00 frames. utt_duration=1101 frames, utt_pad_proportion=0.0378, over 63.00 utterances.], tot_loss[ctc_loss=0.09913, att_loss=0.2478, loss=0.2181, over 3270599.40 frames. utt_duration=1260 frames, utt_pad_proportion=0.05138, over 10398.43 utterances.], batch size: 63, lr: 9.12e-03, grad_scale: 8.0 2023-03-08 06:28:04,196 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0439, 6.2436, 5.7140, 6.0101, 5.9579, 5.5243, 5.6767, 5.3190], device='cuda:0'), covar=tensor([0.1175, 0.0984, 0.0825, 0.0795, 0.0726, 0.1327, 0.2360, 0.2873], device='cuda:0'), in_proj_covar=tensor([0.0447, 0.0517, 0.0391, 0.0392, 0.0377, 0.0436, 0.0533, 0.0470], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 06:28:06,686 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-03-08 06:28:24,614 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.115e+02 2.356e+02 2.763e+02 3.457e+02 6.962e+02, threshold=5.526e+02, percent-clipped=5.0 2023-03-08 06:28:24,659 INFO [train2.py:809] (0/4) Epoch 12, batch 1500, loss[ctc_loss=0.07671, att_loss=0.2241, loss=0.1946, over 16266.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.008224, over 43.00 utterances.], tot_loss[ctc_loss=0.09888, att_loss=0.2475, loss=0.2178, over 3270963.31 frames. utt_duration=1248 frames, utt_pad_proportion=0.05407, over 10493.75 utterances.], batch size: 43, lr: 9.12e-03, grad_scale: 8.0 2023-03-08 06:28:36,247 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8243, 4.8128, 4.7446, 2.4013, 4.6694, 4.4572, 4.1387, 2.4355], device='cuda:0'), covar=tensor([0.0113, 0.0084, 0.0199, 0.1255, 0.0078, 0.0202, 0.0318, 0.1504], device='cuda:0'), in_proj_covar=tensor([0.0061, 0.0085, 0.0078, 0.0104, 0.0071, 0.0096, 0.0095, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0002, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 06:29:06,056 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45348.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 06:29:45,019 INFO [train2.py:809] (0/4) Epoch 12, batch 1550, loss[ctc_loss=0.1073, att_loss=0.2584, loss=0.2282, over 16609.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.005667, over 47.00 utterances.], tot_loss[ctc_loss=0.09884, att_loss=0.2478, loss=0.218, over 3277016.68 frames. utt_duration=1259 frames, utt_pad_proportion=0.05106, over 10426.90 utterances.], batch size: 47, lr: 9.11e-03, grad_scale: 8.0 2023-03-08 06:30:44,785 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45409.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 06:30:49,189 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45412.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 06:31:05,229 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.353e+02 2.825e+02 3.392e+02 6.329e+02, threshold=5.650e+02, percent-clipped=2.0 2023-03-08 06:31:05,272 INFO [train2.py:809] (0/4) Epoch 12, batch 1600, loss[ctc_loss=0.08444, att_loss=0.2507, loss=0.2175, over 17016.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008079, over 51.00 utterances.], tot_loss[ctc_loss=0.1003, att_loss=0.2497, loss=0.2198, over 3279631.69 frames. utt_duration=1218 frames, utt_pad_proportion=0.05995, over 10780.44 utterances.], batch size: 51, lr: 9.11e-03, grad_scale: 16.0 2023-03-08 06:31:28,616 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45437.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:32:07,549 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8297, 4.8655, 4.7537, 4.6690, 5.3869, 4.8910, 4.7964, 2.3973], device='cuda:0'), covar=tensor([0.0176, 0.0225, 0.0240, 0.0275, 0.1014, 0.0152, 0.0228, 0.2252], device='cuda:0'), in_proj_covar=tensor([0.0129, 0.0137, 0.0141, 0.0151, 0.0343, 0.0124, 0.0125, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 06:32:24,617 INFO [train2.py:809] (0/4) Epoch 12, batch 1650, loss[ctc_loss=0.1106, att_loss=0.2675, loss=0.2361, over 17213.00 frames. utt_duration=999.3 frames, utt_pad_proportion=0.05818, over 69.00 utterances.], tot_loss[ctc_loss=0.09939, att_loss=0.2491, loss=0.2191, over 3281535.99 frames. utt_duration=1254 frames, utt_pad_proportion=0.05114, over 10478.35 utterances.], batch size: 69, lr: 9.10e-03, grad_scale: 16.0 2023-03-08 06:33:36,847 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0808, 5.1341, 4.9832, 2.5590, 2.0293, 2.8926, 3.1002, 3.8573], device='cuda:0'), covar=tensor([0.0694, 0.0210, 0.0230, 0.3489, 0.5379, 0.2492, 0.2121, 0.1712], device='cuda:0'), in_proj_covar=tensor([0.0336, 0.0221, 0.0237, 0.0211, 0.0341, 0.0332, 0.0229, 0.0352], device='cuda:0'), out_proj_covar=tensor([1.5062e-04, 8.1669e-05, 1.0166e-04, 9.2914e-05, 1.4855e-04, 1.3390e-04, 9.0761e-05, 1.4792e-04], device='cuda:0') 2023-03-08 06:33:43,905 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.369e+02 2.877e+02 3.430e+02 6.900e+02, threshold=5.754e+02, percent-clipped=3.0 2023-03-08 06:33:43,948 INFO [train2.py:809] (0/4) Epoch 12, batch 1700, loss[ctc_loss=0.1149, att_loss=0.2666, loss=0.2362, over 16950.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008459, over 50.00 utterances.], tot_loss[ctc_loss=0.09942, att_loss=0.2486, loss=0.2188, over 3276064.56 frames. utt_duration=1253 frames, utt_pad_proportion=0.05453, over 10468.19 utterances.], batch size: 50, lr: 9.10e-03, grad_scale: 16.0 2023-03-08 06:33:45,848 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6796, 5.1617, 4.8794, 5.2888, 5.2329, 4.8834, 3.5028, 5.1278], device='cuda:0'), covar=tensor([0.0107, 0.0104, 0.0120, 0.0060, 0.0075, 0.0100, 0.0676, 0.0138], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0070, 0.0087, 0.0053, 0.0059, 0.0069, 0.0091, 0.0090], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:34:06,524 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9358, 3.8653, 3.6955, 3.2940, 3.8307, 3.8831, 3.6404, 2.6775], device='cuda:0'), covar=tensor([0.0925, 0.1200, 0.2143, 0.4939, 0.1607, 0.4410, 0.1015, 0.6665], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0125, 0.0135, 0.0208, 0.0109, 0.0190, 0.0113, 0.0179], device='cuda:0'), out_proj_covar=tensor([9.9991e-05, 1.1113e-04, 1.2279e-04, 1.7106e-04, 1.0279e-04, 1.5895e-04, 1.0102e-04, 1.4886e-04], device='cuda:0') 2023-03-08 06:35:04,018 INFO [train2.py:809] (0/4) Epoch 12, batch 1750, loss[ctc_loss=0.09282, att_loss=0.2467, loss=0.2159, over 16632.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004871, over 47.00 utterances.], tot_loss[ctc_loss=0.09882, att_loss=0.2475, loss=0.2178, over 3269744.30 frames. utt_duration=1275 frames, utt_pad_proportion=0.05102, over 10268.95 utterances.], batch size: 47, lr: 9.09e-03, grad_scale: 16.0 2023-03-08 06:35:17,511 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0514, 3.7477, 3.1280, 3.3607, 3.9541, 3.6597, 2.8902, 4.4326], device='cuda:0'), covar=tensor([0.1038, 0.0575, 0.1112, 0.0772, 0.0759, 0.0753, 0.0974, 0.0485], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0188, 0.0208, 0.0178, 0.0244, 0.0216, 0.0186, 0.0258], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 06:36:24,187 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.564e+02 2.369e+02 2.874e+02 3.424e+02 1.036e+03, threshold=5.747e+02, percent-clipped=2.0 2023-03-08 06:36:24,229 INFO [train2.py:809] (0/4) Epoch 12, batch 1800, loss[ctc_loss=0.1421, att_loss=0.2699, loss=0.2443, over 14516.00 frames. utt_duration=399.2 frames, utt_pad_proportion=0.3045, over 146.00 utterances.], tot_loss[ctc_loss=0.09917, att_loss=0.2476, loss=0.2179, over 3257543.29 frames. utt_duration=1253 frames, utt_pad_proportion=0.05877, over 10409.94 utterances.], batch size: 146, lr: 9.09e-03, grad_scale: 16.0 2023-03-08 06:36:52,420 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-03-08 06:37:43,871 INFO [train2.py:809] (0/4) Epoch 12, batch 1850, loss[ctc_loss=0.1128, att_loss=0.2559, loss=0.2273, over 16750.00 frames. utt_duration=1397 frames, utt_pad_proportion=0.006826, over 48.00 utterances.], tot_loss[ctc_loss=0.09851, att_loss=0.2483, loss=0.2184, over 3268994.42 frames. utt_duration=1259 frames, utt_pad_proportion=0.05298, over 10396.61 utterances.], batch size: 48, lr: 9.08e-03, grad_scale: 8.0 2023-03-08 06:38:28,502 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9874, 3.9169, 3.2794, 3.6537, 4.1358, 3.7315, 3.0388, 4.4975], device='cuda:0'), covar=tensor([0.1092, 0.0455, 0.0990, 0.0573, 0.0592, 0.0667, 0.0857, 0.0513], device='cuda:0'), in_proj_covar=tensor([0.0186, 0.0185, 0.0203, 0.0175, 0.0240, 0.0214, 0.0183, 0.0256], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 06:38:32,222 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45702.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:38:35,063 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=45704.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 06:38:41,726 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.51 vs. limit=2.0 2023-03-08 06:38:47,996 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45712.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 06:39:02,764 INFO [train2.py:809] (0/4) Epoch 12, batch 1900, loss[ctc_loss=0.06104, att_loss=0.2127, loss=0.1824, over 15372.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01115, over 35.00 utterances.], tot_loss[ctc_loss=0.09892, att_loss=0.2482, loss=0.2183, over 3263865.79 frames. utt_duration=1251 frames, utt_pad_proportion=0.05677, over 10449.31 utterances.], batch size: 35, lr: 9.08e-03, grad_scale: 8.0 2023-03-08 06:39:04,236 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.442e+02 2.369e+02 2.834e+02 3.743e+02 8.835e+02, threshold=5.667e+02, percent-clipped=7.0 2023-03-08 06:39:26,574 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=45737.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:40:02,959 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45760.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:40:07,743 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45763.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:40:21,061 INFO [train2.py:809] (0/4) Epoch 12, batch 1950, loss[ctc_loss=0.1115, att_loss=0.259, loss=0.2295, over 16323.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006642, over 45.00 utterances.], tot_loss[ctc_loss=0.09924, att_loss=0.2483, loss=0.2185, over 3257726.20 frames. utt_duration=1262 frames, utt_pad_proportion=0.05461, over 10335.40 utterances.], batch size: 45, lr: 9.07e-03, grad_scale: 8.0 2023-03-08 06:40:24,474 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45774.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:40:41,622 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=45785.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:41:41,125 INFO [train2.py:809] (0/4) Epoch 12, batch 2000, loss[ctc_loss=0.1019, att_loss=0.2457, loss=0.217, over 16190.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.00563, over 41.00 utterances.], tot_loss[ctc_loss=0.09952, att_loss=0.2483, loss=0.2186, over 3266209.18 frames. utt_duration=1272 frames, utt_pad_proportion=0.05083, over 10286.15 utterances.], batch size: 41, lr: 9.07e-03, grad_scale: 8.0 2023-03-08 06:41:42,586 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.549e+02 2.558e+02 3.002e+02 3.965e+02 8.183e+02, threshold=6.004e+02, percent-clipped=7.0 2023-03-08 06:42:02,184 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45835.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:42:37,465 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8886, 3.5794, 4.0013, 3.6152, 4.1112, 4.9711, 4.7493, 3.9255], device='cuda:0'), covar=tensor([0.0308, 0.1306, 0.0907, 0.0970, 0.0741, 0.0584, 0.0378, 0.0949], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0235, 0.0254, 0.0208, 0.0244, 0.0315, 0.0227, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:42:41,060 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=45859.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:43:00,499 INFO [train2.py:809] (0/4) Epoch 12, batch 2050, loss[ctc_loss=0.08359, att_loss=0.231, loss=0.2015, over 16173.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006765, over 41.00 utterances.], tot_loss[ctc_loss=0.09993, att_loss=0.2485, loss=0.2188, over 3270371.99 frames. utt_duration=1257 frames, utt_pad_proportion=0.05155, over 10420.84 utterances.], batch size: 41, lr: 9.06e-03, grad_scale: 8.0 2023-03-08 06:43:56,026 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0631, 4.4992, 3.8115, 4.6629, 2.1512, 4.3448, 2.3431, 2.3549], device='cuda:0'), covar=tensor([0.0358, 0.0163, 0.1066, 0.0139, 0.2208, 0.0215, 0.1886, 0.1594], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0116, 0.0256, 0.0112, 0.0219, 0.0109, 0.0224, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 06:44:18,296 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=45920.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:44:21,027 INFO [train2.py:809] (0/4) Epoch 12, batch 2100, loss[ctc_loss=0.07987, att_loss=0.2208, loss=0.1926, over 15649.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008656, over 37.00 utterances.], tot_loss[ctc_loss=0.1004, att_loss=0.2493, loss=0.2195, over 3269666.67 frames. utt_duration=1207 frames, utt_pad_proportion=0.06362, over 10849.07 utterances.], batch size: 37, lr: 9.06e-03, grad_scale: 8.0 2023-03-08 06:44:22,597 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 2.414e+02 3.055e+02 3.692e+02 7.937e+02, threshold=6.110e+02, percent-clipped=4.0 2023-03-08 06:45:41,045 INFO [train2.py:809] (0/4) Epoch 12, batch 2150, loss[ctc_loss=0.0769, att_loss=0.2283, loss=0.198, over 15962.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006638, over 41.00 utterances.], tot_loss[ctc_loss=0.09964, att_loss=0.2486, loss=0.2188, over 3271720.12 frames. utt_duration=1231 frames, utt_pad_proportion=0.05651, over 10642.49 utterances.], batch size: 41, lr: 9.05e-03, grad_scale: 8.0 2023-03-08 06:46:26,328 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-46000.pt 2023-03-08 06:46:37,253 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46004.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 06:47:04,791 INFO [train2.py:809] (0/4) Epoch 12, batch 2200, loss[ctc_loss=0.08392, att_loss=0.2447, loss=0.2125, over 16266.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.008255, over 43.00 utterances.], tot_loss[ctc_loss=0.09989, att_loss=0.2485, loss=0.2188, over 3256105.65 frames. utt_duration=1225 frames, utt_pad_proportion=0.06218, over 10642.68 utterances.], batch size: 43, lr: 9.05e-03, grad_scale: 8.0 2023-03-08 06:47:06,226 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.315e+02 2.803e+02 3.534e+02 8.357e+02, threshold=5.606e+02, percent-clipped=3.0 2023-03-08 06:47:53,387 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46052.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 06:48:02,424 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46058.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:48:23,928 INFO [train2.py:809] (0/4) Epoch 12, batch 2250, loss[ctc_loss=0.09695, att_loss=0.2596, loss=0.2271, over 17055.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009311, over 53.00 utterances.], tot_loss[ctc_loss=0.09915, att_loss=0.2482, loss=0.2184, over 3260433.73 frames. utt_duration=1238 frames, utt_pad_proportion=0.05793, over 10544.77 utterances.], batch size: 53, lr: 9.04e-03, grad_scale: 8.0 2023-03-08 06:49:42,480 INFO [train2.py:809] (0/4) Epoch 12, batch 2300, loss[ctc_loss=0.102, att_loss=0.2416, loss=0.2137, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007463, over 43.00 utterances.], tot_loss[ctc_loss=0.09959, att_loss=0.2488, loss=0.2189, over 3270535.95 frames. utt_duration=1242 frames, utt_pad_proportion=0.05482, over 10546.25 utterances.], batch size: 43, lr: 9.04e-03, grad_scale: 8.0 2023-03-08 06:49:44,022 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.648e+02 2.436e+02 3.067e+02 3.650e+02 8.781e+02, threshold=6.133e+02, percent-clipped=4.0 2023-03-08 06:49:55,616 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46130.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:50:38,428 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6039, 2.9472, 3.6089, 2.9815, 3.5098, 4.7085, 4.5000, 3.4152], device='cuda:0'), covar=tensor([0.0390, 0.1711, 0.1078, 0.1320, 0.1010, 0.0703, 0.0391, 0.1245], device='cuda:0'), in_proj_covar=tensor([0.0233, 0.0235, 0.0251, 0.0207, 0.0242, 0.0313, 0.0226, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:51:01,256 INFO [train2.py:809] (0/4) Epoch 12, batch 2350, loss[ctc_loss=0.1111, att_loss=0.2386, loss=0.2131, over 16279.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007249, over 43.00 utterances.], tot_loss[ctc_loss=0.09979, att_loss=0.2491, loss=0.2192, over 3272038.14 frames. utt_duration=1244 frames, utt_pad_proportion=0.05466, over 10536.59 utterances.], batch size: 43, lr: 9.03e-03, grad_scale: 8.0 2023-03-08 06:52:09,306 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46215.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:52:19,910 INFO [train2.py:809] (0/4) Epoch 12, batch 2400, loss[ctc_loss=0.06887, att_loss=0.2086, loss=0.1806, over 15360.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01183, over 35.00 utterances.], tot_loss[ctc_loss=0.1007, att_loss=0.2494, loss=0.2196, over 3267386.69 frames. utt_duration=1215 frames, utt_pad_proportion=0.06282, over 10768.35 utterances.], batch size: 35, lr: 9.03e-03, grad_scale: 8.0 2023-03-08 06:52:21,344 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.579e+02 2.431e+02 2.918e+02 3.900e+02 7.255e+02, threshold=5.836e+02, percent-clipped=5.0 2023-03-08 06:53:04,099 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-03-08 06:53:30,424 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7431, 4.5844, 4.7195, 4.5221, 5.1971, 4.7684, 4.6955, 2.6021], device='cuda:0'), covar=tensor([0.0186, 0.0305, 0.0201, 0.0273, 0.1199, 0.0189, 0.0231, 0.2058], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0139, 0.0139, 0.0151, 0.0343, 0.0124, 0.0127, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 06:53:39,240 INFO [train2.py:809] (0/4) Epoch 12, batch 2450, loss[ctc_loss=0.1529, att_loss=0.2815, loss=0.2558, over 13379.00 frames. utt_duration=370.6 frames, utt_pad_proportion=0.3554, over 145.00 utterances.], tot_loss[ctc_loss=0.1008, att_loss=0.2492, loss=0.2195, over 3261820.63 frames. utt_duration=1193 frames, utt_pad_proportion=0.06977, over 10949.92 utterances.], batch size: 145, lr: 9.02e-03, grad_scale: 8.0 2023-03-08 06:53:46,245 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46276.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:54:50,537 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4795, 4.9090, 4.8127, 4.7808, 4.9339, 4.5819, 3.2463, 4.7102], device='cuda:0'), covar=tensor([0.0132, 0.0143, 0.0129, 0.0105, 0.0090, 0.0122, 0.0891, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0072, 0.0088, 0.0054, 0.0060, 0.0071, 0.0093, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:54:58,103 INFO [train2.py:809] (0/4) Epoch 12, batch 2500, loss[ctc_loss=0.08255, att_loss=0.2264, loss=0.1977, over 15653.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.008213, over 37.00 utterances.], tot_loss[ctc_loss=0.1002, att_loss=0.2495, loss=0.2197, over 3275856.53 frames. utt_duration=1205 frames, utt_pad_proportion=0.0617, over 10883.43 utterances.], batch size: 37, lr: 9.02e-03, grad_scale: 8.0 2023-03-08 06:55:00,272 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.509e+02 2.157e+02 2.613e+02 3.362e+02 8.530e+02, threshold=5.227e+02, percent-clipped=3.0 2023-03-08 06:55:22,633 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46337.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:55:56,094 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46358.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:56:18,494 INFO [train2.py:809] (0/4) Epoch 12, batch 2550, loss[ctc_loss=0.0773, att_loss=0.2296, loss=0.1991, over 15875.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009876, over 39.00 utterances.], tot_loss[ctc_loss=0.1001, att_loss=0.2487, loss=0.219, over 3269671.14 frames. utt_duration=1223 frames, utt_pad_proportion=0.05844, over 10710.35 utterances.], batch size: 39, lr: 9.01e-03, grad_scale: 8.0 2023-03-08 06:56:24,846 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3903, 4.8371, 4.6793, 4.6152, 4.8260, 4.5084, 3.2057, 4.5970], device='cuda:0'), covar=tensor([0.0110, 0.0105, 0.0120, 0.0095, 0.0103, 0.0117, 0.0738, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0073, 0.0088, 0.0055, 0.0061, 0.0071, 0.0093, 0.0092], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 06:56:46,234 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2023-03-08 06:57:12,225 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46406.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:57:38,531 INFO [train2.py:809] (0/4) Epoch 12, batch 2600, loss[ctc_loss=0.1006, att_loss=0.2626, loss=0.2302, over 17312.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.01117, over 55.00 utterances.], tot_loss[ctc_loss=0.09951, att_loss=0.2482, loss=0.2185, over 3264958.68 frames. utt_duration=1222 frames, utt_pad_proportion=0.05998, over 10697.91 utterances.], batch size: 55, lr: 9.01e-03, grad_scale: 8.0 2023-03-08 06:57:39,864 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.782e+02 2.587e+02 3.130e+02 3.974e+02 7.202e+02, threshold=6.259e+02, percent-clipped=6.0 2023-03-08 06:57:51,741 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46430.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 06:58:58,111 INFO [train2.py:809] (0/4) Epoch 12, batch 2650, loss[ctc_loss=0.1146, att_loss=0.2533, loss=0.2256, over 15949.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006689, over 41.00 utterances.], tot_loss[ctc_loss=0.09956, att_loss=0.2485, loss=0.2187, over 3270337.06 frames. utt_duration=1236 frames, utt_pad_proportion=0.05518, over 10598.55 utterances.], batch size: 41, lr: 9.00e-03, grad_scale: 8.0 2023-03-08 06:59:03,489 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-03-08 06:59:07,501 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46478.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:00:06,992 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46515.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:00:18,202 INFO [train2.py:809] (0/4) Epoch 12, batch 2700, loss[ctc_loss=0.07496, att_loss=0.222, loss=0.1926, over 15509.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.007704, over 36.00 utterances.], tot_loss[ctc_loss=0.09892, att_loss=0.2485, loss=0.2185, over 3274693.10 frames. utt_duration=1241 frames, utt_pad_proportion=0.05305, over 10570.09 utterances.], batch size: 36, lr: 9.00e-03, grad_scale: 8.0 2023-03-08 07:00:19,662 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.175e+02 2.444e+02 2.933e+02 3.610e+02 6.469e+02, threshold=5.867e+02, percent-clipped=1.0 2023-03-08 07:00:32,300 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-08 07:01:23,485 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46563.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:01:37,868 INFO [train2.py:809] (0/4) Epoch 12, batch 2750, loss[ctc_loss=0.103, att_loss=0.257, loss=0.2262, over 17017.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.007846, over 51.00 utterances.], tot_loss[ctc_loss=0.09813, att_loss=0.2475, loss=0.2176, over 3265502.18 frames. utt_duration=1245 frames, utt_pad_proportion=0.05482, over 10504.28 utterances.], batch size: 51, lr: 9.00e-03, grad_scale: 8.0 2023-03-08 07:02:15,218 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46595.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:02:26,665 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46602.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:02:36,795 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.03 vs. limit=2.0 2023-03-08 07:02:57,841 INFO [train2.py:809] (0/4) Epoch 12, batch 2800, loss[ctc_loss=0.1226, att_loss=0.2649, loss=0.2365, over 17319.00 frames. utt_duration=1005 frames, utt_pad_proportion=0.05146, over 69.00 utterances.], tot_loss[ctc_loss=0.09774, att_loss=0.2479, loss=0.2178, over 3268901.60 frames. utt_duration=1250 frames, utt_pad_proportion=0.05424, over 10470.41 utterances.], batch size: 69, lr: 8.99e-03, grad_scale: 8.0 2023-03-08 07:02:59,235 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.251e+02 2.286e+02 2.805e+02 3.463e+02 9.248e+02, threshold=5.611e+02, percent-clipped=3.0 2023-03-08 07:03:14,273 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46632.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:03:52,934 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46656.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 07:04:04,653 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46663.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:04:18,469 INFO [train2.py:809] (0/4) Epoch 12, batch 2850, loss[ctc_loss=0.0996, att_loss=0.248, loss=0.2183, over 16775.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.005375, over 48.00 utterances.], tot_loss[ctc_loss=0.09777, att_loss=0.2478, loss=0.2178, over 3272311.43 frames. utt_duration=1254 frames, utt_pad_proportion=0.05234, over 10446.92 utterances.], batch size: 48, lr: 8.99e-03, grad_scale: 8.0 2023-03-08 07:05:19,997 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-08 07:05:39,963 INFO [train2.py:809] (0/4) Epoch 12, batch 2900, loss[ctc_loss=0.1111, att_loss=0.2675, loss=0.2362, over 17076.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.00831, over 53.00 utterances.], tot_loss[ctc_loss=0.09872, att_loss=0.2485, loss=0.2185, over 3275003.23 frames. utt_duration=1232 frames, utt_pad_proportion=0.05606, over 10642.69 utterances.], batch size: 53, lr: 8.98e-03, grad_scale: 8.0 2023-03-08 07:05:41,466 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.596e+02 2.133e+02 2.612e+02 3.383e+02 7.389e+02, threshold=5.223e+02, percent-clipped=3.0 2023-03-08 07:07:00,902 INFO [train2.py:809] (0/4) Epoch 12, batch 2950, loss[ctc_loss=0.06779, att_loss=0.21, loss=0.1816, over 15358.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01146, over 35.00 utterances.], tot_loss[ctc_loss=0.09834, att_loss=0.248, loss=0.218, over 3268197.91 frames. utt_duration=1237 frames, utt_pad_proportion=0.05874, over 10577.97 utterances.], batch size: 35, lr: 8.98e-03, grad_scale: 8.0 2023-03-08 07:07:33,709 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7756, 3.5570, 2.8652, 3.2252, 3.6393, 3.4018, 2.5416, 3.8858], device='cuda:0'), covar=tensor([0.1092, 0.0471, 0.1230, 0.0698, 0.0735, 0.0665, 0.1042, 0.0516], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0190, 0.0210, 0.0180, 0.0247, 0.0219, 0.0186, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 07:08:03,443 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=46810.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:08:23,019 INFO [train2.py:809] (0/4) Epoch 12, batch 3000, loss[ctc_loss=0.08704, att_loss=0.2382, loss=0.208, over 16172.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.00617, over 41.00 utterances.], tot_loss[ctc_loss=0.09809, att_loss=0.2484, loss=0.2183, over 3277381.92 frames. utt_duration=1247 frames, utt_pad_proportion=0.05454, over 10528.98 utterances.], batch size: 41, lr: 8.97e-03, grad_scale: 8.0 2023-03-08 07:08:23,022 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 07:08:37,753 INFO [train2.py:843] (0/4) Epoch 12, validation: ctc_loss=0.04782, att_loss=0.2378, loss=0.1998, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 07:08:37,754 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 07:08:39,277 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.287e+02 2.757e+02 3.651e+02 8.301e+02, threshold=5.514e+02, percent-clipped=2.0 2023-03-08 07:09:06,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-08 07:09:38,816 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 07:09:56,621 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=46871.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:09:57,797 INFO [train2.py:809] (0/4) Epoch 12, batch 3050, loss[ctc_loss=0.07738, att_loss=0.2124, loss=0.1854, over 15506.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.007991, over 36.00 utterances.], tot_loss[ctc_loss=0.09857, att_loss=0.2486, loss=0.2186, over 3274409.18 frames. utt_duration=1254 frames, utt_pad_proportion=0.05235, over 10459.77 utterances.], batch size: 36, lr: 8.97e-03, grad_scale: 8.0 2023-03-08 07:11:17,289 INFO [train2.py:809] (0/4) Epoch 12, batch 3100, loss[ctc_loss=0.1208, att_loss=0.2738, loss=0.2432, over 17004.00 frames. utt_duration=688.5 frames, utt_pad_proportion=0.1362, over 99.00 utterances.], tot_loss[ctc_loss=0.09758, att_loss=0.2474, loss=0.2174, over 3269665.03 frames. utt_duration=1285 frames, utt_pad_proportion=0.04604, over 10188.28 utterances.], batch size: 99, lr: 8.96e-03, grad_scale: 8.0 2023-03-08 07:11:18,788 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.489e+02 2.293e+02 2.705e+02 3.311e+02 9.468e+02, threshold=5.410e+02, percent-clipped=5.0 2023-03-08 07:11:32,943 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=46932.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:12:02,256 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46951.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 07:12:13,002 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=46958.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:12:36,180 INFO [train2.py:809] (0/4) Epoch 12, batch 3150, loss[ctc_loss=0.07975, att_loss=0.2314, loss=0.201, over 16176.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006504, over 41.00 utterances.], tot_loss[ctc_loss=0.09808, att_loss=0.2477, loss=0.2178, over 3266416.47 frames. utt_duration=1259 frames, utt_pad_proportion=0.05435, over 10387.22 utterances.], batch size: 41, lr: 8.96e-03, grad_scale: 8.0 2023-03-08 07:12:48,179 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=46980.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:13:55,727 INFO [train2.py:809] (0/4) Epoch 12, batch 3200, loss[ctc_loss=0.1148, att_loss=0.2522, loss=0.2247, over 16157.00 frames. utt_duration=1578 frames, utt_pad_proportion=0.008248, over 41.00 utterances.], tot_loss[ctc_loss=0.09808, att_loss=0.2477, loss=0.2177, over 3270859.55 frames. utt_duration=1257 frames, utt_pad_proportion=0.05267, over 10418.16 utterances.], batch size: 41, lr: 8.95e-03, grad_scale: 8.0 2023-03-08 07:13:57,255 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.636e+02 2.280e+02 2.660e+02 3.739e+02 9.855e+02, threshold=5.321e+02, percent-clipped=6.0 2023-03-08 07:15:10,321 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-03-08 07:15:16,019 INFO [train2.py:809] (0/4) Epoch 12, batch 3250, loss[ctc_loss=0.1218, att_loss=0.2565, loss=0.2296, over 16467.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006794, over 46.00 utterances.], tot_loss[ctc_loss=0.09795, att_loss=0.2479, loss=0.2179, over 3275348.49 frames. utt_duration=1258 frames, utt_pad_proportion=0.05094, over 10424.47 utterances.], batch size: 46, lr: 8.95e-03, grad_scale: 8.0 2023-03-08 07:15:21,545 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-08 07:15:49,738 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.16 vs. limit=2.0 2023-03-08 07:16:08,485 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-03-08 07:16:34,790 INFO [train2.py:809] (0/4) Epoch 12, batch 3300, loss[ctc_loss=0.1128, att_loss=0.2652, loss=0.2347, over 17122.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01485, over 56.00 utterances.], tot_loss[ctc_loss=0.09709, att_loss=0.2474, loss=0.2173, over 3277099.13 frames. utt_duration=1278 frames, utt_pad_proportion=0.0453, over 10266.79 utterances.], batch size: 56, lr: 8.94e-03, grad_scale: 8.0 2023-03-08 07:16:36,329 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.270e+02 2.752e+02 3.318e+02 7.003e+02, threshold=5.505e+02, percent-clipped=5.0 2023-03-08 07:17:43,750 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47166.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:17:53,902 INFO [train2.py:809] (0/4) Epoch 12, batch 3350, loss[ctc_loss=0.1039, att_loss=0.2597, loss=0.2286, over 16888.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006945, over 49.00 utterances.], tot_loss[ctc_loss=0.09635, att_loss=0.247, loss=0.2169, over 3277652.80 frames. utt_duration=1278 frames, utt_pad_proportion=0.045, over 10268.60 utterances.], batch size: 49, lr: 8.94e-03, grad_scale: 8.0 2023-03-08 07:18:06,819 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0781, 5.1195, 5.0606, 2.3168, 2.0297, 2.7665, 3.1357, 3.7476], device='cuda:0'), covar=tensor([0.0641, 0.0227, 0.0203, 0.4470, 0.5516, 0.2422, 0.2144, 0.1881], device='cuda:0'), in_proj_covar=tensor([0.0331, 0.0224, 0.0239, 0.0213, 0.0343, 0.0331, 0.0229, 0.0352], device='cuda:0'), out_proj_covar=tensor([1.4703e-04, 8.3443e-05, 1.0329e-04, 9.3627e-05, 1.4839e-04, 1.3332e-04, 9.0065e-05, 1.4770e-04], device='cuda:0') 2023-03-08 07:18:09,249 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-08 07:19:14,294 INFO [train2.py:809] (0/4) Epoch 12, batch 3400, loss[ctc_loss=0.08215, att_loss=0.2305, loss=0.2008, over 15894.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008739, over 39.00 utterances.], tot_loss[ctc_loss=0.09669, att_loss=0.2467, loss=0.2167, over 3266159.45 frames. utt_duration=1263 frames, utt_pad_proportion=0.05178, over 10352.33 utterances.], batch size: 39, lr: 8.93e-03, grad_scale: 8.0 2023-03-08 07:19:15,802 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.335e+02 2.724e+02 3.467e+02 7.189e+02, threshold=5.448e+02, percent-clipped=4.0 2023-03-08 07:19:36,549 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4121, 2.8374, 3.1689, 4.2482, 3.8320, 3.8556, 2.8498, 1.9156], device='cuda:0'), covar=tensor([0.0684, 0.2178, 0.1086, 0.0696, 0.0817, 0.0444, 0.1582, 0.2729], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0213, 0.0190, 0.0199, 0.0195, 0.0159, 0.0198, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 07:19:44,346 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47241.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:19:59,758 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47251.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:20:10,646 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47258.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:20:24,734 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3738, 2.8729, 3.2866, 4.2892, 3.9095, 3.8515, 2.9027, 1.9333], device='cuda:0'), covar=tensor([0.0748, 0.2187, 0.1084, 0.0694, 0.0758, 0.0468, 0.1600, 0.2772], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0212, 0.0188, 0.0197, 0.0193, 0.0159, 0.0197, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 07:20:33,847 INFO [train2.py:809] (0/4) Epoch 12, batch 3450, loss[ctc_loss=0.07905, att_loss=0.2285, loss=0.1986, over 16282.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.00666, over 43.00 utterances.], tot_loss[ctc_loss=0.09672, att_loss=0.2469, loss=0.2169, over 3254264.57 frames. utt_duration=1256 frames, utt_pad_proportion=0.05639, over 10378.56 utterances.], batch size: 43, lr: 8.93e-03, grad_scale: 8.0 2023-03-08 07:20:57,744 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6947, 2.1897, 2.4487, 1.7454, 2.4666, 2.1815, 2.2115, 2.0957], device='cuda:0'), covar=tensor([0.1594, 0.4528, 0.3560, 0.2793, 0.1789, 0.1925, 0.3317, 0.1674], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0086, 0.0092, 0.0075, 0.0079, 0.0071, 0.0091, 0.0062], device='cuda:0'), out_proj_covar=tensor([5.4509e-05, 6.1047e-05, 6.4327e-05, 5.2897e-05, 5.3215e-05, 5.2373e-05, 6.2536e-05, 4.7119e-05], device='cuda:0') 2023-03-08 07:21:15,924 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47299.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:21:20,712 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47302.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:21:26,639 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47306.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:21:53,258 INFO [train2.py:809] (0/4) Epoch 12, batch 3500, loss[ctc_loss=0.09213, att_loss=0.2488, loss=0.2174, over 16269.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007376, over 43.00 utterances.], tot_loss[ctc_loss=0.09727, att_loss=0.248, loss=0.2178, over 3264800.14 frames. utt_duration=1243 frames, utt_pad_proportion=0.05656, over 10516.20 utterances.], batch size: 43, lr: 8.92e-03, grad_scale: 8.0 2023-03-08 07:21:54,808 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.587e+02 2.392e+02 2.957e+02 3.491e+02 6.570e+02, threshold=5.913e+02, percent-clipped=5.0 2023-03-08 07:23:08,462 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-03-08 07:23:13,803 INFO [train2.py:809] (0/4) Epoch 12, batch 3550, loss[ctc_loss=0.1595, att_loss=0.2845, loss=0.2595, over 14061.00 frames. utt_duration=386.7 frames, utt_pad_proportion=0.325, over 146.00 utterances.], tot_loss[ctc_loss=0.09678, att_loss=0.248, loss=0.2178, over 3266210.19 frames. utt_duration=1236 frames, utt_pad_proportion=0.05815, over 10586.44 utterances.], batch size: 146, lr: 8.92e-03, grad_scale: 8.0 2023-03-08 07:23:15,722 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2596, 5.2133, 5.0788, 3.0812, 4.9689, 4.7356, 4.4647, 2.7256], device='cuda:0'), covar=tensor([0.0108, 0.0082, 0.0202, 0.0986, 0.0091, 0.0188, 0.0280, 0.1468], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0089, 0.0083, 0.0108, 0.0075, 0.0098, 0.0096, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 07:23:25,923 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47379.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:23:40,318 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 07:24:16,155 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-03-08 07:24:34,829 INFO [train2.py:809] (0/4) Epoch 12, batch 3600, loss[ctc_loss=0.1036, att_loss=0.2496, loss=0.2204, over 16672.00 frames. utt_duration=1451 frames, utt_pad_proportion=0.007301, over 46.00 utterances.], tot_loss[ctc_loss=0.09606, att_loss=0.2466, loss=0.2165, over 3261404.93 frames. utt_duration=1264 frames, utt_pad_proportion=0.05356, over 10332.17 utterances.], batch size: 46, lr: 8.92e-03, grad_scale: 8.0 2023-03-08 07:24:36,350 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.815e+02 2.371e+02 2.874e+02 3.552e+02 5.141e+02, threshold=5.747e+02, percent-clipped=0.0 2023-03-08 07:25:04,421 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47440.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:25:15,383 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5979, 5.1134, 4.8684, 5.0479, 5.1325, 4.7893, 3.5542, 4.8870], device='cuda:0'), covar=tensor([0.0099, 0.0088, 0.0110, 0.0066, 0.0073, 0.0093, 0.0670, 0.0201], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0073, 0.0091, 0.0055, 0.0061, 0.0073, 0.0096, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 07:25:47,360 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47466.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:25:56,419 INFO [train2.py:809] (0/4) Epoch 12, batch 3650, loss[ctc_loss=0.09603, att_loss=0.2502, loss=0.2193, over 16331.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006217, over 45.00 utterances.], tot_loss[ctc_loss=0.09593, att_loss=0.2471, loss=0.2168, over 3265432.81 frames. utt_duration=1245 frames, utt_pad_proportion=0.05873, over 10500.30 utterances.], batch size: 45, lr: 8.91e-03, grad_scale: 8.0 2023-03-08 07:26:30,601 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.02 vs. limit=5.0 2023-03-08 07:26:55,624 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.10 vs. limit=5.0 2023-03-08 07:27:03,147 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47514.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:27:13,069 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1606, 2.7143, 3.5028, 2.4637, 3.4364, 4.4420, 4.2100, 2.9284], device='cuda:0'), covar=tensor([0.0414, 0.1665, 0.0948, 0.1529, 0.0981, 0.0703, 0.0500, 0.1541], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0230, 0.0250, 0.0204, 0.0241, 0.0312, 0.0223, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 07:27:16,057 INFO [train2.py:809] (0/4) Epoch 12, batch 3700, loss[ctc_loss=0.108, att_loss=0.2638, loss=0.2326, over 16758.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006174, over 48.00 utterances.], tot_loss[ctc_loss=0.09784, att_loss=0.2483, loss=0.2182, over 3270009.77 frames. utt_duration=1233 frames, utt_pad_proportion=0.06122, over 10619.40 utterances.], batch size: 48, lr: 8.91e-03, grad_scale: 8.0 2023-03-08 07:27:17,595 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.684e+02 2.382e+02 2.831e+02 3.423e+02 6.375e+02, threshold=5.662e+02, percent-clipped=1.0 2023-03-08 07:27:50,228 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.12 vs. limit=5.0 2023-03-08 07:28:38,160 INFO [train2.py:809] (0/4) Epoch 12, batch 3750, loss[ctc_loss=0.0847, att_loss=0.2241, loss=0.1962, over 16175.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006611, over 41.00 utterances.], tot_loss[ctc_loss=0.09941, att_loss=0.2492, loss=0.2192, over 3267395.85 frames. utt_duration=1200 frames, utt_pad_proportion=0.06815, over 10903.47 utterances.], batch size: 41, lr: 8.90e-03, grad_scale: 8.0 2023-03-08 07:29:17,213 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47597.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:29:58,049 INFO [train2.py:809] (0/4) Epoch 12, batch 3800, loss[ctc_loss=0.1022, att_loss=0.2716, loss=0.2377, over 17110.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01472, over 56.00 utterances.], tot_loss[ctc_loss=0.09942, att_loss=0.2495, loss=0.2195, over 3276033.80 frames. utt_duration=1205 frames, utt_pad_proportion=0.06492, over 10885.28 utterances.], batch size: 56, lr: 8.90e-03, grad_scale: 8.0 2023-03-08 07:29:59,587 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.758e+02 2.386e+02 2.816e+02 3.593e+02 7.143e+02, threshold=5.632e+02, percent-clipped=4.0 2023-03-08 07:30:32,376 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9977, 2.5598, 3.4030, 2.4019, 3.3603, 4.2650, 4.0766, 2.8251], device='cuda:0'), covar=tensor([0.0431, 0.1724, 0.0976, 0.1434, 0.0879, 0.0620, 0.0471, 0.1419], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0229, 0.0248, 0.0202, 0.0240, 0.0310, 0.0222, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 07:31:11,674 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6519, 2.9294, 3.5505, 4.4255, 3.9886, 4.1955, 2.9076, 2.4389], device='cuda:0'), covar=tensor([0.0549, 0.2019, 0.0852, 0.0582, 0.0763, 0.0292, 0.1479, 0.2187], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0213, 0.0190, 0.0198, 0.0194, 0.0159, 0.0197, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 07:31:17,520 INFO [train2.py:809] (0/4) Epoch 12, batch 3850, loss[ctc_loss=0.1035, att_loss=0.252, loss=0.2223, over 16883.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006612, over 49.00 utterances.], tot_loss[ctc_loss=0.09798, att_loss=0.2489, loss=0.2187, over 3277969.04 frames. utt_duration=1233 frames, utt_pad_proportion=0.05674, over 10648.72 utterances.], batch size: 49, lr: 8.89e-03, grad_scale: 16.0 2023-03-08 07:31:25,379 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5544, 5.0452, 4.8648, 4.9441, 5.1466, 4.7346, 3.6900, 4.8540], device='cuda:0'), covar=tensor([0.0100, 0.0093, 0.0102, 0.0076, 0.0068, 0.0095, 0.0625, 0.0200], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0075, 0.0093, 0.0057, 0.0063, 0.0075, 0.0097, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 07:32:34,725 INFO [train2.py:809] (0/4) Epoch 12, batch 3900, loss[ctc_loss=0.07615, att_loss=0.2292, loss=0.1986, over 16516.00 frames. utt_duration=1470 frames, utt_pad_proportion=0.007027, over 45.00 utterances.], tot_loss[ctc_loss=0.09799, att_loss=0.2485, loss=0.2184, over 3273756.66 frames. utt_duration=1210 frames, utt_pad_proportion=0.0638, over 10833.42 utterances.], batch size: 45, lr: 8.89e-03, grad_scale: 16.0 2023-03-08 07:32:36,203 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.572e+02 2.278e+02 2.685e+02 3.452e+02 6.474e+02, threshold=5.369e+02, percent-clipped=2.0 2023-03-08 07:32:55,401 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=47735.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:33:00,362 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47738.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:33:52,998 INFO [train2.py:809] (0/4) Epoch 12, batch 3950, loss[ctc_loss=0.0954, att_loss=0.2463, loss=0.2161, over 16783.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005652, over 48.00 utterances.], tot_loss[ctc_loss=0.098, att_loss=0.2485, loss=0.2184, over 3273357.11 frames. utt_duration=1200 frames, utt_pad_proportion=0.06676, over 10929.19 utterances.], batch size: 48, lr: 8.88e-03, grad_scale: 8.0 2023-03-08 07:34:17,730 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47788.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:34:35,120 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47799.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:34:45,125 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-12.pt 2023-03-08 07:35:12,427 INFO [train2.py:809] (0/4) Epoch 13, batch 0, loss[ctc_loss=0.09886, att_loss=0.25, loss=0.2198, over 16481.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005785, over 46.00 utterances.], tot_loss[ctc_loss=0.09886, att_loss=0.25, loss=0.2198, over 16481.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005785, over 46.00 utterances.], batch size: 46, lr: 8.53e-03, grad_scale: 8.0 2023-03-08 07:35:12,429 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 07:35:24,624 INFO [train2.py:843] (0/4) Epoch 13, validation: ctc_loss=0.04799, att_loss=0.2379, loss=0.2, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 07:35:24,625 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 07:35:29,634 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=47809.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 07:35:54,071 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 2.314e+02 2.717e+02 3.534e+02 1.025e+03, threshold=5.435e+02, percent-clipped=1.0 2023-03-08 07:36:34,369 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47849.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:36:45,286 INFO [train2.py:809] (0/4) Epoch 13, batch 50, loss[ctc_loss=0.09482, att_loss=0.2455, loss=0.2153, over 16678.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006885, over 46.00 utterances.], tot_loss[ctc_loss=0.09938, att_loss=0.249, loss=0.2191, over 732162.28 frames. utt_duration=1204 frames, utt_pad_proportion=0.07844, over 2436.01 utterances.], batch size: 46, lr: 8.53e-03, grad_scale: 8.0 2023-03-08 07:37:05,419 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8091, 3.9058, 3.1045, 3.2511, 4.0439, 3.6074, 2.4810, 4.2637], device='cuda:0'), covar=tensor([0.1185, 0.0470, 0.1075, 0.0857, 0.0638, 0.0637, 0.1144, 0.0544], device='cuda:0'), in_proj_covar=tensor([0.0186, 0.0189, 0.0207, 0.0178, 0.0244, 0.0216, 0.0185, 0.0261], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 07:37:08,522 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=47870.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 07:37:36,195 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2023-03-08 07:37:51,303 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=47897.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:37:53,375 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.73 vs. limit=2.0 2023-03-08 07:38:05,791 INFO [train2.py:809] (0/4) Epoch 13, batch 100, loss[ctc_loss=0.1093, att_loss=0.2561, loss=0.2267, over 16528.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.007082, over 45.00 utterances.], tot_loss[ctc_loss=0.09814, att_loss=0.2484, loss=0.2184, over 1299352.55 frames. utt_duration=1238 frames, utt_pad_proportion=0.06051, over 4205.03 utterances.], batch size: 45, lr: 8.52e-03, grad_scale: 8.0 2023-03-08 07:38:35,371 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.519e+02 2.449e+02 2.869e+02 3.545e+02 8.286e+02, threshold=5.739e+02, percent-clipped=3.0 2023-03-08 07:39:08,769 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=47945.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:39:26,425 INFO [train2.py:809] (0/4) Epoch 13, batch 150, loss[ctc_loss=0.08678, att_loss=0.247, loss=0.2149, over 16277.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.006888, over 43.00 utterances.], tot_loss[ctc_loss=0.09608, att_loss=0.2458, loss=0.2158, over 1725909.45 frames. utt_duration=1263 frames, utt_pad_proportion=0.05939, over 5471.27 utterances.], batch size: 43, lr: 8.52e-03, grad_scale: 8.0 2023-03-08 07:40:37,552 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-48000.pt 2023-03-08 07:40:51,746 INFO [train2.py:809] (0/4) Epoch 13, batch 200, loss[ctc_loss=0.09028, att_loss=0.2618, loss=0.2275, over 17075.00 frames. utt_duration=1315 frames, utt_pad_proportion=0.006957, over 52.00 utterances.], tot_loss[ctc_loss=0.09557, att_loss=0.2471, loss=0.2168, over 2079491.01 frames. utt_duration=1280 frames, utt_pad_proportion=0.04748, over 6504.31 utterances.], batch size: 52, lr: 8.52e-03, grad_scale: 8.0 2023-03-08 07:40:56,655 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9233, 4.2691, 4.2484, 4.5201, 2.5367, 4.2864, 2.4390, 1.5806], device='cuda:0'), covar=tensor([0.0378, 0.0142, 0.0758, 0.0168, 0.1887, 0.0196, 0.1678, 0.1873], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0122, 0.0262, 0.0118, 0.0225, 0.0112, 0.0230, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 07:41:20,953 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.209e+02 2.594e+02 3.149e+02 6.379e+02, threshold=5.187e+02, percent-clipped=1.0 2023-03-08 07:41:38,780 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48035.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:42:03,619 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-03-08 07:42:12,098 INFO [train2.py:809] (0/4) Epoch 13, batch 250, loss[ctc_loss=0.08607, att_loss=0.2319, loss=0.2028, over 15849.00 frames. utt_duration=1627 frames, utt_pad_proportion=0.0109, over 39.00 utterances.], tot_loss[ctc_loss=0.09588, att_loss=0.2465, loss=0.2164, over 2341779.64 frames. utt_duration=1282 frames, utt_pad_proportion=0.04834, over 7313.91 utterances.], batch size: 39, lr: 8.51e-03, grad_scale: 8.0 2023-03-08 07:42:38,646 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48072.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:42:55,998 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48083.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:43:12,966 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48094.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:43:28,644 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2853, 4.6804, 4.5615, 4.7042, 4.7527, 4.3514, 3.3304, 4.5954], device='cuda:0'), covar=tensor([0.0120, 0.0119, 0.0134, 0.0082, 0.0096, 0.0119, 0.0692, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0074, 0.0093, 0.0057, 0.0062, 0.0075, 0.0097, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 07:43:31,448 INFO [train2.py:809] (0/4) Epoch 13, batch 300, loss[ctc_loss=0.07589, att_loss=0.2092, loss=0.1825, over 15647.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.007652, over 37.00 utterances.], tot_loss[ctc_loss=0.09568, att_loss=0.2465, loss=0.2163, over 2549741.39 frames. utt_duration=1283 frames, utt_pad_proportion=0.04603, over 7956.10 utterances.], batch size: 37, lr: 8.51e-03, grad_scale: 8.0 2023-03-08 07:44:01,029 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.586e+02 2.171e+02 2.716e+02 3.489e+02 9.070e+02, threshold=5.431e+02, percent-clipped=6.0 2023-03-08 07:44:16,016 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48133.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:44:32,530 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48144.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:44:51,569 INFO [train2.py:809] (0/4) Epoch 13, batch 350, loss[ctc_loss=0.08829, att_loss=0.243, loss=0.2121, over 17113.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01524, over 56.00 utterances.], tot_loss[ctc_loss=0.09737, att_loss=0.248, loss=0.2179, over 2712342.00 frames. utt_duration=1240 frames, utt_pad_proportion=0.05417, over 8756.47 utterances.], batch size: 56, lr: 8.50e-03, grad_scale: 8.0 2023-03-08 07:45:04,061 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-03-08 07:45:06,286 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48165.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 07:45:35,877 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6790, 5.9281, 5.3408, 5.7157, 5.5434, 5.1459, 5.3615, 5.1999], device='cuda:0'), covar=tensor([0.1347, 0.1001, 0.0890, 0.0761, 0.0820, 0.1655, 0.2560, 0.2501], device='cuda:0'), in_proj_covar=tensor([0.0450, 0.0521, 0.0388, 0.0384, 0.0373, 0.0426, 0.0534, 0.0466], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 07:46:10,721 INFO [train2.py:809] (0/4) Epoch 13, batch 400, loss[ctc_loss=0.1169, att_loss=0.2667, loss=0.2368, over 17064.00 frames. utt_duration=691 frames, utt_pad_proportion=0.1319, over 99.00 utterances.], tot_loss[ctc_loss=0.09778, att_loss=0.2485, loss=0.2183, over 2842277.49 frames. utt_duration=1203 frames, utt_pad_proportion=0.06155, over 9465.17 utterances.], batch size: 99, lr: 8.50e-03, grad_scale: 8.0 2023-03-08 07:46:21,578 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48212.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 07:46:40,272 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 2.275e+02 2.732e+02 3.494e+02 6.607e+02, threshold=5.464e+02, percent-clipped=2.0 2023-03-08 07:46:48,647 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3248, 5.1309, 5.1360, 2.8561, 2.1940, 2.8883, 3.0564, 3.7748], device='cuda:0'), covar=tensor([0.0606, 0.0429, 0.0282, 0.3252, 0.6116, 0.2646, 0.2480, 0.2093], device='cuda:0'), in_proj_covar=tensor([0.0339, 0.0230, 0.0245, 0.0218, 0.0348, 0.0338, 0.0236, 0.0360], device='cuda:0'), out_proj_covar=tensor([1.4987e-04, 8.5479e-05, 1.0611e-04, 9.5292e-05, 1.5042e-04, 1.3589e-04, 9.3383e-05, 1.5023e-04], device='cuda:0') 2023-03-08 07:47:03,918 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-03-08 07:47:29,597 INFO [train2.py:809] (0/4) Epoch 13, batch 450, loss[ctc_loss=0.1128, att_loss=0.2668, loss=0.236, over 17101.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01537, over 56.00 utterances.], tot_loss[ctc_loss=0.09635, att_loss=0.2468, loss=0.2167, over 2928106.21 frames. utt_duration=1241 frames, utt_pad_proportion=0.05612, over 9447.97 utterances.], batch size: 56, lr: 8.49e-03, grad_scale: 8.0 2023-03-08 07:47:56,917 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48273.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 07:48:35,791 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4821, 4.7918, 4.3118, 4.8618, 4.2480, 4.5909, 4.9284, 4.6985], device='cuda:0'), covar=tensor([0.0611, 0.0324, 0.0986, 0.0295, 0.0514, 0.0278, 0.0240, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0343, 0.0268, 0.0324, 0.0268, 0.0276, 0.0206, 0.0253, 0.0243], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 07:48:48,218 INFO [train2.py:809] (0/4) Epoch 13, batch 500, loss[ctc_loss=0.1351, att_loss=0.276, loss=0.2478, over 17604.00 frames. utt_duration=893.1 frames, utt_pad_proportion=0.06583, over 79.00 utterances.], tot_loss[ctc_loss=0.09719, att_loss=0.2473, loss=0.2173, over 3006639.24 frames. utt_duration=1239 frames, utt_pad_proportion=0.05622, over 9718.05 utterances.], batch size: 79, lr: 8.49e-03, grad_scale: 8.0 2023-03-08 07:49:16,719 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.612e+02 2.184e+02 2.674e+02 3.467e+02 7.635e+02, threshold=5.348e+02, percent-clipped=2.0 2023-03-08 07:49:17,119 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48324.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:49:30,144 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1724, 4.5127, 4.6808, 4.7205, 2.7171, 4.8127, 2.7423, 1.5809], device='cuda:0'), covar=tensor([0.0298, 0.0190, 0.0527, 0.0134, 0.1633, 0.0121, 0.1483, 0.1851], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0122, 0.0259, 0.0116, 0.0222, 0.0111, 0.0228, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 07:50:05,309 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.58 vs. limit=2.0 2023-03-08 07:50:07,389 INFO [train2.py:809] (0/4) Epoch 13, batch 550, loss[ctc_loss=0.1184, att_loss=0.2753, loss=0.2439, over 16886.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006524, over 49.00 utterances.], tot_loss[ctc_loss=0.09694, att_loss=0.247, loss=0.217, over 3067888.98 frames. utt_duration=1224 frames, utt_pad_proportion=0.05859, over 10040.81 utterances.], batch size: 49, lr: 8.49e-03, grad_scale: 8.0 2023-03-08 07:50:54,962 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48385.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:51:01,044 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5727, 2.6266, 3.1318, 4.4471, 4.1532, 4.2088, 3.1252, 2.1994], device='cuda:0'), covar=tensor([0.0602, 0.2373, 0.1334, 0.0629, 0.0608, 0.0344, 0.1376, 0.2435], device='cuda:0'), in_proj_covar=tensor([0.0171, 0.0212, 0.0189, 0.0195, 0.0193, 0.0159, 0.0199, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 07:51:08,850 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48394.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:51:13,469 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48397.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:51:27,709 INFO [train2.py:809] (0/4) Epoch 13, batch 600, loss[ctc_loss=0.0687, att_loss=0.2181, loss=0.1882, over 15759.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.008672, over 38.00 utterances.], tot_loss[ctc_loss=0.09593, att_loss=0.2466, loss=0.2165, over 3113470.72 frames. utt_duration=1231 frames, utt_pad_proportion=0.05738, over 10128.52 utterances.], batch size: 38, lr: 8.48e-03, grad_scale: 8.0 2023-03-08 07:51:56,287 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48423.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:51:57,443 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.401e+02 2.243e+02 2.721e+02 3.200e+02 7.268e+02, threshold=5.442e+02, percent-clipped=2.0 2023-03-08 07:52:04,406 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48428.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:52:25,669 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48442.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:52:28,973 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48444.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:52:47,829 INFO [train2.py:809] (0/4) Epoch 13, batch 650, loss[ctc_loss=0.1035, att_loss=0.27, loss=0.2367, over 17296.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01224, over 55.00 utterances.], tot_loss[ctc_loss=0.09581, att_loss=0.2465, loss=0.2164, over 3145716.69 frames. utt_duration=1239 frames, utt_pad_proportion=0.05546, over 10171.80 utterances.], batch size: 55, lr: 8.48e-03, grad_scale: 8.0 2023-03-08 07:52:51,846 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48458.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:53:03,126 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48465.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 07:53:11,569 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.26 vs. limit=5.0 2023-03-08 07:53:29,188 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-08 07:53:33,462 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48484.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:53:45,380 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48492.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:54:07,434 INFO [train2.py:809] (0/4) Epoch 13, batch 700, loss[ctc_loss=0.09334, att_loss=0.237, loss=0.2083, over 15897.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008121, over 39.00 utterances.], tot_loss[ctc_loss=0.09585, att_loss=0.2465, loss=0.2164, over 3176912.41 frames. utt_duration=1247 frames, utt_pad_proportion=0.05269, over 10203.72 utterances.], batch size: 39, lr: 8.47e-03, grad_scale: 8.0 2023-03-08 07:54:19,486 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48513.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 07:54:36,145 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.306e+02 2.831e+02 3.598e+02 7.957e+02, threshold=5.663e+02, percent-clipped=6.0 2023-03-08 07:55:26,623 INFO [train2.py:809] (0/4) Epoch 13, batch 750, loss[ctc_loss=0.1249, att_loss=0.2691, loss=0.2403, over 17016.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008006, over 51.00 utterances.], tot_loss[ctc_loss=0.09605, att_loss=0.247, loss=0.2168, over 3204173.75 frames. utt_duration=1263 frames, utt_pad_proportion=0.04849, over 10156.82 utterances.], batch size: 51, lr: 8.47e-03, grad_scale: 8.0 2023-03-08 07:55:46,619 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48568.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 07:56:47,920 INFO [train2.py:809] (0/4) Epoch 13, batch 800, loss[ctc_loss=0.1221, att_loss=0.264, loss=0.2356, over 17029.00 frames. utt_duration=689.8 frames, utt_pad_proportion=0.1335, over 99.00 utterances.], tot_loss[ctc_loss=0.09662, att_loss=0.2472, loss=0.2171, over 3212119.02 frames. utt_duration=1234 frames, utt_pad_proportion=0.0599, over 10424.99 utterances.], batch size: 99, lr: 8.46e-03, grad_scale: 8.0 2023-03-08 07:57:16,742 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.553e+02 2.319e+02 2.822e+02 3.507e+02 7.468e+02, threshold=5.644e+02, percent-clipped=2.0 2023-03-08 07:58:07,622 INFO [train2.py:809] (0/4) Epoch 13, batch 850, loss[ctc_loss=0.1049, att_loss=0.2435, loss=0.2157, over 15738.00 frames. utt_duration=1658 frames, utt_pad_proportion=0.009976, over 38.00 utterances.], tot_loss[ctc_loss=0.09517, att_loss=0.2456, loss=0.2155, over 3223286.75 frames. utt_duration=1269 frames, utt_pad_proportion=0.05197, over 10170.66 utterances.], batch size: 38, lr: 8.46e-03, grad_scale: 8.0 2023-03-08 07:58:29,390 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 07:58:46,037 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48680.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 07:59:28,736 INFO [train2.py:809] (0/4) Epoch 13, batch 900, loss[ctc_loss=0.1037, att_loss=0.2625, loss=0.2308, over 16956.00 frames. utt_duration=686.6 frames, utt_pad_proportion=0.1364, over 99.00 utterances.], tot_loss[ctc_loss=0.09507, att_loss=0.2462, loss=0.216, over 3237963.86 frames. utt_duration=1274 frames, utt_pad_proportion=0.04974, over 10176.61 utterances.], batch size: 99, lr: 8.45e-03, grad_scale: 8.0 2023-03-08 07:59:56,781 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 2.153e+02 2.503e+02 3.048e+02 6.271e+02, threshold=5.007e+02, percent-clipped=1.0 2023-03-08 08:00:04,009 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48728.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:00:43,531 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48753.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:00:49,761 INFO [train2.py:809] (0/4) Epoch 13, batch 950, loss[ctc_loss=0.06979, att_loss=0.2418, loss=0.2074, over 16785.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005593, over 48.00 utterances.], tot_loss[ctc_loss=0.09478, att_loss=0.2464, loss=0.2161, over 3250690.66 frames. utt_duration=1287 frames, utt_pad_proportion=0.04425, over 10118.41 utterances.], batch size: 48, lr: 8.45e-03, grad_scale: 8.0 2023-03-08 08:01:21,802 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48776.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:01:26,494 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=48779.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:02:10,409 INFO [train2.py:809] (0/4) Epoch 13, batch 1000, loss[ctc_loss=0.09798, att_loss=0.2611, loss=0.2285, over 17104.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01586, over 56.00 utterances.], tot_loss[ctc_loss=0.09473, att_loss=0.2464, loss=0.216, over 3256326.12 frames. utt_duration=1249 frames, utt_pad_proportion=0.05243, over 10437.69 utterances.], batch size: 56, lr: 8.45e-03, grad_scale: 8.0 2023-03-08 08:02:33,379 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48821.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:02:37,584 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 2.228e+02 2.748e+02 3.245e+02 7.781e+02, threshold=5.496e+02, percent-clipped=5.0 2023-03-08 08:03:00,431 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=48838.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:03:30,177 INFO [train2.py:809] (0/4) Epoch 13, batch 1050, loss[ctc_loss=0.08496, att_loss=0.2272, loss=0.1988, over 14538.00 frames. utt_duration=1819 frames, utt_pad_proportion=0.02841, over 32.00 utterances.], tot_loss[ctc_loss=0.09443, att_loss=0.2454, loss=0.2152, over 3250554.50 frames. utt_duration=1235 frames, utt_pad_proportion=0.05997, over 10541.28 utterances.], batch size: 32, lr: 8.44e-03, grad_scale: 8.0 2023-03-08 08:03:45,159 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.18 vs. limit=5.0 2023-03-08 08:03:49,011 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48868.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 08:04:07,004 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0125, 5.0748, 4.8896, 2.9095, 4.8457, 4.6375, 4.2300, 2.4550], device='cuda:0'), covar=tensor([0.0104, 0.0090, 0.0274, 0.1087, 0.0096, 0.0163, 0.0347, 0.1565], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0087, 0.0082, 0.0105, 0.0074, 0.0094, 0.0094, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 08:04:09,516 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 08:04:11,753 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48882.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:04:39,018 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=48899.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:04:50,178 INFO [train2.py:809] (0/4) Epoch 13, batch 1100, loss[ctc_loss=0.08155, att_loss=0.2545, loss=0.2199, over 16625.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005244, over 47.00 utterances.], tot_loss[ctc_loss=0.09515, att_loss=0.2461, loss=0.2159, over 3255008.95 frames. utt_duration=1227 frames, utt_pad_proportion=0.06184, over 10622.12 utterances.], batch size: 47, lr: 8.44e-03, grad_scale: 8.0 2023-03-08 08:05:05,418 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=48916.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 08:05:18,178 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.628e+02 2.462e+02 3.131e+02 3.917e+02 9.784e+02, threshold=6.261e+02, percent-clipped=6.0 2023-03-08 08:06:09,390 INFO [train2.py:809] (0/4) Epoch 13, batch 1150, loss[ctc_loss=0.06777, att_loss=0.2035, loss=0.1764, over 15507.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008386, over 36.00 utterances.], tot_loss[ctc_loss=0.09535, att_loss=0.2464, loss=0.2162, over 3251695.58 frames. utt_duration=1237 frames, utt_pad_proportion=0.05951, over 10526.54 utterances.], batch size: 36, lr: 8.43e-03, grad_scale: 8.0 2023-03-08 08:06:39,463 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5849, 1.9932, 1.8943, 2.6679, 3.0454, 2.1207, 2.0211, 2.8878], device='cuda:0'), covar=tensor([0.1934, 0.4727, 0.3963, 0.1293, 0.1248, 0.1793, 0.2951, 0.0833], device='cuda:0'), in_proj_covar=tensor([0.0082, 0.0090, 0.0094, 0.0077, 0.0081, 0.0072, 0.0093, 0.0063], device='cuda:0'), out_proj_covar=tensor([5.6289e-05, 6.4349e-05, 6.6689e-05, 5.4795e-05, 5.5260e-05, 5.4399e-05, 6.4904e-05, 4.8550e-05], device='cuda:0') 2023-03-08 08:06:47,180 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=48980.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:07:28,988 INFO [train2.py:809] (0/4) Epoch 13, batch 1200, loss[ctc_loss=0.09347, att_loss=0.232, loss=0.2043, over 15349.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.0126, over 35.00 utterances.], tot_loss[ctc_loss=0.09499, att_loss=0.2465, loss=0.2162, over 3266229.64 frames. utt_duration=1263 frames, utt_pad_proportion=0.05064, over 10360.05 utterances.], batch size: 35, lr: 8.43e-03, grad_scale: 8.0 2023-03-08 08:07:57,673 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.513e+02 2.494e+02 2.872e+02 3.566e+02 1.058e+03, threshold=5.743e+02, percent-clipped=4.0 2023-03-08 08:08:04,011 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49028.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:08:44,268 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49053.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:08:48,628 INFO [train2.py:809] (0/4) Epoch 13, batch 1250, loss[ctc_loss=0.07121, att_loss=0.2147, loss=0.186, over 13220.00 frames. utt_duration=1825 frames, utt_pad_proportion=0.1028, over 29.00 utterances.], tot_loss[ctc_loss=0.09411, att_loss=0.2462, loss=0.2157, over 3272425.67 frames. utt_duration=1284 frames, utt_pad_proportion=0.04411, over 10206.94 utterances.], batch size: 29, lr: 8.42e-03, grad_scale: 8.0 2023-03-08 08:09:25,431 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49079.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:10:01,106 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49101.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:10:08,581 INFO [train2.py:809] (0/4) Epoch 13, batch 1300, loss[ctc_loss=0.1097, att_loss=0.2414, loss=0.215, over 16002.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007192, over 40.00 utterances.], tot_loss[ctc_loss=0.09446, att_loss=0.2466, loss=0.2162, over 3279098.11 frames. utt_duration=1266 frames, utt_pad_proportion=0.04582, over 10371.00 utterances.], batch size: 40, lr: 8.42e-03, grad_scale: 8.0 2023-03-08 08:10:12,698 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-03-08 08:10:36,806 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 2.317e+02 2.735e+02 3.523e+02 8.419e+02, threshold=5.469e+02, percent-clipped=3.0 2023-03-08 08:10:41,554 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49127.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:11:27,999 INFO [train2.py:809] (0/4) Epoch 13, batch 1350, loss[ctc_loss=0.1342, att_loss=0.2841, loss=0.2541, over 17446.00 frames. utt_duration=1109 frames, utt_pad_proportion=0.02947, over 63.00 utterances.], tot_loss[ctc_loss=0.09451, att_loss=0.2471, loss=0.2166, over 3284037.84 frames. utt_duration=1254 frames, utt_pad_proportion=0.04836, over 10489.57 utterances.], batch size: 63, lr: 8.42e-03, grad_scale: 8.0 2023-03-08 08:12:01,184 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49177.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:12:29,025 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49194.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:12:47,667 INFO [train2.py:809] (0/4) Epoch 13, batch 1400, loss[ctc_loss=0.0696, att_loss=0.2078, loss=0.1802, over 15876.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.00997, over 39.00 utterances.], tot_loss[ctc_loss=0.0946, att_loss=0.247, loss=0.2166, over 3279977.95 frames. utt_duration=1260 frames, utt_pad_proportion=0.04934, over 10422.22 utterances.], batch size: 39, lr: 8.41e-03, grad_scale: 8.0 2023-03-08 08:13:15,975 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.165e+02 2.664e+02 3.367e+02 6.809e+02, threshold=5.329e+02, percent-clipped=2.0 2023-03-08 08:13:35,654 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49236.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:14:07,335 INFO [train2.py:809] (0/4) Epoch 13, batch 1450, loss[ctc_loss=0.1062, att_loss=0.2652, loss=0.2334, over 16544.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006152, over 45.00 utterances.], tot_loss[ctc_loss=0.09409, att_loss=0.2463, loss=0.2159, over 3284905.37 frames. utt_duration=1267 frames, utt_pad_proportion=0.04546, over 10379.30 utterances.], batch size: 45, lr: 8.41e-03, grad_scale: 8.0 2023-03-08 08:15:13,311 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49297.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:15:26,865 INFO [train2.py:809] (0/4) Epoch 13, batch 1500, loss[ctc_loss=0.1066, att_loss=0.2578, loss=0.2275, over 16940.00 frames. utt_duration=686 frames, utt_pad_proportion=0.1382, over 99.00 utterances.], tot_loss[ctc_loss=0.09396, att_loss=0.2462, loss=0.2157, over 3284861.22 frames. utt_duration=1241 frames, utt_pad_proportion=0.05228, over 10596.59 utterances.], batch size: 99, lr: 8.40e-03, grad_scale: 8.0 2023-03-08 08:15:53,293 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 08:15:55,304 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.604e+02 2.233e+02 2.652e+02 3.112e+02 8.079e+02, threshold=5.303e+02, percent-clipped=1.0 2023-03-08 08:16:47,515 INFO [train2.py:809] (0/4) Epoch 13, batch 1550, loss[ctc_loss=0.09809, att_loss=0.2592, loss=0.227, over 17411.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03236, over 63.00 utterances.], tot_loss[ctc_loss=0.09425, att_loss=0.2466, loss=0.2161, over 3276894.03 frames. utt_duration=1216 frames, utt_pad_proportion=0.06086, over 10796.61 utterances.], batch size: 63, lr: 8.40e-03, grad_scale: 8.0 2023-03-08 08:16:53,867 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9625, 2.1304, 2.2047, 2.3747, 3.0838, 2.4968, 2.1748, 2.5806], device='cuda:0'), covar=tensor([0.1504, 0.5331, 0.5189, 0.1907, 0.1708, 0.1914, 0.3070, 0.1371], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0092, 0.0097, 0.0079, 0.0084, 0.0074, 0.0094, 0.0064], device='cuda:0'), out_proj_covar=tensor([5.7340e-05, 6.5367e-05, 6.8683e-05, 5.6131e-05, 5.7650e-05, 5.5839e-05, 6.6031e-05, 4.9422e-05], device='cuda:0') 2023-03-08 08:16:57,821 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-08 08:17:43,768 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6852, 4.5845, 4.6295, 4.5510, 5.1662, 4.6280, 4.6270, 2.3132], device='cuda:0'), covar=tensor([0.0192, 0.0254, 0.0224, 0.0236, 0.0958, 0.0225, 0.0244, 0.2023], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0137, 0.0142, 0.0155, 0.0341, 0.0125, 0.0125, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 08:18:08,484 INFO [train2.py:809] (0/4) Epoch 13, batch 1600, loss[ctc_loss=0.09064, att_loss=0.2362, loss=0.2071, over 15889.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008926, over 39.00 utterances.], tot_loss[ctc_loss=0.09426, att_loss=0.2464, loss=0.216, over 3279051.07 frames. utt_duration=1233 frames, utt_pad_proportion=0.05625, over 10647.13 utterances.], batch size: 39, lr: 8.40e-03, grad_scale: 8.0 2023-03-08 08:18:16,591 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6067, 2.7295, 3.6315, 3.1218, 3.5108, 4.7720, 4.4860, 3.5671], device='cuda:0'), covar=tensor([0.0343, 0.1939, 0.1110, 0.1160, 0.1107, 0.0689, 0.0514, 0.1107], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0233, 0.0254, 0.0204, 0.0249, 0.0319, 0.0230, 0.0227], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:18:34,206 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49422.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:18:36,893 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.327e+02 2.714e+02 3.320e+02 9.456e+02, threshold=5.428e+02, percent-clipped=3.0 2023-03-08 08:18:50,233 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49432.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 08:19:07,546 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 08:19:28,465 INFO [train2.py:809] (0/4) Epoch 13, batch 1650, loss[ctc_loss=0.09031, att_loss=0.2269, loss=0.1996, over 15488.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.009584, over 36.00 utterances.], tot_loss[ctc_loss=0.09458, att_loss=0.2465, loss=0.2161, over 3280918.69 frames. utt_duration=1239 frames, utt_pad_proportion=0.05357, over 10602.12 utterances.], batch size: 36, lr: 8.39e-03, grad_scale: 8.0 2023-03-08 08:20:02,446 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49477.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:20:12,539 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49483.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:20:29,267 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49493.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 08:20:30,717 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49494.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:20:48,665 INFO [train2.py:809] (0/4) Epoch 13, batch 1700, loss[ctc_loss=0.1441, att_loss=0.2602, loss=0.237, over 16419.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.005761, over 44.00 utterances.], tot_loss[ctc_loss=0.09525, att_loss=0.2469, loss=0.2166, over 3275281.96 frames. utt_duration=1223 frames, utt_pad_proportion=0.05872, over 10725.34 utterances.], batch size: 44, lr: 8.39e-03, grad_scale: 8.0 2023-03-08 08:21:03,510 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8437, 5.2108, 5.3833, 5.2787, 5.2208, 5.8369, 4.9988, 5.9116], device='cuda:0'), covar=tensor([0.0761, 0.0726, 0.0876, 0.1173, 0.2175, 0.0980, 0.0819, 0.0778], device='cuda:0'), in_proj_covar=tensor([0.0748, 0.0441, 0.0519, 0.0576, 0.0763, 0.0527, 0.0428, 0.0522], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:21:03,576 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2660, 4.7714, 4.7265, 4.7803, 4.7738, 4.6312, 3.1646, 4.6010], device='cuda:0'), covar=tensor([0.0178, 0.0182, 0.0157, 0.0121, 0.0151, 0.0138, 0.1013, 0.0409], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0073, 0.0091, 0.0057, 0.0062, 0.0072, 0.0093, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:21:16,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.458e+02 2.324e+02 2.788e+02 3.752e+02 8.319e+02, threshold=5.576e+02, percent-clipped=5.0 2023-03-08 08:21:18,726 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49525.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:21:46,000 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49542.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:21:50,507 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8693, 5.1659, 5.4581, 5.3519, 5.2671, 5.8346, 5.0588, 5.9360], device='cuda:0'), covar=tensor([0.0615, 0.0617, 0.0765, 0.1092, 0.1744, 0.0812, 0.0706, 0.0668], device='cuda:0'), in_proj_covar=tensor([0.0745, 0.0440, 0.0517, 0.0574, 0.0763, 0.0526, 0.0428, 0.0522], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:22:07,061 INFO [train2.py:809] (0/4) Epoch 13, batch 1750, loss[ctc_loss=0.09616, att_loss=0.2435, loss=0.2141, over 16963.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007626, over 50.00 utterances.], tot_loss[ctc_loss=0.09487, att_loss=0.2465, loss=0.2162, over 3276976.95 frames. utt_duration=1229 frames, utt_pad_proportion=0.05671, over 10676.90 utterances.], batch size: 50, lr: 8.38e-03, grad_scale: 8.0 2023-03-08 08:22:43,665 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2544, 4.7278, 4.6063, 4.8031, 4.7892, 4.4113, 3.1465, 4.5639], device='cuda:0'), covar=tensor([0.0125, 0.0139, 0.0138, 0.0084, 0.0112, 0.0123, 0.0816, 0.0243], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0075, 0.0092, 0.0057, 0.0063, 0.0073, 0.0095, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:22:58,515 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-03-08 08:23:03,775 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49592.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:23:25,896 INFO [train2.py:809] (0/4) Epoch 13, batch 1800, loss[ctc_loss=0.0918, att_loss=0.2573, loss=0.2242, over 17129.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01444, over 56.00 utterances.], tot_loss[ctc_loss=0.09534, att_loss=0.2466, loss=0.2164, over 3274896.47 frames. utt_duration=1227 frames, utt_pad_proportion=0.05795, over 10691.62 utterances.], batch size: 56, lr: 8.38e-03, grad_scale: 8.0 2023-03-08 08:23:54,567 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.478e+02 2.431e+02 2.885e+02 3.492e+02 8.603e+02, threshold=5.770e+02, percent-clipped=6.0 2023-03-08 08:24:45,025 INFO [train2.py:809] (0/4) Epoch 13, batch 1850, loss[ctc_loss=0.1302, att_loss=0.2738, loss=0.2451, over 17521.00 frames. utt_duration=888.8 frames, utt_pad_proportion=0.06935, over 79.00 utterances.], tot_loss[ctc_loss=0.09443, att_loss=0.2457, loss=0.2155, over 3268494.63 frames. utt_duration=1246 frames, utt_pad_proportion=0.05423, over 10504.28 utterances.], batch size: 79, lr: 8.37e-03, grad_scale: 8.0 2023-03-08 08:25:26,550 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4716, 4.4532, 4.3585, 2.8367, 4.3343, 4.2182, 3.9060, 2.5703], device='cuda:0'), covar=tensor([0.0116, 0.0101, 0.0220, 0.1020, 0.0095, 0.0218, 0.0330, 0.1483], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0088, 0.0084, 0.0104, 0.0074, 0.0095, 0.0093, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 08:25:57,077 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6571, 3.6447, 3.5161, 3.2913, 3.6753, 3.6062, 3.5194, 2.7262], device='cuda:0'), covar=tensor([0.0830, 0.1794, 0.2992, 0.4236, 0.0955, 0.2131, 0.1231, 0.5515], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0131, 0.0141, 0.0208, 0.0109, 0.0196, 0.0117, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 08:26:04,418 INFO [train2.py:809] (0/4) Epoch 13, batch 1900, loss[ctc_loss=0.07288, att_loss=0.2166, loss=0.1879, over 16014.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006862, over 40.00 utterances.], tot_loss[ctc_loss=0.09499, att_loss=0.2463, loss=0.216, over 3270044.39 frames. utt_duration=1231 frames, utt_pad_proportion=0.05711, over 10638.64 utterances.], batch size: 40, lr: 8.37e-03, grad_scale: 8.0 2023-03-08 08:26:15,616 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8721, 3.6724, 3.5647, 3.2031, 3.6159, 3.6104, 3.5573, 2.6269], device='cuda:0'), covar=tensor([0.0914, 0.1649, 0.4119, 0.5416, 0.1642, 0.4470, 0.1233, 0.6396], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0132, 0.0141, 0.0208, 0.0109, 0.0196, 0.0117, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 08:26:33,116 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.157e+02 2.753e+02 3.358e+02 9.197e+02, threshold=5.506e+02, percent-clipped=6.0 2023-03-08 08:26:37,186 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-08 08:27:24,889 INFO [train2.py:809] (0/4) Epoch 13, batch 1950, loss[ctc_loss=0.1066, att_loss=0.2677, loss=0.2355, over 17139.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.0131, over 56.00 utterances.], tot_loss[ctc_loss=0.09432, att_loss=0.2458, loss=0.2155, over 3271020.74 frames. utt_duration=1260 frames, utt_pad_proportion=0.04983, over 10392.64 utterances.], batch size: 56, lr: 8.37e-03, grad_scale: 16.0 2023-03-08 08:27:31,836 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9651, 3.7632, 3.0703, 3.5270, 3.9411, 3.5786, 2.6699, 4.3336], device='cuda:0'), covar=tensor([0.1054, 0.0512, 0.1172, 0.0693, 0.0691, 0.0726, 0.1020, 0.0558], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0193, 0.0210, 0.0181, 0.0247, 0.0220, 0.0189, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 08:27:33,316 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3347, 3.2712, 3.1460, 2.8447, 3.2157, 3.1689, 3.2185, 2.1518], device='cuda:0'), covar=tensor([0.1160, 0.2053, 0.3931, 0.5055, 0.1761, 0.3598, 0.1476, 0.7043], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0130, 0.0139, 0.0206, 0.0107, 0.0193, 0.0116, 0.0177], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0001], device='cuda:0') 2023-03-08 08:27:41,125 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=49766.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:27:59,749 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49778.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:28:17,539 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=49788.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 08:28:45,894 INFO [train2.py:809] (0/4) Epoch 13, batch 2000, loss[ctc_loss=0.08476, att_loss=0.2323, loss=0.2028, over 16183.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.00673, over 41.00 utterances.], tot_loss[ctc_loss=0.095, att_loss=0.2462, loss=0.2159, over 3263259.28 frames. utt_duration=1245 frames, utt_pad_proportion=0.05474, over 10495.63 utterances.], batch size: 41, lr: 8.36e-03, grad_scale: 16.0 2023-03-08 08:28:46,095 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9236, 5.2182, 5.4954, 5.2815, 5.3290, 5.8470, 5.1788, 5.9654], device='cuda:0'), covar=tensor([0.0703, 0.0730, 0.0886, 0.1225, 0.2127, 0.0970, 0.0722, 0.0796], device='cuda:0'), in_proj_covar=tensor([0.0754, 0.0447, 0.0525, 0.0583, 0.0768, 0.0536, 0.0432, 0.0524], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:28:49,258 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4146, 2.7940, 3.5036, 2.9080, 3.4703, 4.5516, 4.3461, 3.3720], device='cuda:0'), covar=tensor([0.0422, 0.1952, 0.1326, 0.1337, 0.1094, 0.0882, 0.0563, 0.1327], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0235, 0.0254, 0.0205, 0.0250, 0.0320, 0.0231, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:29:14,827 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.246e+02 2.552e+02 3.168e+02 7.222e+02, threshold=5.105e+02, percent-clipped=2.0 2023-03-08 08:29:19,831 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=49827.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:29:28,213 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-08 08:30:05,631 INFO [train2.py:809] (0/4) Epoch 13, batch 2050, loss[ctc_loss=0.1054, att_loss=0.2703, loss=0.2373, over 17303.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02363, over 59.00 utterances.], tot_loss[ctc_loss=0.09591, att_loss=0.2473, loss=0.217, over 3271841.18 frames. utt_duration=1223 frames, utt_pad_proportion=0.05743, over 10712.71 utterances.], batch size: 59, lr: 8.36e-03, grad_scale: 16.0 2023-03-08 08:30:19,629 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7587, 4.6731, 4.5779, 4.4467, 5.2444, 4.8063, 4.6845, 2.3511], device='cuda:0'), covar=tensor([0.0171, 0.0272, 0.0270, 0.0305, 0.0966, 0.0177, 0.0244, 0.2009], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0137, 0.0144, 0.0154, 0.0340, 0.0125, 0.0124, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 08:31:03,724 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=49892.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:31:11,696 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3963, 3.9718, 3.3933, 3.7323, 4.0882, 3.7034, 3.2058, 4.5770], device='cuda:0'), covar=tensor([0.0740, 0.0401, 0.0873, 0.0542, 0.0674, 0.0627, 0.0767, 0.0480], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0193, 0.0209, 0.0180, 0.0246, 0.0218, 0.0188, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 08:31:25,303 INFO [train2.py:809] (0/4) Epoch 13, batch 2100, loss[ctc_loss=0.07745, att_loss=0.2386, loss=0.2064, over 16175.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.007205, over 41.00 utterances.], tot_loss[ctc_loss=0.09536, att_loss=0.247, loss=0.2167, over 3276227.99 frames. utt_duration=1243 frames, utt_pad_proportion=0.05271, over 10552.34 utterances.], batch size: 41, lr: 8.35e-03, grad_scale: 16.0 2023-03-08 08:31:53,560 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.781e+02 2.361e+02 2.979e+02 3.749e+02 1.127e+03, threshold=5.957e+02, percent-clipped=6.0 2023-03-08 08:32:19,963 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=49940.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:32:44,496 INFO [train2.py:809] (0/4) Epoch 13, batch 2150, loss[ctc_loss=0.06558, att_loss=0.2144, loss=0.1846, over 15867.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01041, over 39.00 utterances.], tot_loss[ctc_loss=0.09539, att_loss=0.2475, loss=0.2171, over 3282217.34 frames. utt_duration=1245 frames, utt_pad_proportion=0.05123, over 10561.68 utterances.], batch size: 39, lr: 8.35e-03, grad_scale: 16.0 2023-03-08 08:33:54,617 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-50000.pt 2023-03-08 08:34:08,270 INFO [train2.py:809] (0/4) Epoch 13, batch 2200, loss[ctc_loss=0.09983, att_loss=0.2545, loss=0.2236, over 16408.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007193, over 44.00 utterances.], tot_loss[ctc_loss=0.09625, att_loss=0.2482, loss=0.2178, over 3289011.19 frames. utt_duration=1216 frames, utt_pad_proportion=0.05537, over 10832.55 utterances.], batch size: 44, lr: 8.35e-03, grad_scale: 16.0 2023-03-08 08:34:10,122 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5896, 4.4103, 4.3521, 4.3904, 4.9698, 4.5697, 4.4216, 2.2716], device='cuda:0'), covar=tensor([0.0206, 0.0341, 0.0339, 0.0242, 0.0895, 0.0210, 0.0277, 0.2234], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0140, 0.0146, 0.0157, 0.0347, 0.0128, 0.0127, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 08:34:28,750 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8591, 1.6989, 2.3888, 2.5382, 2.9234, 2.7436, 2.4584, 2.6536], device='cuda:0'), covar=tensor([0.1408, 0.4722, 0.3120, 0.1606, 0.2623, 0.1218, 0.2699, 0.1105], device='cuda:0'), in_proj_covar=tensor([0.0084, 0.0093, 0.0094, 0.0079, 0.0085, 0.0074, 0.0093, 0.0064], device='cuda:0'), out_proj_covar=tensor([5.8445e-05, 6.5961e-05, 6.7900e-05, 5.6945e-05, 5.8443e-05, 5.5630e-05, 6.5739e-05, 4.9667e-05], device='cuda:0') 2023-03-08 08:34:30,701 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.00 vs. limit=2.0 2023-03-08 08:34:35,762 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.313e+02 2.828e+02 3.809e+02 8.151e+02, threshold=5.657e+02, percent-clipped=5.0 2023-03-08 08:35:25,942 INFO [train2.py:809] (0/4) Epoch 13, batch 2250, loss[ctc_loss=0.07107, att_loss=0.2273, loss=0.196, over 16014.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.0068, over 40.00 utterances.], tot_loss[ctc_loss=0.0957, att_loss=0.2473, loss=0.217, over 3283852.82 frames. utt_duration=1219 frames, utt_pad_proportion=0.05614, over 10789.47 utterances.], batch size: 40, lr: 8.34e-03, grad_scale: 16.0 2023-03-08 08:35:39,191 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-03-08 08:35:44,979 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5388, 2.8019, 3.6685, 4.4425, 3.9663, 3.9472, 2.8973, 2.0989], device='cuda:0'), covar=tensor([0.0710, 0.2261, 0.0855, 0.0540, 0.0878, 0.0458, 0.1586, 0.2467], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0212, 0.0192, 0.0200, 0.0200, 0.0164, 0.0200, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 08:36:00,904 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50078.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:36:17,715 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50088.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 08:36:19,841 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2023-03-08 08:36:45,461 INFO [train2.py:809] (0/4) Epoch 13, batch 2300, loss[ctc_loss=0.07924, att_loss=0.239, loss=0.2071, over 16280.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007128, over 43.00 utterances.], tot_loss[ctc_loss=0.09514, att_loss=0.2465, loss=0.2162, over 3274571.07 frames. utt_duration=1246 frames, utt_pad_proportion=0.05257, over 10528.24 utterances.], batch size: 43, lr: 8.34e-03, grad_scale: 8.0 2023-03-08 08:37:10,625 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50122.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:37:15,134 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.321e+02 2.441e+02 2.731e+02 3.266e+02 6.421e+02, threshold=5.462e+02, percent-clipped=3.0 2023-03-08 08:37:17,298 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50126.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:37:20,763 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50128.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:37:30,492 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.99 vs. limit=2.0 2023-03-08 08:37:34,117 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50136.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 08:38:04,742 INFO [train2.py:809] (0/4) Epoch 13, batch 2350, loss[ctc_loss=0.1187, att_loss=0.2703, loss=0.24, over 17094.00 frames. utt_duration=1222 frames, utt_pad_proportion=0.01651, over 56.00 utterances.], tot_loss[ctc_loss=0.09508, att_loss=0.2462, loss=0.216, over 3273766.00 frames. utt_duration=1263 frames, utt_pad_proportion=0.04943, over 10377.91 utterances.], batch size: 56, lr: 8.33e-03, grad_scale: 8.0 2023-03-08 08:38:56,983 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1883, 5.1857, 4.9514, 2.9822, 5.0005, 4.8467, 4.4038, 2.7832], device='cuda:0'), covar=tensor([0.0114, 0.0079, 0.0261, 0.1011, 0.0083, 0.0143, 0.0287, 0.1381], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0087, 0.0085, 0.0106, 0.0075, 0.0096, 0.0092, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 08:38:58,532 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50189.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:39:02,913 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6088, 5.0644, 4.8538, 4.9802, 5.1754, 4.7397, 3.6477, 4.9767], device='cuda:0'), covar=tensor([0.0100, 0.0094, 0.0111, 0.0095, 0.0086, 0.0091, 0.0598, 0.0190], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0073, 0.0092, 0.0057, 0.0062, 0.0072, 0.0093, 0.0093], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:39:10,485 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50197.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 08:39:24,033 INFO [train2.py:809] (0/4) Epoch 13, batch 2400, loss[ctc_loss=0.1003, att_loss=0.2544, loss=0.2236, over 17262.00 frames. utt_duration=1172 frames, utt_pad_proportion=0.02517, over 59.00 utterances.], tot_loss[ctc_loss=0.09553, att_loss=0.2469, loss=0.2167, over 3279114.78 frames. utt_duration=1246 frames, utt_pad_proportion=0.05232, over 10536.26 utterances.], batch size: 59, lr: 8.33e-03, grad_scale: 8.0 2023-03-08 08:39:54,444 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.486e+02 2.261e+02 2.728e+02 3.440e+02 5.730e+02, threshold=5.455e+02, percent-clipped=1.0 2023-03-08 08:40:42,741 INFO [train2.py:809] (0/4) Epoch 13, batch 2450, loss[ctc_loss=0.08492, att_loss=0.2381, loss=0.2074, over 16409.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007268, over 44.00 utterances.], tot_loss[ctc_loss=0.09519, att_loss=0.2464, loss=0.2162, over 3275109.16 frames. utt_duration=1250 frames, utt_pad_proportion=0.05331, over 10490.97 utterances.], batch size: 44, lr: 8.32e-03, grad_scale: 8.0 2023-03-08 08:40:46,183 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50258.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 08:41:59,980 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 08:42:02,222 INFO [train2.py:809] (0/4) Epoch 13, batch 2500, loss[ctc_loss=0.08697, att_loss=0.2437, loss=0.2124, over 16007.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007156, over 40.00 utterances.], tot_loss[ctc_loss=0.09431, att_loss=0.246, loss=0.2157, over 3268586.51 frames. utt_duration=1262 frames, utt_pad_proportion=0.05313, over 10374.22 utterances.], batch size: 40, lr: 8.32e-03, grad_scale: 8.0 2023-03-08 08:42:05,466 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1030, 5.3275, 5.6876, 5.4896, 5.5411, 6.0947, 5.2409, 6.1724], device='cuda:0'), covar=tensor([0.0655, 0.0683, 0.0674, 0.1162, 0.1725, 0.0686, 0.0587, 0.0557], device='cuda:0'), in_proj_covar=tensor([0.0764, 0.0445, 0.0530, 0.0580, 0.0769, 0.0531, 0.0431, 0.0526], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:42:25,965 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.15 vs. limit=2.0 2023-03-08 08:42:33,321 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.276e+02 2.773e+02 3.463e+02 7.694e+02, threshold=5.547e+02, percent-clipped=7.0 2023-03-08 08:43:21,287 INFO [train2.py:809] (0/4) Epoch 13, batch 2550, loss[ctc_loss=0.06345, att_loss=0.2204, loss=0.189, over 16124.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.005729, over 42.00 utterances.], tot_loss[ctc_loss=0.09411, att_loss=0.2462, loss=0.2158, over 3276804.01 frames. utt_duration=1269 frames, utt_pad_proportion=0.04964, over 10338.58 utterances.], batch size: 42, lr: 8.32e-03, grad_scale: 8.0 2023-03-08 08:44:36,453 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-08 08:44:40,229 INFO [train2.py:809] (0/4) Epoch 13, batch 2600, loss[ctc_loss=0.09437, att_loss=0.2291, loss=0.2021, over 15998.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008375, over 40.00 utterances.], tot_loss[ctc_loss=0.09465, att_loss=0.2465, loss=0.2162, over 3276496.90 frames. utt_duration=1262 frames, utt_pad_proportion=0.05145, over 10397.31 utterances.], batch size: 40, lr: 8.31e-03, grad_scale: 8.0 2023-03-08 08:44:42,078 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9843, 5.2658, 4.7788, 5.3488, 4.6789, 4.9766, 5.4569, 5.1579], device='cuda:0'), covar=tensor([0.0546, 0.0308, 0.0876, 0.0257, 0.0444, 0.0230, 0.0203, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0273, 0.0329, 0.0279, 0.0280, 0.0212, 0.0260, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 08:45:05,592 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50422.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:45:10,504 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.653e+02 2.288e+02 2.850e+02 3.496e+02 8.572e+02, threshold=5.699e+02, percent-clipped=4.0 2023-03-08 08:45:43,139 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-03-08 08:45:59,190 INFO [train2.py:809] (0/4) Epoch 13, batch 2650, loss[ctc_loss=0.07369, att_loss=0.2413, loss=0.2078, over 16475.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006101, over 46.00 utterances.], tot_loss[ctc_loss=0.09443, att_loss=0.246, loss=0.2157, over 3271285.71 frames. utt_duration=1253 frames, utt_pad_proportion=0.05553, over 10457.93 utterances.], batch size: 46, lr: 8.31e-03, grad_scale: 8.0 2023-03-08 08:46:22,032 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50470.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:46:44,836 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50484.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:46:53,696 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0491, 5.2850, 5.5547, 5.3927, 5.4809, 6.0031, 5.1778, 6.1032], device='cuda:0'), covar=tensor([0.0576, 0.0716, 0.0803, 0.1141, 0.1708, 0.0766, 0.0676, 0.0564], device='cuda:0'), in_proj_covar=tensor([0.0778, 0.0455, 0.0542, 0.0596, 0.0786, 0.0544, 0.0436, 0.0534], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:47:17,951 INFO [train2.py:809] (0/4) Epoch 13, batch 2700, loss[ctc_loss=0.1245, att_loss=0.2613, loss=0.2339, over 16336.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005685, over 45.00 utterances.], tot_loss[ctc_loss=0.09508, att_loss=0.2461, loss=0.2159, over 3273612.24 frames. utt_duration=1251 frames, utt_pad_proportion=0.05511, over 10478.39 utterances.], batch size: 45, lr: 8.30e-03, grad_scale: 8.0 2023-03-08 08:47:20,228 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-03-08 08:47:49,716 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.590e+02 2.365e+02 2.788e+02 3.367e+02 6.677e+02, threshold=5.576e+02, percent-clipped=4.0 2023-03-08 08:48:11,806 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9900, 6.2372, 5.6565, 5.9825, 5.8885, 5.3857, 5.5798, 5.4572], device='cuda:0'), covar=tensor([0.1315, 0.0835, 0.0864, 0.0760, 0.0805, 0.1581, 0.2445, 0.2283], device='cuda:0'), in_proj_covar=tensor([0.0452, 0.0526, 0.0394, 0.0395, 0.0378, 0.0431, 0.0544, 0.0474], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 08:48:23,008 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50546.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:48:33,724 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50553.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 08:48:38,308 INFO [train2.py:809] (0/4) Epoch 13, batch 2750, loss[ctc_loss=0.09413, att_loss=0.2517, loss=0.2202, over 16967.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.00767, over 50.00 utterances.], tot_loss[ctc_loss=0.09542, att_loss=0.2458, loss=0.2158, over 3266122.25 frames. utt_duration=1249 frames, utt_pad_proportion=0.05875, over 10474.98 utterances.], batch size: 50, lr: 8.30e-03, grad_scale: 8.0 2023-03-08 08:49:35,948 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8143, 5.1383, 5.3996, 5.3502, 5.2650, 5.8109, 5.1710, 5.9029], device='cuda:0'), covar=tensor([0.0672, 0.0727, 0.0705, 0.1071, 0.1751, 0.0801, 0.0688, 0.0575], device='cuda:0'), in_proj_covar=tensor([0.0776, 0.0456, 0.0541, 0.0595, 0.0784, 0.0542, 0.0434, 0.0531], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 08:49:57,335 INFO [train2.py:809] (0/4) Epoch 13, batch 2800, loss[ctc_loss=0.0744, att_loss=0.2178, loss=0.1891, over 15747.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.01003, over 38.00 utterances.], tot_loss[ctc_loss=0.09473, att_loss=0.2453, loss=0.2152, over 3272199.65 frames. utt_duration=1274 frames, utt_pad_proportion=0.05069, over 10284.08 utterances.], batch size: 38, lr: 8.30e-03, grad_scale: 8.0 2023-03-08 08:49:59,310 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50607.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:50:01,292 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-08 08:50:29,966 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.488e+02 2.143e+02 2.586e+02 3.162e+02 8.446e+02, threshold=5.172e+02, percent-clipped=2.0 2023-03-08 08:50:53,967 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.90 vs. limit=5.0 2023-03-08 08:51:14,497 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-08 08:51:17,763 INFO [train2.py:809] (0/4) Epoch 13, batch 2850, loss[ctc_loss=0.1144, att_loss=0.2812, loss=0.2479, over 17297.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01218, over 55.00 utterances.], tot_loss[ctc_loss=0.09395, att_loss=0.2445, loss=0.2144, over 3268106.93 frames. utt_duration=1281 frames, utt_pad_proportion=0.05063, over 10215.34 utterances.], batch size: 55, lr: 8.29e-03, grad_scale: 8.0 2023-03-08 08:52:37,438 INFO [train2.py:809] (0/4) Epoch 13, batch 2900, loss[ctc_loss=0.1105, att_loss=0.2531, loss=0.2246, over 15953.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.005728, over 41.00 utterances.], tot_loss[ctc_loss=0.09365, att_loss=0.2448, loss=0.2145, over 3274932.41 frames. utt_duration=1282 frames, utt_pad_proportion=0.04825, over 10233.91 utterances.], batch size: 41, lr: 8.29e-03, grad_scale: 8.0 2023-03-08 08:52:45,157 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6399, 5.9493, 5.4105, 5.7243, 5.5696, 5.2242, 5.3453, 5.0423], device='cuda:0'), covar=tensor([0.1390, 0.0892, 0.0961, 0.0795, 0.0899, 0.1498, 0.2408, 0.2396], device='cuda:0'), in_proj_covar=tensor([0.0452, 0.0526, 0.0394, 0.0395, 0.0378, 0.0427, 0.0542, 0.0475], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 08:53:05,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 08:53:08,851 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.231e+02 2.674e+02 3.196e+02 5.262e+02, threshold=5.348e+02, percent-clipped=1.0 2023-03-08 08:53:56,557 INFO [train2.py:809] (0/4) Epoch 13, batch 2950, loss[ctc_loss=0.07575, att_loss=0.2177, loss=0.1893, over 14518.00 frames. utt_duration=1816 frames, utt_pad_proportion=0.0426, over 32.00 utterances.], tot_loss[ctc_loss=0.09345, att_loss=0.2448, loss=0.2145, over 3272657.69 frames. utt_duration=1277 frames, utt_pad_proportion=0.05009, over 10260.07 utterances.], batch size: 32, lr: 8.28e-03, grad_scale: 8.0 2023-03-08 08:54:39,269 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50782.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:54:42,252 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50784.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:55:16,327 INFO [train2.py:809] (0/4) Epoch 13, batch 3000, loss[ctc_loss=0.09338, att_loss=0.2595, loss=0.2263, over 17017.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.00773, over 51.00 utterances.], tot_loss[ctc_loss=0.094, att_loss=0.2452, loss=0.215, over 3269404.64 frames. utt_duration=1270 frames, utt_pad_proportion=0.05354, over 10307.22 utterances.], batch size: 51, lr: 8.28e-03, grad_scale: 8.0 2023-03-08 08:55:16,330 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 08:55:30,014 INFO [train2.py:843] (0/4) Epoch 13, validation: ctc_loss=0.04571, att_loss=0.2368, loss=0.1986, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 08:55:30,015 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 08:55:33,388 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50808.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 08:56:00,977 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.736e+02 2.441e+02 3.101e+02 4.045e+02 6.345e+02, threshold=6.202e+02, percent-clipped=8.0 2023-03-08 08:56:09,393 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7303, 4.7939, 4.7257, 4.5971, 5.1335, 5.0153, 4.7563, 2.4708], device='cuda:0'), covar=tensor([0.0128, 0.0150, 0.0154, 0.0119, 0.0698, 0.0090, 0.0149, 0.1795], device='cuda:0'), in_proj_covar=tensor([0.0130, 0.0138, 0.0145, 0.0157, 0.0342, 0.0126, 0.0129, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 08:56:12,026 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50832.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:56:21,567 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7738, 2.4391, 5.0120, 4.3152, 3.1544, 4.5308, 5.0510, 4.9328], device='cuda:0'), covar=tensor([0.0133, 0.1540, 0.0181, 0.0746, 0.1699, 0.0167, 0.0091, 0.0125], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0241, 0.0144, 0.0307, 0.0270, 0.0187, 0.0129, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 08:56:29,182 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50843.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:56:44,444 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2597, 5.2214, 5.0541, 2.6637, 5.0645, 4.8645, 4.5232, 3.0925], device='cuda:0'), covar=tensor([0.0101, 0.0089, 0.0200, 0.1136, 0.0091, 0.0127, 0.0269, 0.1162], device='cuda:0'), in_proj_covar=tensor([0.0063, 0.0087, 0.0085, 0.0104, 0.0075, 0.0096, 0.0092, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 08:56:44,455 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=50853.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 08:56:48,708 INFO [train2.py:809] (0/4) Epoch 13, batch 3050, loss[ctc_loss=0.07419, att_loss=0.2347, loss=0.2026, over 16344.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005335, over 45.00 utterances.], tot_loss[ctc_loss=0.0945, att_loss=0.2458, loss=0.2156, over 3267804.19 frames. utt_duration=1260 frames, utt_pad_proportion=0.05507, over 10389.32 utterances.], batch size: 45, lr: 8.28e-03, grad_scale: 8.0 2023-03-08 08:57:01,493 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6554, 4.9221, 4.4309, 4.9629, 4.3425, 4.6856, 5.0573, 4.8587], device='cuda:0'), covar=tensor([0.0574, 0.0316, 0.0862, 0.0322, 0.0463, 0.0321, 0.0230, 0.0201], device='cuda:0'), in_proj_covar=tensor([0.0341, 0.0271, 0.0324, 0.0275, 0.0275, 0.0211, 0.0256, 0.0242], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-08 08:57:08,752 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 08:57:10,229 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=50869.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 08:57:50,322 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1191, 2.6178, 3.0452, 4.2868, 3.6866, 3.7519, 2.6616, 1.8556], device='cuda:0'), covar=tensor([0.0846, 0.2297, 0.1136, 0.0555, 0.0879, 0.0509, 0.1639, 0.2668], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0213, 0.0190, 0.0201, 0.0202, 0.0162, 0.0198, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 08:58:00,716 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=50901.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 08:58:02,422 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=50902.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:58:08,260 INFO [train2.py:809] (0/4) Epoch 13, batch 3100, loss[ctc_loss=0.08746, att_loss=0.2456, loss=0.214, over 16620.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005648, over 47.00 utterances.], tot_loss[ctc_loss=0.09495, att_loss=0.2463, loss=0.2161, over 3271824.16 frames. utt_duration=1231 frames, utt_pad_proportion=0.05979, over 10640.32 utterances.], batch size: 47, lr: 8.27e-03, grad_scale: 8.0 2023-03-08 08:58:32,081 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-03-08 08:58:39,136 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.262e+02 2.755e+02 3.486e+02 1.234e+03, threshold=5.511e+02, percent-clipped=2.0 2023-03-08 08:59:11,626 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=50946.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 08:59:26,548 INFO [train2.py:809] (0/4) Epoch 13, batch 3150, loss[ctc_loss=0.07358, att_loss=0.2396, loss=0.2064, over 16471.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006267, over 46.00 utterances.], tot_loss[ctc_loss=0.09546, att_loss=0.2469, loss=0.2166, over 3271310.65 frames. utt_duration=1207 frames, utt_pad_proportion=0.06578, over 10851.92 utterances.], batch size: 46, lr: 8.27e-03, grad_scale: 8.0 2023-03-08 09:00:45,786 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-08 09:00:46,289 INFO [train2.py:809] (0/4) Epoch 13, batch 3200, loss[ctc_loss=0.08386, att_loss=0.2243, loss=0.1962, over 15889.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008973, over 39.00 utterances.], tot_loss[ctc_loss=0.09497, att_loss=0.246, loss=0.2158, over 3269266.25 frames. utt_duration=1227 frames, utt_pad_proportion=0.06107, over 10669.88 utterances.], batch size: 39, lr: 8.26e-03, grad_scale: 8.0 2023-03-08 09:00:48,220 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51007.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:00:52,691 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.2451, 5.3294, 5.7620, 5.5943, 5.7145, 6.1444, 5.2749, 6.2306], device='cuda:0'), covar=tensor([0.0594, 0.0675, 0.0657, 0.1011, 0.1676, 0.0777, 0.0595, 0.0625], device='cuda:0'), in_proj_covar=tensor([0.0763, 0.0452, 0.0533, 0.0591, 0.0782, 0.0540, 0.0431, 0.0522], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:01:17,932 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.247e+02 2.628e+02 3.429e+02 8.157e+02, threshold=5.255e+02, percent-clipped=5.0 2023-03-08 09:02:05,326 INFO [train2.py:809] (0/4) Epoch 13, batch 3250, loss[ctc_loss=0.1138, att_loss=0.2642, loss=0.2342, over 17056.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009499, over 53.00 utterances.], tot_loss[ctc_loss=0.09511, att_loss=0.2466, loss=0.2163, over 3268509.66 frames. utt_duration=1203 frames, utt_pad_proportion=0.0676, over 10878.00 utterances.], batch size: 53, lr: 8.26e-03, grad_scale: 8.0 2023-03-08 09:02:14,788 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51062.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:03:23,230 INFO [train2.py:809] (0/4) Epoch 13, batch 3300, loss[ctc_loss=0.08605, att_loss=0.2237, loss=0.1962, over 15383.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01049, over 35.00 utterances.], tot_loss[ctc_loss=0.09509, att_loss=0.2465, loss=0.2162, over 3266913.37 frames. utt_duration=1219 frames, utt_pad_proportion=0.06537, over 10737.23 utterances.], batch size: 35, lr: 8.26e-03, grad_scale: 8.0 2023-03-08 09:03:34,517 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6862, 2.6129, 5.1587, 4.2071, 2.8786, 4.5855, 5.0645, 4.7374], device='cuda:0'), covar=tensor([0.0241, 0.1575, 0.0213, 0.0773, 0.1897, 0.0201, 0.0104, 0.0252], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0243, 0.0146, 0.0307, 0.0270, 0.0187, 0.0130, 0.0161], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 09:03:51,907 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51123.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:03:54,635 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.818e+02 2.303e+02 2.804e+02 3.427e+02 6.539e+02, threshold=5.609e+02, percent-clipped=6.0 2023-03-08 09:03:56,939 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 09:04:14,234 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51138.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:04:41,662 INFO [train2.py:809] (0/4) Epoch 13, batch 3350, loss[ctc_loss=0.06457, att_loss=0.2122, loss=0.1827, over 14500.00 frames. utt_duration=1814 frames, utt_pad_proportion=0.032, over 32.00 utterances.], tot_loss[ctc_loss=0.09462, att_loss=0.246, loss=0.2158, over 3266725.62 frames. utt_duration=1234 frames, utt_pad_proportion=0.06074, over 10599.15 utterances.], batch size: 32, lr: 8.25e-03, grad_scale: 8.0 2023-03-08 09:04:54,974 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51164.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 09:05:08,516 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51172.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 09:05:55,383 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51202.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:06:01,136 INFO [train2.py:809] (0/4) Epoch 13, batch 3400, loss[ctc_loss=0.1276, att_loss=0.2725, loss=0.2435, over 17433.00 frames. utt_duration=1108 frames, utt_pad_proportion=0.03054, over 63.00 utterances.], tot_loss[ctc_loss=0.09639, att_loss=0.2471, loss=0.217, over 3264416.40 frames. utt_duration=1186 frames, utt_pad_proportion=0.07237, over 11020.02 utterances.], batch size: 63, lr: 8.25e-03, grad_scale: 8.0 2023-03-08 09:06:33,019 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.487e+02 2.909e+02 3.549e+02 7.479e+02, threshold=5.817e+02, percent-clipped=4.0 2023-03-08 09:06:45,875 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51233.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 09:07:11,545 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51250.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:07:20,299 INFO [train2.py:809] (0/4) Epoch 13, batch 3450, loss[ctc_loss=0.08678, att_loss=0.2511, loss=0.2182, over 16618.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005708, over 47.00 utterances.], tot_loss[ctc_loss=0.09566, att_loss=0.2472, loss=0.2169, over 3260371.92 frames. utt_duration=1194 frames, utt_pad_proportion=0.07015, over 10940.02 utterances.], batch size: 47, lr: 8.24e-03, grad_scale: 8.0 2023-03-08 09:08:33,561 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51302.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:08:39,554 INFO [train2.py:809] (0/4) Epoch 13, batch 3500, loss[ctc_loss=0.08854, att_loss=0.2444, loss=0.2132, over 16559.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005297, over 45.00 utterances.], tot_loss[ctc_loss=0.0948, att_loss=0.2471, loss=0.2167, over 3266884.43 frames. utt_duration=1209 frames, utt_pad_proportion=0.06603, over 10824.26 utterances.], batch size: 45, lr: 8.24e-03, grad_scale: 8.0 2023-03-08 09:09:10,403 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.552e+02 2.305e+02 2.663e+02 3.285e+02 9.490e+02, threshold=5.326e+02, percent-clipped=4.0 2023-03-08 09:09:57,628 INFO [train2.py:809] (0/4) Epoch 13, batch 3550, loss[ctc_loss=0.09022, att_loss=0.2597, loss=0.2258, over 17144.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01372, over 56.00 utterances.], tot_loss[ctc_loss=0.09412, att_loss=0.2465, loss=0.216, over 3276224.70 frames. utt_duration=1219 frames, utt_pad_proportion=0.05958, over 10766.30 utterances.], batch size: 56, lr: 8.24e-03, grad_scale: 8.0 2023-03-08 09:10:39,763 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51382.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:10:54,961 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51392.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:11:16,642 INFO [train2.py:809] (0/4) Epoch 13, batch 3600, loss[ctc_loss=0.08858, att_loss=0.2189, loss=0.1928, over 14553.00 frames. utt_duration=1821 frames, utt_pad_proportion=0.03303, over 32.00 utterances.], tot_loss[ctc_loss=0.09353, att_loss=0.2461, loss=0.2156, over 3272355.27 frames. utt_duration=1236 frames, utt_pad_proportion=0.05756, over 10600.05 utterances.], batch size: 32, lr: 8.23e-03, grad_scale: 8.0 2023-03-08 09:11:36,222 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51418.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:11:45,131 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9793, 3.7882, 3.2367, 3.5517, 4.0081, 3.7391, 2.9476, 4.2600], device='cuda:0'), covar=tensor([0.1119, 0.0545, 0.1073, 0.0653, 0.0649, 0.0652, 0.0914, 0.0541], device='cuda:0'), in_proj_covar=tensor([0.0187, 0.0192, 0.0205, 0.0181, 0.0242, 0.0219, 0.0185, 0.0259], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 09:11:45,612 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2023-03-08 09:11:46,269 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.606e+02 2.412e+02 2.884e+02 3.551e+02 7.030e+02, threshold=5.767e+02, percent-clipped=4.0 2023-03-08 09:12:05,842 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51438.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:12:07,432 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5654, 2.7587, 3.6827, 2.9812, 3.5585, 4.7027, 4.4589, 3.2249], device='cuda:0'), covar=tensor([0.0446, 0.2018, 0.1213, 0.1462, 0.1108, 0.0786, 0.0630, 0.1390], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0235, 0.0256, 0.0209, 0.0252, 0.0328, 0.0234, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:12:13,319 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51443.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:12:29,022 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51453.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:12:33,509 INFO [train2.py:809] (0/4) Epoch 13, batch 3650, loss[ctc_loss=0.07595, att_loss=0.2343, loss=0.2026, over 16107.00 frames. utt_duration=1535 frames, utt_pad_proportion=0.007526, over 42.00 utterances.], tot_loss[ctc_loss=0.0945, att_loss=0.2465, loss=0.2161, over 3278893.65 frames. utt_duration=1245 frames, utt_pad_proportion=0.05316, over 10548.21 utterances.], batch size: 42, lr: 8.23e-03, grad_scale: 8.0 2023-03-08 09:12:47,022 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51464.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 09:13:16,142 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8163, 3.0359, 3.8854, 3.1929, 3.7765, 4.8464, 4.6473, 3.3852], device='cuda:0'), covar=tensor([0.0319, 0.1747, 0.0987, 0.1338, 0.0904, 0.0859, 0.0529, 0.1264], device='cuda:0'), in_proj_covar=tensor([0.0233, 0.0233, 0.0253, 0.0208, 0.0251, 0.0325, 0.0232, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:13:20,392 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51486.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:13:20,639 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5506, 2.7710, 3.7527, 2.9791, 3.5432, 4.6377, 4.4299, 3.1763], device='cuda:0'), covar=tensor([0.0397, 0.1946, 0.1022, 0.1410, 0.1066, 0.0782, 0.0556, 0.1392], device='cuda:0'), in_proj_covar=tensor([0.0232, 0.0233, 0.0253, 0.0207, 0.0250, 0.0325, 0.0232, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:13:34,814 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51495.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:13:52,933 INFO [train2.py:809] (0/4) Epoch 13, batch 3700, loss[ctc_loss=0.1069, att_loss=0.2654, loss=0.2337, over 17365.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03509, over 63.00 utterances.], tot_loss[ctc_loss=0.09478, att_loss=0.2469, loss=0.2164, over 3261171.97 frames. utt_duration=1221 frames, utt_pad_proportion=0.06394, over 10700.38 utterances.], batch size: 63, lr: 8.22e-03, grad_scale: 8.0 2023-03-08 09:14:03,447 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51512.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 09:14:23,031 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.328e+02 2.806e+02 3.830e+02 9.896e+02, threshold=5.611e+02, percent-clipped=3.0 2023-03-08 09:14:28,086 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51528.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 09:15:05,532 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.02 vs. limit=2.0 2023-03-08 09:15:13,026 INFO [train2.py:809] (0/4) Epoch 13, batch 3750, loss[ctc_loss=0.07672, att_loss=0.2401, loss=0.2075, over 16113.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.005794, over 42.00 utterances.], tot_loss[ctc_loss=0.09515, att_loss=0.2473, loss=0.2169, over 3270391.39 frames. utt_duration=1204 frames, utt_pad_proportion=0.06614, over 10880.09 utterances.], batch size: 42, lr: 8.22e-03, grad_scale: 8.0 2023-03-08 09:15:13,446 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51556.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:16:10,357 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.06 vs. limit=5.0 2023-03-08 09:16:15,867 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8818, 5.1059, 5.1444, 5.0144, 5.1153, 5.0794, 4.8192, 4.6480], device='cuda:0'), covar=tensor([0.1156, 0.0565, 0.0268, 0.0550, 0.0365, 0.0361, 0.0373, 0.0375], device='cuda:0'), in_proj_covar=tensor([0.0470, 0.0309, 0.0280, 0.0306, 0.0367, 0.0383, 0.0311, 0.0346], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 09:16:26,914 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51602.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:16:32,779 INFO [train2.py:809] (0/4) Epoch 13, batch 3800, loss[ctc_loss=0.0832, att_loss=0.2364, loss=0.2058, over 16170.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006232, over 41.00 utterances.], tot_loss[ctc_loss=0.09474, att_loss=0.247, loss=0.2165, over 3264789.47 frames. utt_duration=1212 frames, utt_pad_proportion=0.06648, over 10792.26 utterances.], batch size: 41, lr: 8.22e-03, grad_scale: 8.0 2023-03-08 09:17:02,027 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-08 09:17:02,532 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.490e+02 2.194e+02 2.674e+02 3.110e+02 7.500e+02, threshold=5.349e+02, percent-clipped=1.0 2023-03-08 09:17:41,329 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51650.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:17:51,574 INFO [train2.py:809] (0/4) Epoch 13, batch 3850, loss[ctc_loss=0.08092, att_loss=0.2433, loss=0.2108, over 16753.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006663, over 48.00 utterances.], tot_loss[ctc_loss=0.09387, att_loss=0.2463, loss=0.2158, over 3264498.65 frames. utt_duration=1229 frames, utt_pad_proportion=0.06035, over 10639.83 utterances.], batch size: 48, lr: 8.21e-03, grad_scale: 8.0 2023-03-08 09:18:04,587 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2023-03-08 09:19:07,769 INFO [train2.py:809] (0/4) Epoch 13, batch 3900, loss[ctc_loss=0.1022, att_loss=0.257, loss=0.2261, over 16953.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008254, over 50.00 utterances.], tot_loss[ctc_loss=0.09412, att_loss=0.2466, loss=0.2161, over 3263144.84 frames. utt_duration=1209 frames, utt_pad_proportion=0.06608, over 10807.65 utterances.], batch size: 50, lr: 8.21e-03, grad_scale: 8.0 2023-03-08 09:19:26,174 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51718.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:19:36,508 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.485e+02 2.131e+02 2.698e+02 3.249e+02 8.076e+02, threshold=5.395e+02, percent-clipped=6.0 2023-03-08 09:19:56,058 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51738.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:20:10,164 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=51747.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:20:11,508 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51748.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:20:23,591 INFO [train2.py:809] (0/4) Epoch 13, batch 3950, loss[ctc_loss=0.1055, att_loss=0.2598, loss=0.229, over 17284.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02447, over 59.00 utterances.], tot_loss[ctc_loss=0.09475, att_loss=0.247, loss=0.2166, over 3262096.35 frames. utt_duration=1191 frames, utt_pad_proportion=0.07131, over 10968.45 utterances.], batch size: 59, lr: 8.20e-03, grad_scale: 8.0 2023-03-08 09:20:38,801 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51766.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:21:14,591 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-13.pt 2023-03-08 09:21:41,017 INFO [train2.py:809] (0/4) Epoch 14, batch 0, loss[ctc_loss=0.1054, att_loss=0.2564, loss=0.2262, over 17071.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008673, over 53.00 utterances.], tot_loss[ctc_loss=0.1054, att_loss=0.2564, loss=0.2262, over 17071.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008673, over 53.00 utterances.], batch size: 53, lr: 7.90e-03, grad_scale: 8.0 2023-03-08 09:21:41,019 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 09:21:52,753 INFO [train2.py:843] (0/4) Epoch 14, validation: ctc_loss=0.04501, att_loss=0.2367, loss=0.1984, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 09:21:52,754 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 09:22:20,781 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=51808.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:22:46,826 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.322e+02 2.890e+02 3.740e+02 8.433e+02, threshold=5.780e+02, percent-clipped=5.0 2023-03-08 09:22:49,378 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 09:22:52,362 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=51828.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 09:23:04,216 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-03-08 09:23:11,220 INFO [train2.py:809] (0/4) Epoch 14, batch 50, loss[ctc_loss=0.1069, att_loss=0.2609, loss=0.2301, over 17031.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009927, over 52.00 utterances.], tot_loss[ctc_loss=0.09376, att_loss=0.2463, loss=0.2158, over 747260.94 frames. utt_duration=1290 frames, utt_pad_proportion=0.03575, over 2319.18 utterances.], batch size: 52, lr: 7.90e-03, grad_scale: 8.0 2023-03-08 09:23:29,030 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=51851.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:24:08,129 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=51876.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 09:24:29,805 INFO [train2.py:809] (0/4) Epoch 14, batch 100, loss[ctc_loss=0.06714, att_loss=0.2282, loss=0.196, over 16629.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.0052, over 47.00 utterances.], tot_loss[ctc_loss=0.09189, att_loss=0.246, loss=0.2152, over 1309999.27 frames. utt_duration=1241 frames, utt_pad_proportion=0.04922, over 4228.21 utterances.], batch size: 47, lr: 7.90e-03, grad_scale: 8.0 2023-03-08 09:25:03,376 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5617, 2.2321, 5.0067, 3.9188, 2.9961, 4.3155, 4.6614, 4.6834], device='cuda:0'), covar=tensor([0.0184, 0.1756, 0.0102, 0.0901, 0.1777, 0.0232, 0.0111, 0.0190], device='cuda:0'), in_proj_covar=tensor([0.0155, 0.0239, 0.0145, 0.0302, 0.0265, 0.0186, 0.0130, 0.0160], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 09:25:24,994 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.393e+02 2.937e+02 3.541e+02 1.195e+03, threshold=5.873e+02, percent-clipped=2.0 2023-03-08 09:25:48,792 INFO [train2.py:809] (0/4) Epoch 14, batch 150, loss[ctc_loss=0.1117, att_loss=0.2481, loss=0.2208, over 16547.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005792, over 45.00 utterances.], tot_loss[ctc_loss=0.09427, att_loss=0.2478, loss=0.2171, over 1748235.86 frames. utt_duration=1221 frames, utt_pad_proportion=0.05679, over 5735.65 utterances.], batch size: 45, lr: 7.89e-03, grad_scale: 8.0 2023-03-08 09:26:28,867 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8221, 2.3963, 2.6164, 3.2741, 3.1174, 3.2700, 2.5427, 2.2186], device='cuda:0'), covar=tensor([0.0639, 0.1777, 0.0949, 0.0717, 0.0769, 0.0465, 0.1383, 0.1829], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0214, 0.0191, 0.0206, 0.0204, 0.0163, 0.0200, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 09:26:56,457 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0850, 5.1170, 5.0638, 2.3197, 1.9244, 2.9142, 2.2930, 3.8141], device='cuda:0'), covar=tensor([0.0707, 0.0241, 0.0202, 0.4759, 0.6265, 0.2581, 0.3426, 0.1754], device='cuda:0'), in_proj_covar=tensor([0.0346, 0.0239, 0.0246, 0.0225, 0.0351, 0.0338, 0.0236, 0.0357], device='cuda:0'), out_proj_covar=tensor([1.5275e-04, 8.8419e-05, 1.0621e-04, 9.8406e-05, 1.5032e-04, 1.3506e-04, 9.3965e-05, 1.4913e-04], device='cuda:0') 2023-03-08 09:27:07,210 INFO [train2.py:809] (0/4) Epoch 14, batch 200, loss[ctc_loss=0.1209, att_loss=0.2713, loss=0.2412, over 17060.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009137, over 53.00 utterances.], tot_loss[ctc_loss=0.09299, att_loss=0.2468, loss=0.216, over 2092403.73 frames. utt_duration=1243 frames, utt_pad_proportion=0.05052, over 6740.37 utterances.], batch size: 53, lr: 7.89e-03, grad_scale: 8.0 2023-03-08 09:27:22,383 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-52000.pt 2023-03-08 09:28:06,026 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.576e+02 2.279e+02 2.700e+02 3.291e+02 7.221e+02, threshold=5.400e+02, percent-clipped=2.0 2023-03-08 09:28:27,663 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52038.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:28:30,294 INFO [train2.py:809] (0/4) Epoch 14, batch 250, loss[ctc_loss=0.1298, att_loss=0.2747, loss=0.2457, over 17025.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007497, over 51.00 utterances.], tot_loss[ctc_loss=0.09152, att_loss=0.2447, loss=0.2141, over 2350446.19 frames. utt_duration=1260 frames, utt_pad_proportion=0.04881, over 7472.21 utterances.], batch size: 51, lr: 7.88e-03, grad_scale: 8.0 2023-03-08 09:28:30,640 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52040.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:28:42,582 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52048.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:29:34,334 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0871, 5.4129, 5.6383, 5.5463, 5.5550, 6.0971, 5.3267, 6.1942], device='cuda:0'), covar=tensor([0.0654, 0.0684, 0.0734, 0.0989, 0.1840, 0.0776, 0.0582, 0.0597], device='cuda:0'), in_proj_covar=tensor([0.0758, 0.0448, 0.0532, 0.0579, 0.0773, 0.0539, 0.0436, 0.0523], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:29:42,210 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52086.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:29:45,590 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4424, 3.4402, 3.2136, 2.8422, 3.3025, 3.2839, 3.3313, 2.3850], device='cuda:0'), covar=tensor([0.1058, 0.1750, 0.3109, 0.6485, 0.1201, 0.4527, 0.1090, 0.5903], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0138, 0.0152, 0.0221, 0.0113, 0.0204, 0.0125, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 09:29:48,098 INFO [train2.py:809] (0/4) Epoch 14, batch 300, loss[ctc_loss=0.09277, att_loss=0.2355, loss=0.2069, over 16007.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007388, over 40.00 utterances.], tot_loss[ctc_loss=0.09342, att_loss=0.246, loss=0.2155, over 2548454.25 frames. utt_duration=1231 frames, utt_pad_proportion=0.06092, over 8293.44 utterances.], batch size: 40, lr: 7.88e-03, grad_scale: 16.0 2023-03-08 09:29:57,154 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52096.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:29:58,953 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1967, 3.8893, 3.2037, 3.6205, 4.0245, 3.6794, 3.2427, 4.4090], device='cuda:0'), covar=tensor([0.0863, 0.0390, 0.0966, 0.0591, 0.0655, 0.0598, 0.0683, 0.0414], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0194, 0.0210, 0.0182, 0.0246, 0.0221, 0.0186, 0.0262], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 09:30:05,843 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52101.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:30:05,868 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1969, 4.6452, 4.6201, 4.9634, 2.6462, 4.8045, 3.0383, 1.7792], device='cuda:0'), covar=tensor([0.0324, 0.0157, 0.0554, 0.0120, 0.1731, 0.0138, 0.1289, 0.1757], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0128, 0.0257, 0.0121, 0.0223, 0.0114, 0.0226, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 09:30:08,641 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52103.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:30:42,751 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 2.191e+02 2.821e+02 3.545e+02 1.097e+03, threshold=5.643e+02, percent-clipped=10.0 2023-03-08 09:31:06,325 INFO [train2.py:809] (0/4) Epoch 14, batch 350, loss[ctc_loss=0.07973, att_loss=0.2222, loss=0.1937, over 15527.00 frames. utt_duration=1727 frames, utt_pad_proportion=0.006964, over 36.00 utterances.], tot_loss[ctc_loss=0.09274, att_loss=0.2453, loss=0.2148, over 2709518.39 frames. utt_duration=1259 frames, utt_pad_proportion=0.05371, over 8616.48 utterances.], batch size: 36, lr: 7.88e-03, grad_scale: 16.0 2023-03-08 09:31:23,437 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9997, 5.3484, 4.8357, 5.4158, 4.7403, 5.0286, 5.4590, 5.2291], device='cuda:0'), covar=tensor([0.0581, 0.0243, 0.0755, 0.0236, 0.0415, 0.0243, 0.0196, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0270, 0.0322, 0.0279, 0.0279, 0.0210, 0.0261, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0005, 0.0005, 0.0004, 0.0005, 0.0004], device='cuda:0') 2023-03-08 09:31:23,497 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52151.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:32:25,893 INFO [train2.py:809] (0/4) Epoch 14, batch 400, loss[ctc_loss=0.1037, att_loss=0.2405, loss=0.2131, over 16551.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005672, over 45.00 utterances.], tot_loss[ctc_loss=0.09327, att_loss=0.2461, loss=0.2155, over 2839531.26 frames. utt_duration=1240 frames, utt_pad_proportion=0.05632, over 9168.17 utterances.], batch size: 45, lr: 7.87e-03, grad_scale: 16.0 2023-03-08 09:32:31,104 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52193.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:32:40,136 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52199.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:33:21,767 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.597e+02 2.286e+02 2.626e+02 3.493e+02 1.383e+03, threshold=5.251e+02, percent-clipped=5.0 2023-03-08 09:33:46,075 INFO [train2.py:809] (0/4) Epoch 14, batch 450, loss[ctc_loss=0.055, att_loss=0.2048, loss=0.1748, over 15344.00 frames. utt_duration=1755 frames, utt_pad_proportion=0.01294, over 35.00 utterances.], tot_loss[ctc_loss=0.09257, att_loss=0.2453, loss=0.2148, over 2936351.67 frames. utt_duration=1260 frames, utt_pad_proportion=0.05181, over 9329.77 utterances.], batch size: 35, lr: 7.87e-03, grad_scale: 16.0 2023-03-08 09:33:49,609 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5738, 2.9283, 3.6464, 2.9747, 3.5175, 4.6849, 4.4143, 3.3604], device='cuda:0'), covar=tensor([0.0367, 0.1904, 0.1251, 0.1416, 0.1181, 0.0883, 0.0589, 0.1309], device='cuda:0'), in_proj_covar=tensor([0.0232, 0.0235, 0.0253, 0.0208, 0.0250, 0.0325, 0.0231, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:34:09,173 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52254.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:35:04,913 INFO [train2.py:809] (0/4) Epoch 14, batch 500, loss[ctc_loss=0.08402, att_loss=0.2283, loss=0.1995, over 16399.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006956, over 44.00 utterances.], tot_loss[ctc_loss=0.09267, att_loss=0.2451, loss=0.2146, over 3010434.22 frames. utt_duration=1263 frames, utt_pad_proportion=0.0518, over 9547.60 utterances.], batch size: 44, lr: 7.87e-03, grad_scale: 16.0 2023-03-08 09:35:29,371 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-03-08 09:35:39,562 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52312.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:36:01,138 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.299e+02 2.275e+02 2.805e+02 3.649e+02 6.948e+02, threshold=5.610e+02, percent-clipped=5.0 2023-03-08 09:36:04,600 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7953, 2.9567, 3.8068, 3.1363, 3.6558, 4.8166, 4.6117, 3.4814], device='cuda:0'), covar=tensor([0.0310, 0.1918, 0.1094, 0.1329, 0.1027, 0.0805, 0.0497, 0.1240], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0232, 0.0253, 0.0207, 0.0250, 0.0324, 0.0231, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:36:23,207 INFO [train2.py:809] (0/4) Epoch 14, batch 550, loss[ctc_loss=0.09732, att_loss=0.2695, loss=0.2351, over 17310.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01154, over 55.00 utterances.], tot_loss[ctc_loss=0.09234, att_loss=0.2449, loss=0.2144, over 3062907.26 frames. utt_duration=1248 frames, utt_pad_proportion=0.05746, over 9832.75 utterances.], batch size: 55, lr: 7.86e-03, grad_scale: 8.0 2023-03-08 09:37:15,090 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52373.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:37:42,122 INFO [train2.py:809] (0/4) Epoch 14, batch 600, loss[ctc_loss=0.08318, att_loss=0.2418, loss=0.2101, over 16636.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004766, over 47.00 utterances.], tot_loss[ctc_loss=0.09154, att_loss=0.2442, loss=0.2136, over 3101464.68 frames. utt_duration=1258 frames, utt_pad_proportion=0.0575, over 9870.83 utterances.], batch size: 47, lr: 7.86e-03, grad_scale: 8.0 2023-03-08 09:37:51,401 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52396.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:38:03,110 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52403.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:38:39,774 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.147e+02 2.714e+02 3.419e+02 5.544e+02, threshold=5.429e+02, percent-clipped=0.0 2023-03-08 09:39:01,291 INFO [train2.py:809] (0/4) Epoch 14, batch 650, loss[ctc_loss=0.1057, att_loss=0.2491, loss=0.2204, over 17028.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007453, over 51.00 utterances.], tot_loss[ctc_loss=0.09154, att_loss=0.2442, loss=0.2136, over 3140101.54 frames. utt_duration=1259 frames, utt_pad_proportion=0.05621, over 9986.80 utterances.], batch size: 51, lr: 7.85e-03, grad_scale: 8.0 2023-03-08 09:39:18,702 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52451.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:39:51,691 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2058, 5.2656, 5.0648, 2.5365, 2.1298, 2.9362, 2.8480, 3.8879], device='cuda:0'), covar=tensor([0.0669, 0.0341, 0.0259, 0.4771, 0.5647, 0.2537, 0.2621, 0.1754], device='cuda:0'), in_proj_covar=tensor([0.0342, 0.0237, 0.0245, 0.0223, 0.0348, 0.0335, 0.0231, 0.0353], device='cuda:0'), out_proj_covar=tensor([1.5068e-04, 8.8237e-05, 1.0640e-04, 9.7536e-05, 1.4882e-04, 1.3395e-04, 9.1641e-05, 1.4740e-04], device='cuda:0') 2023-03-08 09:40:19,852 INFO [train2.py:809] (0/4) Epoch 14, batch 700, loss[ctc_loss=0.06352, att_loss=0.2082, loss=0.1793, over 14072.00 frames. utt_duration=1817 frames, utt_pad_proportion=0.03124, over 31.00 utterances.], tot_loss[ctc_loss=0.09113, att_loss=0.2439, loss=0.2134, over 3169384.77 frames. utt_duration=1279 frames, utt_pad_proportion=0.05001, over 9923.25 utterances.], batch size: 31, lr: 7.85e-03, grad_scale: 8.0 2023-03-08 09:41:15,851 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 1.961e+02 2.440e+02 3.056e+02 5.134e+02, threshold=4.881e+02, percent-clipped=0.0 2023-03-08 09:41:22,236 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-03-08 09:41:27,564 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3349, 4.6073, 4.6329, 4.9040, 2.7473, 4.7547, 2.7400, 2.0198], device='cuda:0'), covar=tensor([0.0266, 0.0170, 0.0597, 0.0110, 0.1686, 0.0126, 0.1541, 0.1628], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0130, 0.0258, 0.0120, 0.0224, 0.0115, 0.0227, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 09:41:37,731 INFO [train2.py:809] (0/4) Epoch 14, batch 750, loss[ctc_loss=0.1023, att_loss=0.2644, loss=0.232, over 17353.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03584, over 63.00 utterances.], tot_loss[ctc_loss=0.09153, att_loss=0.2443, loss=0.2138, over 3190320.85 frames. utt_duration=1275 frames, utt_pad_proportion=0.04966, over 10023.72 utterances.], batch size: 63, lr: 7.85e-03, grad_scale: 8.0 2023-03-08 09:41:52,172 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52549.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:41:58,960 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0573, 5.3443, 4.8145, 5.3982, 4.7700, 5.0602, 5.4771, 5.2288], device='cuda:0'), covar=tensor([0.0508, 0.0264, 0.0804, 0.0225, 0.0389, 0.0198, 0.0186, 0.0172], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0274, 0.0327, 0.0284, 0.0282, 0.0213, 0.0260, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 09:42:02,138 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52555.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 09:42:56,735 INFO [train2.py:809] (0/4) Epoch 14, batch 800, loss[ctc_loss=0.0889, att_loss=0.2426, loss=0.2118, over 16472.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006132, over 46.00 utterances.], tot_loss[ctc_loss=0.09164, att_loss=0.2452, loss=0.2145, over 3217502.77 frames. utt_duration=1278 frames, utt_pad_proportion=0.04496, over 10080.77 utterances.], batch size: 46, lr: 7.84e-03, grad_scale: 8.0 2023-03-08 09:43:38,056 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52616.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 09:43:39,424 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4691, 2.9747, 3.6240, 2.8400, 3.5045, 4.6211, 4.4172, 3.2353], device='cuda:0'), covar=tensor([0.0387, 0.1630, 0.1044, 0.1403, 0.0963, 0.0740, 0.0483, 0.1226], device='cuda:0'), in_proj_covar=tensor([0.0232, 0.0232, 0.0253, 0.0208, 0.0248, 0.0325, 0.0232, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 09:43:43,823 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 09:43:54,264 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.071e+02 2.560e+02 3.132e+02 4.899e+02, threshold=5.119e+02, percent-clipped=1.0 2023-03-08 09:44:16,111 INFO [train2.py:809] (0/4) Epoch 14, batch 850, loss[ctc_loss=0.08366, att_loss=0.2555, loss=0.2212, over 17304.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01184, over 55.00 utterances.], tot_loss[ctc_loss=0.09227, att_loss=0.2452, loss=0.2146, over 3223136.84 frames. utt_duration=1247 frames, utt_pad_proportion=0.05507, over 10348.28 utterances.], batch size: 55, lr: 7.84e-03, grad_scale: 8.0 2023-03-08 09:44:20,931 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8905, 5.1660, 5.0885, 5.0547, 5.2068, 5.1592, 4.8515, 4.7001], device='cuda:0'), covar=tensor([0.0939, 0.0495, 0.0303, 0.0460, 0.0296, 0.0286, 0.0353, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0475, 0.0314, 0.0279, 0.0307, 0.0360, 0.0378, 0.0311, 0.0346], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 09:45:00,848 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52668.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:45:32,680 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1494, 4.5868, 4.2431, 4.6997, 2.5885, 4.4122, 2.4111, 2.3242], device='cuda:0'), covar=tensor([0.0399, 0.0135, 0.0698, 0.0135, 0.1886, 0.0169, 0.1761, 0.1442], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0130, 0.0259, 0.0122, 0.0226, 0.0116, 0.0229, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 09:45:35,371 INFO [train2.py:809] (0/4) Epoch 14, batch 900, loss[ctc_loss=0.09161, att_loss=0.2405, loss=0.2108, over 16394.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007364, over 44.00 utterances.], tot_loss[ctc_loss=0.09302, att_loss=0.2459, loss=0.2153, over 3238127.95 frames. utt_duration=1225 frames, utt_pad_proportion=0.05938, over 10585.28 utterances.], batch size: 44, lr: 7.84e-03, grad_scale: 8.0 2023-03-08 09:45:44,717 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52696.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:46:32,364 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.422e+02 2.184e+02 2.605e+02 3.188e+02 5.544e+02, threshold=5.210e+02, percent-clipped=2.0 2023-03-08 09:46:53,846 INFO [train2.py:809] (0/4) Epoch 14, batch 950, loss[ctc_loss=0.07925, att_loss=0.2477, loss=0.214, over 16890.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006244, over 49.00 utterances.], tot_loss[ctc_loss=0.09254, att_loss=0.2455, loss=0.2149, over 3252696.29 frames. utt_duration=1241 frames, utt_pad_proportion=0.05486, over 10496.88 utterances.], batch size: 49, lr: 7.83e-03, grad_scale: 8.0 2023-03-08 09:47:00,131 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52744.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:47:12,385 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=52752.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:47:59,774 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-03-08 09:48:12,276 INFO [train2.py:809] (0/4) Epoch 14, batch 1000, loss[ctc_loss=0.07618, att_loss=0.2212, loss=0.1922, over 15771.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008604, over 38.00 utterances.], tot_loss[ctc_loss=0.09246, att_loss=0.2454, loss=0.2148, over 3255993.82 frames. utt_duration=1236 frames, utt_pad_proportion=0.05797, over 10545.78 utterances.], batch size: 38, lr: 7.83e-03, grad_scale: 8.0 2023-03-08 09:48:48,521 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=52813.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:49:09,713 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.532e+02 2.243e+02 2.663e+02 3.401e+02 6.618e+02, threshold=5.327e+02, percent-clipped=3.0 2023-03-08 09:49:31,215 INFO [train2.py:809] (0/4) Epoch 14, batch 1050, loss[ctc_loss=0.08698, att_loss=0.2407, loss=0.21, over 16113.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006987, over 42.00 utterances.], tot_loss[ctc_loss=0.09129, att_loss=0.2449, loss=0.2142, over 3257708.13 frames. utt_duration=1238 frames, utt_pad_proportion=0.05787, over 10540.40 utterances.], batch size: 42, lr: 7.82e-03, grad_scale: 4.0 2023-03-08 09:49:45,242 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52849.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:50:49,485 INFO [train2.py:809] (0/4) Epoch 14, batch 1100, loss[ctc_loss=0.07991, att_loss=0.2452, loss=0.2121, over 16312.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.007098, over 45.00 utterances.], tot_loss[ctc_loss=0.09134, att_loss=0.2452, loss=0.2144, over 3260912.34 frames. utt_duration=1242 frames, utt_pad_proportion=0.05646, over 10512.66 utterances.], batch size: 45, lr: 7.82e-03, grad_scale: 4.0 2023-03-08 09:51:00,109 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=52897.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:51:23,329 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=52911.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 09:51:47,355 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.88 vs. limit=2.0 2023-03-08 09:51:48,092 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.396e+02 2.277e+02 2.760e+02 3.306e+02 6.438e+02, threshold=5.521e+02, percent-clipped=4.0 2023-03-08 09:52:07,935 INFO [train2.py:809] (0/4) Epoch 14, batch 1150, loss[ctc_loss=0.1816, att_loss=0.2899, loss=0.2682, over 14037.00 frames. utt_duration=385.9 frames, utt_pad_proportion=0.3265, over 146.00 utterances.], tot_loss[ctc_loss=0.09333, att_loss=0.2461, loss=0.2156, over 3266305.43 frames. utt_duration=1222 frames, utt_pad_proportion=0.06055, over 10701.46 utterances.], batch size: 146, lr: 7.82e-03, grad_scale: 4.0 2023-03-08 09:52:52,779 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=52968.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:53:26,827 INFO [train2.py:809] (0/4) Epoch 14, batch 1200, loss[ctc_loss=0.0842, att_loss=0.2324, loss=0.2027, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.006673, over 44.00 utterances.], tot_loss[ctc_loss=0.09226, att_loss=0.2452, loss=0.2146, over 3258811.78 frames. utt_duration=1230 frames, utt_pad_proportion=0.05995, over 10613.71 utterances.], batch size: 44, lr: 7.81e-03, grad_scale: 8.0 2023-03-08 09:53:35,399 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-03-08 09:53:47,760 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5365, 4.6018, 4.5366, 4.6313, 4.9949, 4.6993, 4.5882, 2.4323], device='cuda:0'), covar=tensor([0.0177, 0.0206, 0.0247, 0.0151, 0.0824, 0.0162, 0.0213, 0.2062], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0141, 0.0147, 0.0159, 0.0344, 0.0127, 0.0131, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 09:54:08,668 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53016.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:54:25,773 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 2.184e+02 2.648e+02 3.563e+02 1.260e+03, threshold=5.296e+02, percent-clipped=9.0 2023-03-08 09:54:33,642 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0033, 5.3131, 5.3281, 5.3001, 5.3886, 5.2969, 5.1091, 4.8060], device='cuda:0'), covar=tensor([0.1039, 0.0468, 0.0228, 0.0376, 0.0258, 0.0266, 0.0250, 0.0321], device='cuda:0'), in_proj_covar=tensor([0.0471, 0.0314, 0.0279, 0.0308, 0.0362, 0.0377, 0.0309, 0.0346], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 09:54:45,536 INFO [train2.py:809] (0/4) Epoch 14, batch 1250, loss[ctc_loss=0.09628, att_loss=0.2695, loss=0.2348, over 17125.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01379, over 56.00 utterances.], tot_loss[ctc_loss=0.09208, att_loss=0.2456, loss=0.2149, over 3264564.65 frames. utt_duration=1219 frames, utt_pad_proportion=0.0621, over 10725.92 utterances.], batch size: 56, lr: 7.81e-03, grad_scale: 8.0 2023-03-08 09:55:10,501 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53056.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:56:03,883 INFO [train2.py:809] (0/4) Epoch 14, batch 1300, loss[ctc_loss=0.135, att_loss=0.2769, loss=0.2485, over 13813.00 frames. utt_duration=380 frames, utt_pad_proportion=0.338, over 146.00 utterances.], tot_loss[ctc_loss=0.09259, att_loss=0.2457, loss=0.2151, over 3259677.45 frames. utt_duration=1208 frames, utt_pad_proportion=0.06573, over 10803.28 utterances.], batch size: 146, lr: 7.81e-03, grad_scale: 8.0 2023-03-08 09:56:32,021 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53108.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:56:47,500 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53117.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 09:57:03,198 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.477e+02 2.168e+02 2.572e+02 3.596e+02 7.210e+02, threshold=5.144e+02, percent-clipped=5.0 2023-03-08 09:57:12,950 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8040, 3.5931, 3.4807, 3.0742, 3.5883, 3.6755, 3.6179, 2.6710], device='cuda:0'), covar=tensor([0.1059, 0.1716, 0.4605, 0.5817, 0.1597, 0.2581, 0.0992, 0.5986], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0138, 0.0149, 0.0221, 0.0114, 0.0205, 0.0127, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 09:57:23,233 INFO [train2.py:809] (0/4) Epoch 14, batch 1350, loss[ctc_loss=0.07735, att_loss=0.2366, loss=0.2047, over 15939.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007404, over 41.00 utterances.], tot_loss[ctc_loss=0.09235, att_loss=0.2457, loss=0.215, over 3259820.46 frames. utt_duration=1195 frames, utt_pad_proportion=0.06878, over 10923.05 utterances.], batch size: 41, lr: 7.80e-03, grad_scale: 8.0 2023-03-08 09:58:26,004 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.61 vs. limit=2.0 2023-03-08 09:58:41,979 INFO [train2.py:809] (0/4) Epoch 14, batch 1400, loss[ctc_loss=0.07496, att_loss=0.2171, loss=0.1887, over 15851.00 frames. utt_duration=1627 frames, utt_pad_proportion=0.00952, over 39.00 utterances.], tot_loss[ctc_loss=0.09268, att_loss=0.2455, loss=0.2149, over 3255219.55 frames. utt_duration=1197 frames, utt_pad_proportion=0.06773, over 10894.07 utterances.], batch size: 39, lr: 7.80e-03, grad_scale: 8.0 2023-03-08 09:59:16,391 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53211.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 09:59:41,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.260e+02 2.643e+02 3.236e+02 5.232e+02, threshold=5.286e+02, percent-clipped=1.0 2023-03-08 09:59:52,543 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3476, 2.8371, 3.0711, 4.2706, 4.0152, 4.0769, 3.1509, 2.2820], device='cuda:0'), covar=tensor([0.0775, 0.2402, 0.1371, 0.0728, 0.0719, 0.0436, 0.1267, 0.2369], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0214, 0.0191, 0.0206, 0.0206, 0.0163, 0.0199, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 10:00:00,306 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5970, 4.5066, 4.5163, 4.6575, 5.1321, 4.6236, 4.6482, 2.2785], device='cuda:0'), covar=tensor([0.0191, 0.0264, 0.0263, 0.0215, 0.0824, 0.0183, 0.0215, 0.2034], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0143, 0.0150, 0.0160, 0.0349, 0.0129, 0.0133, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 10:00:01,540 INFO [train2.py:809] (0/4) Epoch 14, batch 1450, loss[ctc_loss=0.05761, att_loss=0.234, loss=0.1987, over 17021.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008655, over 51.00 utterances.], tot_loss[ctc_loss=0.09192, att_loss=0.2455, loss=0.2148, over 3262607.34 frames. utt_duration=1204 frames, utt_pad_proportion=0.06627, over 10856.59 utterances.], batch size: 51, lr: 7.80e-03, grad_scale: 8.0 2023-03-08 10:00:03,437 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53241.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:00:31,932 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53259.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 10:00:46,111 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.52 vs. limit=5.0 2023-03-08 10:01:21,379 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-03-08 10:01:21,617 INFO [train2.py:809] (0/4) Epoch 14, batch 1500, loss[ctc_loss=0.1027, att_loss=0.2649, loss=0.2325, over 17514.00 frames. utt_duration=1017 frames, utt_pad_proportion=0.04173, over 69.00 utterances.], tot_loss[ctc_loss=0.09109, att_loss=0.2455, loss=0.2146, over 3271921.96 frames. utt_duration=1229 frames, utt_pad_proportion=0.05774, over 10659.03 utterances.], batch size: 69, lr: 7.79e-03, grad_scale: 8.0 2023-03-08 10:01:41,540 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53302.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:01:44,587 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53304.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:02:02,590 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53315.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:02:20,570 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.202e+02 2.545e+02 3.206e+02 8.434e+02, threshold=5.090e+02, percent-clipped=3.0 2023-03-08 10:02:41,147 INFO [train2.py:809] (0/4) Epoch 14, batch 1550, loss[ctc_loss=0.07771, att_loss=0.2181, loss=0.1901, over 15782.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.006229, over 38.00 utterances.], tot_loss[ctc_loss=0.09115, att_loss=0.2455, loss=0.2146, over 3267718.86 frames. utt_duration=1210 frames, utt_pad_proportion=0.06489, over 10811.31 utterances.], batch size: 38, lr: 7.79e-03, grad_scale: 8.0 2023-03-08 10:03:21,746 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53365.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:03:24,978 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=53367.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:03:38,813 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53376.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:03:59,967 INFO [train2.py:809] (0/4) Epoch 14, batch 1600, loss[ctc_loss=0.07766, att_loss=0.2346, loss=0.2032, over 16135.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005144, over 42.00 utterances.], tot_loss[ctc_loss=0.09087, att_loss=0.2454, loss=0.2145, over 3272902.33 frames. utt_duration=1233 frames, utt_pad_proportion=0.05862, over 10630.64 utterances.], batch size: 42, lr: 7.78e-03, grad_scale: 8.0 2023-03-08 10:04:29,524 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53408.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:04:36,047 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53412.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:04:58,719 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.287e+02 2.681e+02 3.372e+02 7.467e+02, threshold=5.361e+02, percent-clipped=4.0 2023-03-08 10:05:00,593 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=53428.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 10:05:19,366 INFO [train2.py:809] (0/4) Epoch 14, batch 1650, loss[ctc_loss=0.07867, att_loss=0.2512, loss=0.2167, over 17042.00 frames. utt_duration=1338 frames, utt_pad_proportion=0.007375, over 51.00 utterances.], tot_loss[ctc_loss=0.09168, att_loss=0.2462, loss=0.2153, over 3277437.92 frames. utt_duration=1224 frames, utt_pad_proportion=0.05904, over 10726.94 utterances.], batch size: 51, lr: 7.78e-03, grad_scale: 8.0 2023-03-08 10:05:19,791 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6540, 4.4551, 4.4572, 4.5627, 4.9627, 4.6906, 4.4952, 2.3634], device='cuda:0'), covar=tensor([0.0173, 0.0261, 0.0262, 0.0199, 0.0847, 0.0180, 0.0269, 0.2028], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0145, 0.0151, 0.0162, 0.0352, 0.0131, 0.0134, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 10:05:44,640 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53456.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:06:38,740 INFO [train2.py:809] (0/4) Epoch 14, batch 1700, loss[ctc_loss=0.06878, att_loss=0.2503, loss=0.214, over 16979.00 frames. utt_duration=1360 frames, utt_pad_proportion=0.006808, over 50.00 utterances.], tot_loss[ctc_loss=0.08979, att_loss=0.2447, loss=0.2138, over 3277244.32 frames. utt_duration=1263 frames, utt_pad_proportion=0.05048, over 10393.31 utterances.], batch size: 50, lr: 7.78e-03, grad_scale: 8.0 2023-03-08 10:06:58,006 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0435, 5.3499, 4.8322, 5.4843, 4.7634, 5.0733, 5.5166, 5.2811], device='cuda:0'), covar=tensor([0.0585, 0.0350, 0.0875, 0.0252, 0.0448, 0.0234, 0.0237, 0.0179], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0278, 0.0334, 0.0288, 0.0287, 0.0218, 0.0266, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 10:07:14,029 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3592, 4.6088, 4.1909, 4.6778, 4.1714, 4.3034, 4.6920, 4.5400], device='cuda:0'), covar=tensor([0.0662, 0.0312, 0.0884, 0.0317, 0.0468, 0.0422, 0.0303, 0.0215], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0277, 0.0333, 0.0287, 0.0286, 0.0217, 0.0265, 0.0248], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 10:07:37,131 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.130e+02 2.657e+02 3.171e+02 6.492e+02, threshold=5.314e+02, percent-clipped=1.0 2023-03-08 10:07:57,990 INFO [train2.py:809] (0/4) Epoch 14, batch 1750, loss[ctc_loss=0.1063, att_loss=0.2608, loss=0.2299, over 17030.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.00816, over 51.00 utterances.], tot_loss[ctc_loss=0.08989, att_loss=0.244, loss=0.2132, over 3275095.83 frames. utt_duration=1262 frames, utt_pad_proportion=0.05175, over 10396.11 utterances.], batch size: 51, lr: 7.77e-03, grad_scale: 8.0 2023-03-08 10:08:06,904 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.08 vs. limit=5.0 2023-03-08 10:08:20,434 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5235, 2.8639, 3.3685, 4.4961, 3.9962, 4.0250, 2.9014, 2.2411], device='cuda:0'), covar=tensor([0.0665, 0.2097, 0.1002, 0.0455, 0.0719, 0.0424, 0.1568, 0.2249], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0207, 0.0187, 0.0197, 0.0200, 0.0159, 0.0194, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 10:08:56,677 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2312, 5.3307, 5.0989, 2.8327, 5.0459, 4.8993, 4.2895, 2.9616], device='cuda:0'), covar=tensor([0.0147, 0.0075, 0.0271, 0.1101, 0.0088, 0.0161, 0.0349, 0.1262], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0089, 0.0086, 0.0106, 0.0075, 0.0099, 0.0095, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 10:09:17,125 INFO [train2.py:809] (0/4) Epoch 14, batch 1800, loss[ctc_loss=0.08462, att_loss=0.227, loss=0.1985, over 16121.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006006, over 42.00 utterances.], tot_loss[ctc_loss=0.09036, att_loss=0.2441, loss=0.2133, over 3279921.61 frames. utt_duration=1265 frames, utt_pad_proportion=0.04924, over 10385.78 utterances.], batch size: 42, lr: 7.77e-03, grad_scale: 8.0 2023-03-08 10:09:28,919 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53597.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:09:51,266 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0357, 5.0413, 4.9180, 2.7928, 4.7744, 4.7173, 4.1842, 2.6755], device='cuda:0'), covar=tensor([0.0134, 0.0102, 0.0235, 0.1081, 0.0097, 0.0173, 0.0330, 0.1328], device='cuda:0'), in_proj_covar=tensor([0.0064, 0.0089, 0.0085, 0.0105, 0.0074, 0.0099, 0.0094, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 10:10:17,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.192e+02 2.789e+02 3.495e+02 6.708e+02, threshold=5.578e+02, percent-clipped=5.0 2023-03-08 10:10:39,125 INFO [train2.py:809] (0/4) Epoch 14, batch 1850, loss[ctc_loss=0.1039, att_loss=0.2574, loss=0.2267, over 17412.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03068, over 63.00 utterances.], tot_loss[ctc_loss=0.09122, att_loss=0.2448, loss=0.2141, over 3268498.08 frames. utt_duration=1230 frames, utt_pad_proportion=0.06284, over 10644.64 utterances.], batch size: 63, lr: 7.77e-03, grad_scale: 8.0 2023-03-08 10:11:12,777 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53660.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:11:20,341 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0999, 2.0174, 2.1223, 2.6718, 2.8331, 2.1827, 2.2137, 2.9740], device='cuda:0'), covar=tensor([0.1049, 0.3661, 0.2659, 0.1196, 0.1200, 0.1378, 0.3389, 0.0830], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0093, 0.0095, 0.0079, 0.0084, 0.0076, 0.0099, 0.0068], device='cuda:0'), out_proj_covar=tensor([6.0154e-05, 6.7700e-05, 7.0192e-05, 5.8637e-05, 5.9257e-05, 5.8075e-05, 6.9373e-05, 5.2456e-05], device='cuda:0') 2023-03-08 10:11:29,325 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53671.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:11:42,351 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-03-08 10:11:59,260 INFO [train2.py:809] (0/4) Epoch 14, batch 1900, loss[ctc_loss=0.07006, att_loss=0.2279, loss=0.1963, over 16122.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006572, over 42.00 utterances.], tot_loss[ctc_loss=0.09109, att_loss=0.245, loss=0.2142, over 3275423.69 frames. utt_duration=1238 frames, utt_pad_proportion=0.05869, over 10593.28 utterances.], batch size: 42, lr: 7.76e-03, grad_scale: 8.0 2023-03-08 10:12:35,605 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53712.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:12:41,879 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9885, 4.9254, 5.1094, 2.2817, 1.8826, 2.5954, 2.3603, 3.5883], device='cuda:0'), covar=tensor([0.0935, 0.0477, 0.0275, 0.4018, 0.6976, 0.3288, 0.3262, 0.2262], device='cuda:0'), in_proj_covar=tensor([0.0337, 0.0236, 0.0243, 0.0219, 0.0341, 0.0331, 0.0230, 0.0352], device='cuda:0'), out_proj_covar=tensor([1.4772e-04, 8.7379e-05, 1.0459e-04, 9.5945e-05, 1.4595e-04, 1.3194e-04, 9.1465e-05, 1.4591e-04], device='cuda:0') 2023-03-08 10:12:49,339 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0995, 1.9314, 2.3141, 2.7002, 2.8936, 2.4303, 2.3090, 3.0753], device='cuda:0'), covar=tensor([0.1695, 0.4633, 0.3045, 0.1753, 0.1897, 0.1851, 0.3751, 0.1183], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0095, 0.0096, 0.0081, 0.0085, 0.0077, 0.0101, 0.0068], device='cuda:0'), out_proj_covar=tensor([6.1355e-05, 6.8913e-05, 7.1045e-05, 5.9647e-05, 6.0351e-05, 5.8875e-05, 7.0630e-05, 5.3188e-05], device='cuda:0') 2023-03-08 10:12:52,332 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=53723.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 10:12:58,261 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.175e+02 2.093e+02 2.485e+02 3.000e+02 6.734e+02, threshold=4.970e+02, percent-clipped=1.0 2023-03-08 10:13:18,790 INFO [train2.py:809] (0/4) Epoch 14, batch 1950, loss[ctc_loss=0.0916, att_loss=0.2371, loss=0.208, over 14456.00 frames. utt_duration=1809 frames, utt_pad_proportion=0.03894, over 32.00 utterances.], tot_loss[ctc_loss=0.0919, att_loss=0.2459, loss=0.2151, over 3275969.35 frames. utt_duration=1206 frames, utt_pad_proportion=0.06586, over 10878.10 utterances.], batch size: 32, lr: 7.76e-03, grad_scale: 8.0 2023-03-08 10:13:50,777 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53760.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:14:38,396 INFO [train2.py:809] (0/4) Epoch 14, batch 2000, loss[ctc_loss=0.0882, att_loss=0.2322, loss=0.2034, over 16276.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006827, over 43.00 utterances.], tot_loss[ctc_loss=0.092, att_loss=0.2458, loss=0.215, over 3281444.45 frames. utt_duration=1225 frames, utt_pad_proportion=0.05973, over 10727.38 utterances.], batch size: 43, lr: 7.76e-03, grad_scale: 8.0 2023-03-08 10:15:36,776 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.520e+02 2.201e+02 2.632e+02 3.371e+02 1.571e+03, threshold=5.264e+02, percent-clipped=3.0 2023-03-08 10:15:57,724 INFO [train2.py:809] (0/4) Epoch 14, batch 2050, loss[ctc_loss=0.07028, att_loss=0.2226, loss=0.1921, over 15759.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.008253, over 38.00 utterances.], tot_loss[ctc_loss=0.09187, att_loss=0.2458, loss=0.215, over 3285954.21 frames. utt_duration=1237 frames, utt_pad_proportion=0.05507, over 10639.39 utterances.], batch size: 38, lr: 7.75e-03, grad_scale: 8.0 2023-03-08 10:17:18,527 INFO [train2.py:809] (0/4) Epoch 14, batch 2100, loss[ctc_loss=0.06366, att_loss=0.206, loss=0.1775, over 15641.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009241, over 37.00 utterances.], tot_loss[ctc_loss=0.09091, att_loss=0.2453, loss=0.2144, over 3280851.67 frames. utt_duration=1222 frames, utt_pad_proportion=0.05943, over 10754.32 utterances.], batch size: 37, lr: 7.75e-03, grad_scale: 8.0 2023-03-08 10:17:30,149 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53897.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:18:17,164 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.409e+02 2.088e+02 2.471e+02 3.126e+02 7.099e+02, threshold=4.942e+02, percent-clipped=2.0 2023-03-08 10:18:37,840 INFO [train2.py:809] (0/4) Epoch 14, batch 2150, loss[ctc_loss=0.05819, att_loss=0.2185, loss=0.1864, over 15887.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009191, over 39.00 utterances.], tot_loss[ctc_loss=0.09077, att_loss=0.2453, loss=0.2144, over 3282256.13 frames. utt_duration=1239 frames, utt_pad_proportion=0.05496, over 10613.41 utterances.], batch size: 39, lr: 7.75e-03, grad_scale: 8.0 2023-03-08 10:18:45,617 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=53945.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:19:10,018 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53960.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:19:27,841 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=53971.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:19:29,386 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5394, 4.9228, 4.7292, 4.9441, 5.0623, 4.6228, 3.4080, 4.7811], device='cuda:0'), covar=tensor([0.0108, 0.0120, 0.0137, 0.0073, 0.0095, 0.0121, 0.0699, 0.0222], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0077, 0.0095, 0.0059, 0.0064, 0.0075, 0.0095, 0.0096], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 10:19:48,707 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-03-08 10:19:57,013 INFO [train2.py:809] (0/4) Epoch 14, batch 2200, loss[ctc_loss=0.08412, att_loss=0.2468, loss=0.2143, over 16480.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005574, over 46.00 utterances.], tot_loss[ctc_loss=0.09031, att_loss=0.2452, loss=0.2142, over 3289161.44 frames. utt_duration=1259 frames, utt_pad_proportion=0.04802, over 10458.43 utterances.], batch size: 46, lr: 7.74e-03, grad_scale: 8.0 2023-03-08 10:20:13,318 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-54000.pt 2023-03-08 10:20:30,049 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54008.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:20:47,495 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54019.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:20:53,792 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=54023.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 10:20:59,197 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 2.188e+02 2.816e+02 3.562e+02 5.650e+02, threshold=5.631e+02, percent-clipped=6.0 2023-03-08 10:21:20,203 INFO [train2.py:809] (0/4) Epoch 14, batch 2250, loss[ctc_loss=0.0657, att_loss=0.2186, loss=0.188, over 16164.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006554, over 41.00 utterances.], tot_loss[ctc_loss=0.09068, att_loss=0.2449, loss=0.2141, over 3282941.70 frames. utt_duration=1231 frames, utt_pad_proportion=0.0556, over 10684.41 utterances.], batch size: 41, lr: 7.74e-03, grad_scale: 8.0 2023-03-08 10:22:09,526 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=54071.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:22:39,616 INFO [train2.py:809] (0/4) Epoch 14, batch 2300, loss[ctc_loss=0.0767, att_loss=0.2358, loss=0.204, over 16273.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007706, over 43.00 utterances.], tot_loss[ctc_loss=0.09085, att_loss=0.245, loss=0.2142, over 3276982.15 frames. utt_duration=1218 frames, utt_pad_proportion=0.06202, over 10779.28 utterances.], batch size: 43, lr: 7.73e-03, grad_scale: 8.0 2023-03-08 10:22:53,266 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-03-08 10:23:31,082 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7649, 5.1464, 4.9262, 5.0599, 5.2826, 4.7900, 3.6977, 5.0680], device='cuda:0'), covar=tensor([0.0086, 0.0095, 0.0102, 0.0076, 0.0054, 0.0097, 0.0603, 0.0128], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0077, 0.0094, 0.0058, 0.0064, 0.0075, 0.0095, 0.0095], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 10:23:36,761 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.455e+02 2.251e+02 2.657e+02 3.358e+02 7.808e+02, threshold=5.315e+02, percent-clipped=1.0 2023-03-08 10:23:57,482 INFO [train2.py:809] (0/4) Epoch 14, batch 2350, loss[ctc_loss=0.1057, att_loss=0.2655, loss=0.2335, over 17318.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02261, over 59.00 utterances.], tot_loss[ctc_loss=0.09159, att_loss=0.2458, loss=0.215, over 3290353.48 frames. utt_duration=1227 frames, utt_pad_proportion=0.05532, over 10741.03 utterances.], batch size: 59, lr: 7.73e-03, grad_scale: 8.0 2023-03-08 10:25:16,437 INFO [train2.py:809] (0/4) Epoch 14, batch 2400, loss[ctc_loss=0.06683, att_loss=0.2286, loss=0.1963, over 17430.00 frames. utt_duration=884.2 frames, utt_pad_proportion=0.07509, over 79.00 utterances.], tot_loss[ctc_loss=0.09085, att_loss=0.2445, loss=0.2138, over 3283374.88 frames. utt_duration=1236 frames, utt_pad_proportion=0.05392, over 10637.35 utterances.], batch size: 79, lr: 7.73e-03, grad_scale: 8.0 2023-03-08 10:25:39,408 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1968, 3.9446, 3.3334, 3.5124, 4.1531, 3.8263, 2.9933, 4.5060], device='cuda:0'), covar=tensor([0.0986, 0.0529, 0.0997, 0.0646, 0.0660, 0.0631, 0.0867, 0.0413], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0195, 0.0211, 0.0184, 0.0250, 0.0220, 0.0189, 0.0267], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 10:26:05,189 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1767, 5.2110, 5.0960, 2.0713, 2.0003, 2.6300, 2.6358, 3.8973], device='cuda:0'), covar=tensor([0.0687, 0.0222, 0.0207, 0.4795, 0.6136, 0.2926, 0.2492, 0.1785], device='cuda:0'), in_proj_covar=tensor([0.0343, 0.0240, 0.0246, 0.0222, 0.0346, 0.0337, 0.0233, 0.0356], device='cuda:0'), out_proj_covar=tensor([1.4981e-04, 8.9021e-05, 1.0593e-04, 9.7525e-05, 1.4776e-04, 1.3387e-04, 9.2805e-05, 1.4779e-04], device='cuda:0') 2023-03-08 10:26:15,330 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.443e+02 2.183e+02 2.531e+02 3.194e+02 7.343e+02, threshold=5.062e+02, percent-clipped=2.0 2023-03-08 10:26:17,218 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3281, 5.2291, 5.1582, 3.0288, 5.1464, 4.8452, 4.3674, 3.0955], device='cuda:0'), covar=tensor([0.0091, 0.0063, 0.0221, 0.0985, 0.0065, 0.0149, 0.0316, 0.1113], device='cuda:0'), in_proj_covar=tensor([0.0066, 0.0090, 0.0086, 0.0107, 0.0076, 0.0100, 0.0096, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 10:26:17,265 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8026, 3.2744, 3.8068, 3.1725, 3.6208, 4.8122, 4.5905, 3.2948], device='cuda:0'), covar=tensor([0.0328, 0.1483, 0.0976, 0.1294, 0.0943, 0.0698, 0.0511, 0.1299], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0230, 0.0251, 0.0204, 0.0242, 0.0326, 0.0232, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 10:26:27,170 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0611, 5.2084, 5.1206, 2.5899, 2.0546, 2.9921, 2.9075, 3.8278], device='cuda:0'), covar=tensor([0.0733, 0.0245, 0.0215, 0.4112, 0.6008, 0.2460, 0.2585, 0.1827], device='cuda:0'), in_proj_covar=tensor([0.0340, 0.0239, 0.0245, 0.0221, 0.0344, 0.0334, 0.0232, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.4883e-04, 8.8617e-05, 1.0520e-04, 9.7102e-05, 1.4681e-04, 1.3298e-04, 9.2253e-05, 1.4696e-04], device='cuda:0') 2023-03-08 10:26:35,973 INFO [train2.py:809] (0/4) Epoch 14, batch 2450, loss[ctc_loss=0.08546, att_loss=0.2498, loss=0.2169, over 16636.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004841, over 47.00 utterances.], tot_loss[ctc_loss=0.09034, att_loss=0.2441, loss=0.2134, over 3278807.16 frames. utt_duration=1266 frames, utt_pad_proportion=0.04737, over 10374.56 utterances.], batch size: 47, lr: 7.72e-03, grad_scale: 8.0 2023-03-08 10:27:54,980 INFO [train2.py:809] (0/4) Epoch 14, batch 2500, loss[ctc_loss=0.07992, att_loss=0.2397, loss=0.2078, over 16622.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005543, over 47.00 utterances.], tot_loss[ctc_loss=0.09091, att_loss=0.2448, loss=0.214, over 3280716.03 frames. utt_duration=1242 frames, utt_pad_proportion=0.05172, over 10577.14 utterances.], batch size: 47, lr: 7.72e-03, grad_scale: 8.0 2023-03-08 10:28:46,334 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3663, 4.5640, 4.5238, 4.6081, 4.6475, 4.6380, 4.3677, 4.2230], device='cuda:0'), covar=tensor([0.0939, 0.0698, 0.0341, 0.0419, 0.0301, 0.0334, 0.0372, 0.0369], device='cuda:0'), in_proj_covar=tensor([0.0480, 0.0320, 0.0284, 0.0312, 0.0369, 0.0387, 0.0314, 0.0354], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 10:28:53,990 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.435e+02 2.776e+02 3.474e+02 9.434e+02, threshold=5.553e+02, percent-clipped=7.0 2023-03-08 10:29:14,825 INFO [train2.py:809] (0/4) Epoch 14, batch 2550, loss[ctc_loss=0.0739, att_loss=0.2289, loss=0.1979, over 16275.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.006888, over 43.00 utterances.], tot_loss[ctc_loss=0.0906, att_loss=0.2445, loss=0.2137, over 3272448.04 frames. utt_duration=1240 frames, utt_pad_proportion=0.05404, over 10567.57 utterances.], batch size: 43, lr: 7.72e-03, grad_scale: 8.0 2023-03-08 10:29:34,825 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1764, 5.2185, 5.0507, 2.4589, 2.0509, 3.0006, 2.8029, 3.8034], device='cuda:0'), covar=tensor([0.0715, 0.0398, 0.0298, 0.4831, 0.6045, 0.2554, 0.2771, 0.2062], device='cuda:0'), in_proj_covar=tensor([0.0342, 0.0240, 0.0245, 0.0222, 0.0347, 0.0335, 0.0232, 0.0356], device='cuda:0'), out_proj_covar=tensor([1.4974e-04, 8.9012e-05, 1.0512e-04, 9.7213e-05, 1.4789e-04, 1.3327e-04, 9.2236e-05, 1.4783e-04], device='cuda:0') 2023-03-08 10:30:31,081 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9388, 5.2264, 5.5191, 5.3256, 5.3729, 5.8829, 5.1865, 5.9843], device='cuda:0'), covar=tensor([0.0815, 0.0704, 0.0762, 0.1255, 0.1936, 0.0998, 0.0671, 0.0716], device='cuda:0'), in_proj_covar=tensor([0.0779, 0.0456, 0.0536, 0.0600, 0.0791, 0.0545, 0.0440, 0.0529], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 10:30:37,349 INFO [train2.py:809] (0/4) Epoch 14, batch 2600, loss[ctc_loss=0.09653, att_loss=0.2397, loss=0.2111, over 15939.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.00799, over 41.00 utterances.], tot_loss[ctc_loss=0.09057, att_loss=0.2443, loss=0.2136, over 3267691.64 frames. utt_duration=1251 frames, utt_pad_proportion=0.05366, over 10457.59 utterances.], batch size: 41, lr: 7.71e-03, grad_scale: 8.0 2023-03-08 10:31:39,668 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.230e+02 2.611e+02 3.212e+02 5.306e+02, threshold=5.223e+02, percent-clipped=0.0 2023-03-08 10:32:01,194 INFO [train2.py:809] (0/4) Epoch 14, batch 2650, loss[ctc_loss=0.1164, att_loss=0.2643, loss=0.2347, over 16623.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005454, over 47.00 utterances.], tot_loss[ctc_loss=0.09136, att_loss=0.2447, loss=0.2141, over 3268281.44 frames. utt_duration=1242 frames, utt_pad_proportion=0.0569, over 10539.72 utterances.], batch size: 47, lr: 7.71e-03, grad_scale: 8.0 2023-03-08 10:33:18,525 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.65 vs. limit=5.0 2023-03-08 10:33:23,999 INFO [train2.py:809] (0/4) Epoch 14, batch 2700, loss[ctc_loss=0.08433, att_loss=0.2328, loss=0.2031, over 16343.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005411, over 45.00 utterances.], tot_loss[ctc_loss=0.09169, att_loss=0.2452, loss=0.2145, over 3277046.99 frames. utt_duration=1237 frames, utt_pad_proportion=0.05489, over 10612.87 utterances.], batch size: 45, lr: 7.71e-03, grad_scale: 8.0 2023-03-08 10:33:51,065 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2471, 2.7639, 2.8715, 4.2821, 3.8179, 3.9083, 2.9007, 2.0042], device='cuda:0'), covar=tensor([0.0795, 0.2266, 0.1317, 0.0619, 0.0897, 0.0464, 0.1626, 0.2584], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0210, 0.0188, 0.0198, 0.0199, 0.0163, 0.0197, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 10:34:25,552 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.188e+02 2.730e+02 3.256e+02 6.934e+02, threshold=5.460e+02, percent-clipped=5.0 2023-03-08 10:34:27,666 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3066, 5.2743, 5.0835, 2.8116, 2.2740, 2.9842, 3.1666, 3.9811], device='cuda:0'), covar=tensor([0.0671, 0.0293, 0.0315, 0.4249, 0.5687, 0.2716, 0.2429, 0.1834], device='cuda:0'), in_proj_covar=tensor([0.0340, 0.0237, 0.0245, 0.0222, 0.0343, 0.0335, 0.0232, 0.0353], device='cuda:0'), out_proj_covar=tensor([1.4863e-04, 8.7785e-05, 1.0535e-04, 9.7292e-05, 1.4660e-04, 1.3300e-04, 9.2151e-05, 1.4663e-04], device='cuda:0') 2023-03-08 10:34:47,228 INFO [train2.py:809] (0/4) Epoch 14, batch 2750, loss[ctc_loss=0.08265, att_loss=0.2095, loss=0.1841, over 15779.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.008039, over 38.00 utterances.], tot_loss[ctc_loss=0.0925, att_loss=0.2456, loss=0.215, over 3278163.79 frames. utt_duration=1222 frames, utt_pad_proportion=0.05883, over 10740.03 utterances.], batch size: 38, lr: 7.70e-03, grad_scale: 8.0 2023-03-08 10:35:04,721 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0055, 3.8509, 3.2453, 3.5229, 4.0798, 3.8139, 2.9697, 4.4221], device='cuda:0'), covar=tensor([0.1144, 0.0485, 0.1075, 0.0732, 0.0663, 0.0665, 0.0904, 0.0561], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0197, 0.0213, 0.0187, 0.0253, 0.0223, 0.0191, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 10:35:30,802 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8915, 5.2519, 4.7046, 5.3505, 4.6479, 5.0019, 5.4272, 5.1849], device='cuda:0'), covar=tensor([0.0632, 0.0306, 0.0913, 0.0257, 0.0457, 0.0223, 0.0188, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0356, 0.0278, 0.0333, 0.0288, 0.0284, 0.0217, 0.0268, 0.0251], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0005, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 10:36:10,902 INFO [train2.py:809] (0/4) Epoch 14, batch 2800, loss[ctc_loss=0.06718, att_loss=0.2186, loss=0.1883, over 14597.00 frames. utt_duration=1826 frames, utt_pad_proportion=0.03935, over 32.00 utterances.], tot_loss[ctc_loss=0.09101, att_loss=0.2439, loss=0.2133, over 3271704.21 frames. utt_duration=1262 frames, utt_pad_proportion=0.05152, over 10384.51 utterances.], batch size: 32, lr: 7.70e-03, grad_scale: 8.0 2023-03-08 10:37:12,836 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.211e+02 2.653e+02 3.453e+02 1.073e+03, threshold=5.307e+02, percent-clipped=4.0 2023-03-08 10:37:34,801 INFO [train2.py:809] (0/4) Epoch 14, batch 2850, loss[ctc_loss=0.08304, att_loss=0.2439, loss=0.2117, over 17055.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008926, over 52.00 utterances.], tot_loss[ctc_loss=0.09118, att_loss=0.2442, loss=0.2136, over 3273754.36 frames. utt_duration=1260 frames, utt_pad_proportion=0.05206, over 10409.10 utterances.], batch size: 52, lr: 7.70e-03, grad_scale: 8.0 2023-03-08 10:37:48,691 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2219, 5.1375, 5.0099, 2.3953, 1.9806, 2.8673, 2.5376, 3.9463], device='cuda:0'), covar=tensor([0.0607, 0.0246, 0.0246, 0.4589, 0.5721, 0.2602, 0.3109, 0.1693], device='cuda:0'), in_proj_covar=tensor([0.0337, 0.0236, 0.0242, 0.0219, 0.0340, 0.0331, 0.0230, 0.0349], device='cuda:0'), out_proj_covar=tensor([1.4711e-04, 8.7027e-05, 1.0393e-04, 9.5912e-05, 1.4520e-04, 1.3155e-04, 9.1472e-05, 1.4513e-04], device='cuda:0') 2023-03-08 10:38:05,402 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8796, 4.8912, 4.6575, 4.6728, 5.3705, 4.8187, 4.7967, 2.4899], device='cuda:0'), covar=tensor([0.0165, 0.0214, 0.0222, 0.0277, 0.0850, 0.0186, 0.0244, 0.1973], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0145, 0.0151, 0.0161, 0.0347, 0.0131, 0.0137, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 10:38:13,020 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4282, 2.8588, 3.6617, 2.8895, 3.4934, 4.6079, 4.4487, 3.0509], device='cuda:0'), covar=tensor([0.0378, 0.1844, 0.1076, 0.1422, 0.0990, 0.0720, 0.0459, 0.1498], device='cuda:0'), in_proj_covar=tensor([0.0234, 0.0234, 0.0254, 0.0206, 0.0247, 0.0329, 0.0234, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 10:38:14,622 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54664.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 10:38:57,449 INFO [train2.py:809] (0/4) Epoch 14, batch 2900, loss[ctc_loss=0.09152, att_loss=0.2521, loss=0.22, over 16327.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006399, over 45.00 utterances.], tot_loss[ctc_loss=0.09062, att_loss=0.2442, loss=0.2135, over 3278056.98 frames. utt_duration=1253 frames, utt_pad_proportion=0.0516, over 10481.03 utterances.], batch size: 45, lr: 7.69e-03, grad_scale: 8.0 2023-03-08 10:39:55,558 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54725.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 10:39:58,422 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.074e+02 2.453e+02 3.116e+02 5.349e+02, threshold=4.905e+02, percent-clipped=1.0 2023-03-08 10:40:19,749 INFO [train2.py:809] (0/4) Epoch 14, batch 2950, loss[ctc_loss=0.08307, att_loss=0.2422, loss=0.2104, over 16104.00 frames. utt_duration=1535 frames, utt_pad_proportion=0.007618, over 42.00 utterances.], tot_loss[ctc_loss=0.09056, att_loss=0.2444, loss=0.2136, over 3275049.26 frames. utt_duration=1249 frames, utt_pad_proportion=0.05502, over 10501.73 utterances.], batch size: 42, lr: 7.69e-03, grad_scale: 8.0 2023-03-08 10:41:41,645 INFO [train2.py:809] (0/4) Epoch 14, batch 3000, loss[ctc_loss=0.1135, att_loss=0.2556, loss=0.2272, over 16471.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006373, over 46.00 utterances.], tot_loss[ctc_loss=0.09082, att_loss=0.2447, loss=0.2139, over 3273210.54 frames. utt_duration=1243 frames, utt_pad_proportion=0.05633, over 10543.28 utterances.], batch size: 46, lr: 7.69e-03, grad_scale: 8.0 2023-03-08 10:41:41,648 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 10:41:56,406 INFO [train2.py:843] (0/4) Epoch 14, validation: ctc_loss=0.0441, att_loss=0.2368, loss=0.1983, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 10:41:56,407 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 10:42:55,796 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.163e+02 2.238e+02 2.720e+02 3.334e+02 7.498e+02, threshold=5.441e+02, percent-clipped=5.0 2023-03-08 10:43:16,796 INFO [train2.py:809] (0/4) Epoch 14, batch 3050, loss[ctc_loss=0.1371, att_loss=0.2782, loss=0.25, over 14078.00 frames. utt_duration=387.2 frames, utt_pad_proportion=0.3231, over 146.00 utterances.], tot_loss[ctc_loss=0.08989, att_loss=0.2434, loss=0.2127, over 3271161.24 frames. utt_duration=1260 frames, utt_pad_proportion=0.05331, over 10395.17 utterances.], batch size: 146, lr: 7.68e-03, grad_scale: 16.0 2023-03-08 10:43:52,880 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=54862.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:44:29,849 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2616, 4.9922, 5.0954, 5.1409, 4.9351, 5.0910, 4.9368, 4.6056], device='cuda:0'), covar=tensor([0.2099, 0.0867, 0.0412, 0.0556, 0.0923, 0.0478, 0.0485, 0.0511], device='cuda:0'), in_proj_covar=tensor([0.0483, 0.0320, 0.0285, 0.0317, 0.0375, 0.0393, 0.0318, 0.0358], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 10:44:38,673 INFO [train2.py:809] (0/4) Epoch 14, batch 3100, loss[ctc_loss=0.07692, att_loss=0.2242, loss=0.1947, over 15784.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.007866, over 38.00 utterances.], tot_loss[ctc_loss=0.09041, att_loss=0.2438, loss=0.2131, over 3278091.03 frames. utt_duration=1279 frames, utt_pad_proportion=0.04685, over 10264.64 utterances.], batch size: 38, lr: 7.68e-03, grad_scale: 16.0 2023-03-08 10:45:32,168 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=54923.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 10:45:38,025 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 2.291e+02 2.720e+02 3.507e+02 7.867e+02, threshold=5.439e+02, percent-clipped=2.0 2023-03-08 10:45:59,420 INFO [train2.py:809] (0/4) Epoch 14, batch 3150, loss[ctc_loss=0.07107, att_loss=0.2245, loss=0.1939, over 15896.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008864, over 39.00 utterances.], tot_loss[ctc_loss=0.08966, att_loss=0.2434, loss=0.2126, over 3277324.58 frames. utt_duration=1264 frames, utt_pad_proportion=0.05015, over 10379.81 utterances.], batch size: 39, lr: 7.68e-03, grad_scale: 16.0 2023-03-08 10:46:07,479 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4872, 3.0718, 3.3895, 4.4827, 3.9542, 3.9178, 3.0235, 1.9802], device='cuda:0'), covar=tensor([0.0650, 0.1912, 0.1001, 0.0488, 0.0754, 0.0469, 0.1366, 0.2530], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0208, 0.0187, 0.0197, 0.0199, 0.0162, 0.0195, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 10:46:26,540 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5472, 4.5718, 4.4719, 4.4467, 5.0234, 4.6610, 4.5338, 2.4602], device='cuda:0'), covar=tensor([0.0185, 0.0260, 0.0276, 0.0254, 0.1184, 0.0172, 0.0291, 0.2152], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0144, 0.0151, 0.0163, 0.0347, 0.0130, 0.0137, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 10:46:32,914 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6982, 2.9805, 3.5476, 4.6014, 4.0134, 3.9878, 3.0896, 2.1171], device='cuda:0'), covar=tensor([0.0504, 0.1880, 0.0853, 0.0438, 0.0639, 0.0429, 0.1288, 0.2256], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0207, 0.0185, 0.0196, 0.0198, 0.0161, 0.0194, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 10:47:20,536 INFO [train2.py:809] (0/4) Epoch 14, batch 3200, loss[ctc_loss=0.09462, att_loss=0.2486, loss=0.2178, over 16766.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006718, over 48.00 utterances.], tot_loss[ctc_loss=0.09008, att_loss=0.2438, loss=0.213, over 3281803.77 frames. utt_duration=1272 frames, utt_pad_proportion=0.04707, over 10329.54 utterances.], batch size: 48, lr: 7.67e-03, grad_scale: 16.0 2023-03-08 10:47:25,855 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4930, 2.9043, 2.9322, 4.4122, 3.9689, 3.9832, 3.0803, 2.1364], device='cuda:0'), covar=tensor([0.0579, 0.1976, 0.1403, 0.0520, 0.0736, 0.0394, 0.1250, 0.2197], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0209, 0.0187, 0.0197, 0.0200, 0.0162, 0.0195, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 10:47:46,341 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 10:48:05,138 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1396, 5.4137, 5.3395, 5.4085, 5.4710, 5.3979, 5.1615, 4.9233], device='cuda:0'), covar=tensor([0.0899, 0.0471, 0.0243, 0.0473, 0.0267, 0.0280, 0.0357, 0.0351], device='cuda:0'), in_proj_covar=tensor([0.0495, 0.0327, 0.0291, 0.0323, 0.0381, 0.0400, 0.0325, 0.0364], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 10:48:12,466 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55020.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 10:48:23,969 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.213e+02 2.608e+02 3.360e+02 8.145e+02, threshold=5.216e+02, percent-clipped=2.0 2023-03-08 10:48:46,136 INFO [train2.py:809] (0/4) Epoch 14, batch 3250, loss[ctc_loss=0.09739, att_loss=0.2556, loss=0.224, over 16479.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006459, over 46.00 utterances.], tot_loss[ctc_loss=0.08994, att_loss=0.2436, loss=0.2129, over 3276892.85 frames. utt_duration=1271 frames, utt_pad_proportion=0.04919, over 10325.61 utterances.], batch size: 46, lr: 7.67e-03, grad_scale: 16.0 2023-03-08 10:48:51,134 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-08 10:48:51,829 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0034, 5.2587, 5.5668, 5.4224, 5.4057, 5.9116, 5.2426, 6.0445], device='cuda:0'), covar=tensor([0.0580, 0.0708, 0.0669, 0.1051, 0.1717, 0.0879, 0.0594, 0.0507], device='cuda:0'), in_proj_covar=tensor([0.0770, 0.0455, 0.0538, 0.0598, 0.0791, 0.0543, 0.0438, 0.0534], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 10:49:01,104 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-03-08 10:50:12,431 INFO [train2.py:809] (0/4) Epoch 14, batch 3300, loss[ctc_loss=0.121, att_loss=0.2642, loss=0.2356, over 17035.00 frames. utt_duration=1338 frames, utt_pad_proportion=0.007709, over 51.00 utterances.], tot_loss[ctc_loss=0.09082, att_loss=0.2444, loss=0.2137, over 3271478.77 frames. utt_duration=1239 frames, utt_pad_proportion=0.05656, over 10571.59 utterances.], batch size: 51, lr: 7.66e-03, grad_scale: 16.0 2023-03-08 10:50:56,278 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0042, 5.3232, 5.5659, 5.4077, 5.4583, 5.9684, 5.2380, 6.0810], device='cuda:0'), covar=tensor([0.0719, 0.0690, 0.0742, 0.1154, 0.1794, 0.0838, 0.0613, 0.0547], device='cuda:0'), in_proj_covar=tensor([0.0776, 0.0460, 0.0540, 0.0602, 0.0793, 0.0546, 0.0442, 0.0536], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 10:51:15,768 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.495e+02 2.260e+02 2.683e+02 3.437e+02 9.607e+02, threshold=5.365e+02, percent-clipped=9.0 2023-03-08 10:51:31,911 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.96 vs. limit=5.0 2023-03-08 10:51:37,890 INFO [train2.py:809] (0/4) Epoch 14, batch 3350, loss[ctc_loss=0.1048, att_loss=0.2521, loss=0.2226, over 16466.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.007302, over 46.00 utterances.], tot_loss[ctc_loss=0.0912, att_loss=0.2454, loss=0.2146, over 3278036.29 frames. utt_duration=1224 frames, utt_pad_proportion=0.05892, over 10729.50 utterances.], batch size: 46, lr: 7.66e-03, grad_scale: 16.0 2023-03-08 10:51:50,869 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55148.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:52:36,793 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-03-08 10:53:01,319 INFO [train2.py:809] (0/4) Epoch 14, batch 3400, loss[ctc_loss=0.06361, att_loss=0.2278, loss=0.1949, over 16123.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006526, over 42.00 utterances.], tot_loss[ctc_loss=0.09045, att_loss=0.2445, loss=0.2137, over 3275433.42 frames. utt_duration=1247 frames, utt_pad_proportion=0.05353, over 10523.53 utterances.], batch size: 42, lr: 7.66e-03, grad_scale: 16.0 2023-03-08 10:53:34,782 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55209.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:53:38,105 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55211.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:53:50,356 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55218.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 10:53:59,826 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-03-08 10:54:05,573 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.493e+02 2.249e+02 2.663e+02 3.402e+02 7.529e+02, threshold=5.327e+02, percent-clipped=2.0 2023-03-08 10:54:27,347 INFO [train2.py:809] (0/4) Epoch 14, batch 3450, loss[ctc_loss=0.07208, att_loss=0.2373, loss=0.2043, over 16682.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006145, over 46.00 utterances.], tot_loss[ctc_loss=0.09125, att_loss=0.2447, loss=0.214, over 3269832.23 frames. utt_duration=1237 frames, utt_pad_proportion=0.05766, over 10587.42 utterances.], batch size: 46, lr: 7.65e-03, grad_scale: 16.0 2023-03-08 10:55:20,420 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1198, 4.4320, 4.2290, 4.5534, 2.5707, 4.7513, 2.8319, 1.7455], device='cuda:0'), covar=tensor([0.0344, 0.0163, 0.0770, 0.0170, 0.1828, 0.0141, 0.1464, 0.1764], device='cuda:0'), in_proj_covar=tensor([0.0158, 0.0132, 0.0249, 0.0121, 0.0217, 0.0114, 0.0223, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 10:55:22,013 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55272.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 10:55:52,458 INFO [train2.py:809] (0/4) Epoch 14, batch 3500, loss[ctc_loss=0.08801, att_loss=0.249, loss=0.2168, over 17151.00 frames. utt_duration=869.9 frames, utt_pad_proportion=0.08908, over 79.00 utterances.], tot_loss[ctc_loss=0.09106, att_loss=0.2439, loss=0.2133, over 3269507.71 frames. utt_duration=1242 frames, utt_pad_proportion=0.05718, over 10546.26 utterances.], batch size: 79, lr: 7.65e-03, grad_scale: 8.0 2023-03-08 10:56:43,458 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55320.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 10:56:56,801 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.444e+02 2.024e+02 2.386e+02 2.873e+02 9.037e+02, threshold=4.771e+02, percent-clipped=2.0 2023-03-08 10:57:17,386 INFO [train2.py:809] (0/4) Epoch 14, batch 3550, loss[ctc_loss=0.08375, att_loss=0.2226, loss=0.1948, over 15363.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.0117, over 35.00 utterances.], tot_loss[ctc_loss=0.09131, att_loss=0.2447, loss=0.214, over 3280463.52 frames. utt_duration=1264 frames, utt_pad_proportion=0.04894, over 10390.18 utterances.], batch size: 35, lr: 7.65e-03, grad_scale: 8.0 2023-03-08 10:58:05,152 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55368.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 10:58:21,701 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-03-08 10:58:41,902 INFO [train2.py:809] (0/4) Epoch 14, batch 3600, loss[ctc_loss=0.09655, att_loss=0.2438, loss=0.2144, over 16285.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006945, over 43.00 utterances.], tot_loss[ctc_loss=0.09011, att_loss=0.2438, loss=0.2131, over 3273162.06 frames. utt_duration=1266 frames, utt_pad_proportion=0.05039, over 10357.37 utterances.], batch size: 43, lr: 7.64e-03, grad_scale: 8.0 2023-03-08 10:59:40,607 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5124, 4.6088, 4.5399, 4.4852, 5.0394, 4.6077, 4.6007, 2.0770], device='cuda:0'), covar=tensor([0.0197, 0.0248, 0.0270, 0.0261, 0.0910, 0.0198, 0.0265, 0.2189], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0146, 0.0155, 0.0165, 0.0353, 0.0133, 0.0139, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 10:59:46,533 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.040e+02 2.412e+02 2.968e+02 1.250e+03, threshold=4.823e+02, percent-clipped=3.0 2023-03-08 11:00:07,018 INFO [train2.py:809] (0/4) Epoch 14, batch 3650, loss[ctc_loss=0.09268, att_loss=0.2481, loss=0.217, over 16939.00 frames. utt_duration=686 frames, utt_pad_proportion=0.1349, over 99.00 utterances.], tot_loss[ctc_loss=0.08997, att_loss=0.2441, loss=0.2133, over 3276113.26 frames. utt_duration=1251 frames, utt_pad_proportion=0.05212, over 10487.52 utterances.], batch size: 99, lr: 7.64e-03, grad_scale: 8.0 2023-03-08 11:00:49,526 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3683, 4.8326, 4.6520, 4.7761, 4.9588, 4.5280, 3.3196, 4.7538], device='cuda:0'), covar=tensor([0.0109, 0.0096, 0.0121, 0.0079, 0.0089, 0.0101, 0.0684, 0.0193], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0077, 0.0095, 0.0058, 0.0065, 0.0076, 0.0095, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:01:08,311 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4032, 2.7895, 3.5865, 4.4350, 3.7964, 4.0015, 2.7312, 1.8980], device='cuda:0'), covar=tensor([0.0643, 0.2197, 0.0794, 0.0559, 0.0887, 0.0416, 0.1682, 0.2485], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0212, 0.0188, 0.0199, 0.0204, 0.0162, 0.0197, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 11:01:29,821 INFO [train2.py:809] (0/4) Epoch 14, batch 3700, loss[ctc_loss=0.07099, att_loss=0.216, loss=0.187, over 14055.00 frames. utt_duration=1815 frames, utt_pad_proportion=0.05083, over 31.00 utterances.], tot_loss[ctc_loss=0.09107, att_loss=0.2445, loss=0.2138, over 3272820.12 frames. utt_duration=1228 frames, utt_pad_proportion=0.06001, over 10675.18 utterances.], batch size: 31, lr: 7.64e-03, grad_scale: 8.0 2023-03-08 11:01:46,830 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6395, 3.3634, 2.7876, 3.1544, 3.5316, 3.3092, 2.3784, 3.5719], device='cuda:0'), covar=tensor([0.0926, 0.0412, 0.0937, 0.0579, 0.0592, 0.0610, 0.0972, 0.0459], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0197, 0.0210, 0.0187, 0.0253, 0.0224, 0.0191, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:01:53,174 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55504.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:02:16,921 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55518.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 11:02:33,336 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.607e+02 2.162e+02 2.668e+02 3.280e+02 7.175e+02, threshold=5.335e+02, percent-clipped=8.0 2023-03-08 11:02:53,079 INFO [train2.py:809] (0/4) Epoch 14, batch 3750, loss[ctc_loss=0.1329, att_loss=0.2758, loss=0.2472, over 17408.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03238, over 63.00 utterances.], tot_loss[ctc_loss=0.09151, att_loss=0.2445, loss=0.2139, over 3270920.35 frames. utt_duration=1240 frames, utt_pad_proportion=0.0564, over 10562.27 utterances.], batch size: 63, lr: 7.63e-03, grad_scale: 8.0 2023-03-08 11:02:53,504 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9715, 4.9074, 4.9236, 4.8033, 5.3478, 4.9883, 4.7984, 2.1583], device='cuda:0'), covar=tensor([0.0109, 0.0147, 0.0126, 0.0135, 0.0741, 0.0109, 0.0174, 0.2091], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0147, 0.0155, 0.0166, 0.0356, 0.0134, 0.0140, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:03:36,185 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55566.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:03:37,989 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=55567.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:04:00,079 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6896, 1.8079, 2.2436, 2.1915, 2.4219, 2.8058, 2.2734, 2.9814], device='cuda:0'), covar=tensor([0.1573, 0.4095, 0.2848, 0.1579, 0.1469, 0.0993, 0.2438, 0.0555], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0094, 0.0098, 0.0082, 0.0086, 0.0078, 0.0100, 0.0068], device='cuda:0'), out_proj_covar=tensor([6.2026e-05, 6.8842e-05, 7.2186e-05, 6.0486e-05, 6.0783e-05, 6.0013e-05, 7.0752e-05, 5.3375e-05], device='cuda:0') 2023-03-08 11:04:16,304 INFO [train2.py:809] (0/4) Epoch 14, batch 3800, loss[ctc_loss=0.09114, att_loss=0.242, loss=0.2118, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007112, over 43.00 utterances.], tot_loss[ctc_loss=0.09109, att_loss=0.2444, loss=0.2137, over 3269193.09 frames. utt_duration=1244 frames, utt_pad_proportion=0.05654, over 10525.53 utterances.], batch size: 43, lr: 7.63e-03, grad_scale: 8.0 2023-03-08 11:04:31,472 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7259, 3.0541, 3.7202, 4.7125, 4.1054, 4.1740, 2.9532, 2.2265], device='cuda:0'), covar=tensor([0.0517, 0.1804, 0.0841, 0.0365, 0.0606, 0.0353, 0.1405, 0.2218], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0215, 0.0192, 0.0201, 0.0206, 0.0164, 0.0199, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 11:05:01,989 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1239, 5.0650, 5.0122, 2.3069, 1.9264, 2.6000, 2.6309, 3.7085], device='cuda:0'), covar=tensor([0.0658, 0.0267, 0.0217, 0.4466, 0.6111, 0.3004, 0.2914, 0.1900], device='cuda:0'), in_proj_covar=tensor([0.0343, 0.0240, 0.0245, 0.0221, 0.0346, 0.0339, 0.0233, 0.0356], device='cuda:0'), out_proj_covar=tensor([1.4994e-04, 8.9782e-05, 1.0561e-04, 9.6552e-05, 1.4743e-04, 1.3427e-04, 9.2789e-05, 1.4784e-04], device='cuda:0') 2023-03-08 11:05:05,909 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0746, 4.6226, 4.9461, 2.0405, 2.0249, 2.3775, 2.3157, 3.5205], device='cuda:0'), covar=tensor([0.0831, 0.0412, 0.0265, 0.3970, 0.5938, 0.3393, 0.3024, 0.2193], device='cuda:0'), in_proj_covar=tensor([0.0343, 0.0240, 0.0245, 0.0221, 0.0346, 0.0339, 0.0233, 0.0356], device='cuda:0'), out_proj_covar=tensor([1.4992e-04, 8.9780e-05, 1.0559e-04, 9.6544e-05, 1.4744e-04, 1.3427e-04, 9.2770e-05, 1.4786e-04], device='cuda:0') 2023-03-08 11:05:20,070 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 2.212e+02 2.620e+02 3.323e+02 8.677e+02, threshold=5.239e+02, percent-clipped=3.0 2023-03-08 11:05:40,114 INFO [train2.py:809] (0/4) Epoch 14, batch 3850, loss[ctc_loss=0.07079, att_loss=0.2164, loss=0.1873, over 15650.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008909, over 37.00 utterances.], tot_loss[ctc_loss=0.09099, att_loss=0.2439, loss=0.2134, over 3259443.67 frames. utt_duration=1207 frames, utt_pad_proportion=0.06846, over 10817.65 utterances.], batch size: 37, lr: 7.63e-03, grad_scale: 8.0 2023-03-08 11:06:33,821 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9845, 4.9073, 4.8898, 4.8125, 5.2703, 4.9932, 4.7351, 2.1479], device='cuda:0'), covar=tensor([0.0125, 0.0170, 0.0146, 0.0152, 0.0899, 0.0136, 0.0194, 0.2079], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0149, 0.0155, 0.0167, 0.0360, 0.0135, 0.0140, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:06:49,841 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1530, 5.0663, 4.9932, 2.9734, 4.9121, 4.7496, 4.4248, 2.5577], device='cuda:0'), covar=tensor([0.0105, 0.0091, 0.0209, 0.1009, 0.0087, 0.0170, 0.0291, 0.1527], device='cuda:0'), in_proj_covar=tensor([0.0065, 0.0090, 0.0086, 0.0105, 0.0075, 0.0101, 0.0095, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 11:07:00,900 INFO [train2.py:809] (0/4) Epoch 14, batch 3900, loss[ctc_loss=0.08771, att_loss=0.2283, loss=0.2002, over 15502.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008562, over 36.00 utterances.], tot_loss[ctc_loss=0.09026, att_loss=0.2443, loss=0.2135, over 3268957.87 frames. utt_duration=1227 frames, utt_pad_proportion=0.06172, over 10672.67 utterances.], batch size: 36, lr: 7.62e-03, grad_scale: 8.0 2023-03-08 11:08:01,631 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.280e+02 2.167e+02 2.705e+02 3.104e+02 1.128e+03, threshold=5.409e+02, percent-clipped=4.0 2023-03-08 11:08:19,382 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 11:08:20,150 INFO [train2.py:809] (0/4) Epoch 14, batch 3950, loss[ctc_loss=0.1078, att_loss=0.2605, loss=0.2299, over 16816.00 frames. utt_duration=687.7 frames, utt_pad_proportion=0.1382, over 98.00 utterances.], tot_loss[ctc_loss=0.08939, att_loss=0.2434, loss=0.2126, over 3260854.66 frames. utt_duration=1239 frames, utt_pad_proportion=0.06119, over 10543.57 utterances.], batch size: 98, lr: 7.62e-03, grad_scale: 8.0 2023-03-08 11:08:32,819 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55748.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:08:36,438 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-03-08 11:09:10,812 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-14.pt 2023-03-08 11:09:39,028 INFO [train2.py:809] (0/4) Epoch 15, batch 0, loss[ctc_loss=0.09596, att_loss=0.259, loss=0.2264, over 17434.00 frames. utt_duration=1012 frames, utt_pad_proportion=0.04509, over 69.00 utterances.], tot_loss[ctc_loss=0.09596, att_loss=0.259, loss=0.2264, over 17434.00 frames. utt_duration=1012 frames, utt_pad_proportion=0.04509, over 69.00 utterances.], batch size: 69, lr: 7.36e-03, grad_scale: 8.0 2023-03-08 11:09:39,031 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 11:09:47,482 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4897, 4.4639, 4.4705, 4.3593, 4.9697, 4.3500, 4.4099, 2.3576], device='cuda:0'), covar=tensor([0.0201, 0.0307, 0.0302, 0.0295, 0.0665, 0.0235, 0.0314, 0.1959], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0148, 0.0154, 0.0166, 0.0355, 0.0135, 0.0139, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:09:51,762 INFO [train2.py:843] (0/4) Epoch 15, validation: ctc_loss=0.04404, att_loss=0.2365, loss=0.198, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 11:09:51,762 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 11:09:55,284 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7205, 3.5397, 2.9790, 3.2064, 3.7345, 3.4140, 2.6632, 3.8592], device='cuda:0'), covar=tensor([0.1076, 0.0456, 0.1032, 0.0726, 0.0669, 0.0741, 0.0964, 0.0573], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0200, 0.0211, 0.0186, 0.0254, 0.0225, 0.0191, 0.0271], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:10:43,361 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55804.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:10:44,866 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7783, 5.0781, 5.3287, 5.2556, 5.2264, 5.7519, 5.0168, 5.8918], device='cuda:0'), covar=tensor([0.0707, 0.0714, 0.0743, 0.1116, 0.1925, 0.0821, 0.0760, 0.0594], device='cuda:0'), in_proj_covar=tensor([0.0780, 0.0464, 0.0547, 0.0606, 0.0802, 0.0556, 0.0453, 0.0539], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:10:46,648 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1440, 4.5169, 4.7076, 4.8837, 3.0653, 4.5186, 2.9770, 2.1465], device='cuda:0'), covar=tensor([0.0313, 0.0201, 0.0552, 0.0117, 0.1444, 0.0178, 0.1260, 0.1423], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0135, 0.0254, 0.0124, 0.0221, 0.0116, 0.0226, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 11:10:51,514 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55809.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:11:05,746 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6692, 3.5970, 3.5496, 3.1474, 3.5843, 3.6597, 3.6298, 2.6588], device='cuda:0'), covar=tensor([0.0973, 0.1470, 0.2407, 0.5004, 0.1202, 0.2867, 0.1281, 0.5902], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0145, 0.0156, 0.0229, 0.0120, 0.0210, 0.0133, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:11:13,891 INFO [train2.py:809] (0/4) Epoch 15, batch 50, loss[ctc_loss=0.07613, att_loss=0.2203, loss=0.1915, over 15629.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009922, over 37.00 utterances.], tot_loss[ctc_loss=0.08976, att_loss=0.2438, loss=0.213, over 742207.92 frames. utt_duration=1228 frames, utt_pad_proportion=0.05792, over 2419.99 utterances.], batch size: 37, lr: 7.35e-03, grad_scale: 8.0 2023-03-08 11:11:21,609 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.330e+02 2.261e+02 2.719e+02 3.409e+02 7.287e+02, threshold=5.439e+02, percent-clipped=2.0 2023-03-08 11:11:42,966 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.90 vs. limit=5.0 2023-03-08 11:11:55,461 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=55848.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:12:01,377 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55852.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:12:25,378 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=55867.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:12:35,389 INFO [train2.py:809] (0/4) Epoch 15, batch 100, loss[ctc_loss=0.1181, att_loss=0.2562, loss=0.2286, over 17314.00 frames. utt_duration=878 frames, utt_pad_proportion=0.08155, over 79.00 utterances.], tot_loss[ctc_loss=0.09311, att_loss=0.2454, loss=0.215, over 1302220.54 frames. utt_duration=1183 frames, utt_pad_proportion=0.06973, over 4409.34 utterances.], batch size: 79, lr: 7.35e-03, grad_scale: 8.0 2023-03-08 11:13:03,916 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8958, 5.1985, 5.4886, 5.4315, 5.3381, 5.8299, 5.0963, 5.9808], device='cuda:0'), covar=tensor([0.0700, 0.0703, 0.0756, 0.1077, 0.1878, 0.0943, 0.0835, 0.0609], device='cuda:0'), in_proj_covar=tensor([0.0789, 0.0466, 0.0549, 0.0611, 0.0804, 0.0560, 0.0457, 0.0541], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:13:35,206 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=55909.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:13:44,372 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=55915.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:13:57,609 INFO [train2.py:809] (0/4) Epoch 15, batch 150, loss[ctc_loss=0.07063, att_loss=0.2231, loss=0.1926, over 16008.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007357, over 40.00 utterances.], tot_loss[ctc_loss=0.09079, att_loss=0.2442, loss=0.2135, over 1733892.86 frames. utt_duration=1209 frames, utt_pad_proportion=0.06295, over 5742.99 utterances.], batch size: 40, lr: 7.35e-03, grad_scale: 8.0 2023-03-08 11:14:05,581 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.292e+02 2.863e+02 3.453e+02 8.582e+02, threshold=5.726e+02, percent-clipped=4.0 2023-03-08 11:14:13,696 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7332, 3.5671, 3.4652, 3.0179, 3.5992, 3.6094, 3.6391, 2.6846], device='cuda:0'), covar=tensor([0.1197, 0.1589, 0.2233, 0.6089, 0.1307, 0.3459, 0.1042, 0.6295], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0143, 0.0154, 0.0227, 0.0118, 0.0208, 0.0131, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:15:19,358 INFO [train2.py:809] (0/4) Epoch 15, batch 200, loss[ctc_loss=0.1115, att_loss=0.2685, loss=0.2371, over 16994.00 frames. utt_duration=688 frames, utt_pad_proportion=0.1368, over 99.00 utterances.], tot_loss[ctc_loss=0.08917, att_loss=0.244, loss=0.2131, over 2079659.44 frames. utt_duration=1222 frames, utt_pad_proportion=0.05731, over 6813.81 utterances.], batch size: 99, lr: 7.34e-03, grad_scale: 8.0 2023-03-08 11:15:57,260 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-03-08 11:16:04,588 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-56000.pt 2023-03-08 11:16:45,491 INFO [train2.py:809] (0/4) Epoch 15, batch 250, loss[ctc_loss=0.09069, att_loss=0.2468, loss=0.2155, over 17304.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01106, over 55.00 utterances.], tot_loss[ctc_loss=0.08931, att_loss=0.244, loss=0.2131, over 2340001.92 frames. utt_duration=1189 frames, utt_pad_proportion=0.06927, over 7880.31 utterances.], batch size: 55, lr: 7.34e-03, grad_scale: 8.0 2023-03-08 11:16:53,151 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 2.127e+02 2.516e+02 2.865e+02 8.492e+02, threshold=5.032e+02, percent-clipped=2.0 2023-03-08 11:17:15,635 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-08 11:17:37,364 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0600, 6.2671, 5.7948, 6.0440, 5.9389, 5.5032, 5.7966, 5.5180], device='cuda:0'), covar=tensor([0.1365, 0.0999, 0.0725, 0.0760, 0.0732, 0.1396, 0.2322, 0.2546], device='cuda:0'), in_proj_covar=tensor([0.0468, 0.0544, 0.0416, 0.0405, 0.0389, 0.0439, 0.0563, 0.0494], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 11:18:06,087 INFO [train2.py:809] (0/4) Epoch 15, batch 300, loss[ctc_loss=0.1325, att_loss=0.268, loss=0.2409, over 14002.00 frames. utt_duration=387.8 frames, utt_pad_proportion=0.3268, over 145.00 utterances.], tot_loss[ctc_loss=0.08995, att_loss=0.2441, loss=0.2133, over 2549905.01 frames. utt_duration=1198 frames, utt_pad_proportion=0.06637, over 8521.64 utterances.], batch size: 145, lr: 7.34e-03, grad_scale: 8.0 2023-03-08 11:18:36,290 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5311, 2.5183, 3.6497, 3.0961, 3.5478, 4.6601, 4.3742, 3.2853], device='cuda:0'), covar=tensor([0.0340, 0.2241, 0.1179, 0.1350, 0.1082, 0.0911, 0.0646, 0.1416], device='cuda:0'), in_proj_covar=tensor([0.0235, 0.0235, 0.0257, 0.0206, 0.0248, 0.0331, 0.0236, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:18:55,665 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56104.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:19:25,553 INFO [train2.py:809] (0/4) Epoch 15, batch 350, loss[ctc_loss=0.08296, att_loss=0.234, loss=0.2038, over 15941.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007389, over 41.00 utterances.], tot_loss[ctc_loss=0.09074, att_loss=0.2449, loss=0.2141, over 2713053.67 frames. utt_duration=1211 frames, utt_pad_proportion=0.06088, over 8971.70 utterances.], batch size: 41, lr: 7.34e-03, grad_scale: 8.0 2023-03-08 11:19:34,068 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 2.148e+02 2.622e+02 3.077e+02 6.565e+02, threshold=5.243e+02, percent-clipped=3.0 2023-03-08 11:20:01,807 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56145.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:20:45,055 INFO [train2.py:809] (0/4) Epoch 15, batch 400, loss[ctc_loss=0.1014, att_loss=0.2264, loss=0.2014, over 15755.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009499, over 38.00 utterances.], tot_loss[ctc_loss=0.0906, att_loss=0.2444, loss=0.2137, over 2830586.64 frames. utt_duration=1206 frames, utt_pad_proportion=0.06443, over 9396.89 utterances.], batch size: 38, lr: 7.33e-03, grad_scale: 8.0 2023-03-08 11:21:26,955 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3412, 2.7038, 3.3462, 4.3128, 3.7261, 3.9081, 2.8349, 1.9537], device='cuda:0'), covar=tensor([0.0625, 0.2025, 0.0901, 0.0458, 0.0782, 0.0408, 0.1504, 0.2382], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0216, 0.0193, 0.0201, 0.0207, 0.0165, 0.0200, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 11:21:34,922 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56204.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:21:38,225 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56206.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:22:04,491 INFO [train2.py:809] (0/4) Epoch 15, batch 450, loss[ctc_loss=0.07682, att_loss=0.2401, loss=0.2075, over 16321.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006566, over 45.00 utterances.], tot_loss[ctc_loss=0.08956, att_loss=0.2438, loss=0.2129, over 2931942.14 frames. utt_duration=1233 frames, utt_pad_proportion=0.05764, over 9520.46 utterances.], batch size: 45, lr: 7.33e-03, grad_scale: 8.0 2023-03-08 11:22:12,113 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 2.121e+02 2.729e+02 3.274e+02 6.212e+02, threshold=5.457e+02, percent-clipped=2.0 2023-03-08 11:22:15,581 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7268, 4.8125, 4.6236, 4.4960, 5.2125, 4.8266, 4.6795, 2.4607], device='cuda:0'), covar=tensor([0.0194, 0.0202, 0.0252, 0.0312, 0.1082, 0.0190, 0.0262, 0.1939], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0143, 0.0150, 0.0164, 0.0345, 0.0131, 0.0136, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:22:38,276 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 11:22:55,032 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0233, 3.8533, 3.2784, 3.5678, 4.1553, 3.7167, 3.1165, 4.4668], device='cuda:0'), covar=tensor([0.1135, 0.0508, 0.1125, 0.0724, 0.0609, 0.0705, 0.0886, 0.0444], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0200, 0.0212, 0.0186, 0.0254, 0.0226, 0.0191, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:23:06,353 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7063, 2.1916, 5.0326, 3.9456, 2.9390, 4.3663, 4.8136, 4.7536], device='cuda:0'), covar=tensor([0.0144, 0.1821, 0.0124, 0.0926, 0.1814, 0.0204, 0.0087, 0.0146], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0241, 0.0153, 0.0304, 0.0264, 0.0190, 0.0138, 0.0168], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:23:23,983 INFO [train2.py:809] (0/4) Epoch 15, batch 500, loss[ctc_loss=0.1373, att_loss=0.2675, loss=0.2415, over 14257.00 frames. utt_duration=392.1 frames, utt_pad_proportion=0.3181, over 146.00 utterances.], tot_loss[ctc_loss=0.08958, att_loss=0.2441, loss=0.2132, over 3010494.43 frames. utt_duration=1226 frames, utt_pad_proportion=0.05972, over 9832.29 utterances.], batch size: 146, lr: 7.33e-03, grad_scale: 8.0 2023-03-08 11:23:25,935 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6097, 3.5526, 3.4723, 3.0667, 3.6096, 3.4645, 3.5996, 2.5293], device='cuda:0'), covar=tensor([0.1249, 0.2215, 0.3345, 0.4480, 0.0967, 0.4324, 0.1231, 0.6004], device='cuda:0'), in_proj_covar=tensor([0.0125, 0.0145, 0.0157, 0.0225, 0.0118, 0.0210, 0.0132, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:23:34,216 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7403, 1.8723, 2.2169, 1.9636, 2.6752, 2.2332, 2.3239, 3.1303], device='cuda:0'), covar=tensor([0.1800, 0.4555, 0.3327, 0.2326, 0.1473, 0.1878, 0.3087, 0.0746], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0097, 0.0098, 0.0084, 0.0087, 0.0080, 0.0103, 0.0069], device='cuda:0'), out_proj_covar=tensor([6.3791e-05, 7.0686e-05, 7.2939e-05, 6.2242e-05, 6.2157e-05, 6.1807e-05, 7.2780e-05, 5.4143e-05], device='cuda:0') 2023-03-08 11:24:03,485 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1765, 4.5357, 4.4570, 4.6774, 2.5271, 4.6031, 2.7497, 1.7181], device='cuda:0'), covar=tensor([0.0348, 0.0159, 0.0621, 0.0131, 0.1920, 0.0127, 0.1467, 0.1805], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0134, 0.0253, 0.0123, 0.0219, 0.0115, 0.0224, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 11:24:12,470 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4203, 5.0013, 4.7587, 4.8367, 5.0469, 4.6352, 3.4088, 4.9077], device='cuda:0'), covar=tensor([0.0137, 0.0107, 0.0146, 0.0098, 0.0097, 0.0120, 0.0742, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0078, 0.0097, 0.0061, 0.0066, 0.0077, 0.0095, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:24:43,548 INFO [train2.py:809] (0/4) Epoch 15, batch 550, loss[ctc_loss=0.07669, att_loss=0.2195, loss=0.1909, over 15885.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.00797, over 39.00 utterances.], tot_loss[ctc_loss=0.08947, att_loss=0.2443, loss=0.2133, over 3076738.04 frames. utt_duration=1230 frames, utt_pad_proportion=0.05629, over 10017.45 utterances.], batch size: 39, lr: 7.32e-03, grad_scale: 8.0 2023-03-08 11:24:45,621 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5731, 1.8534, 2.0721, 2.0845, 2.7524, 2.0829, 2.1471, 2.9807], device='cuda:0'), covar=tensor([0.1590, 0.3645, 0.2839, 0.2695, 0.1824, 0.2303, 0.3066, 0.0764], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0097, 0.0098, 0.0084, 0.0087, 0.0081, 0.0103, 0.0068], device='cuda:0'), out_proj_covar=tensor([6.3815e-05, 7.0776e-05, 7.3035e-05, 6.2456e-05, 6.2318e-05, 6.2092e-05, 7.2985e-05, 5.4027e-05], device='cuda:0') 2023-03-08 11:24:51,304 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.338e+02 2.199e+02 2.714e+02 3.543e+02 1.087e+03, threshold=5.428e+02, percent-clipped=4.0 2023-03-08 11:25:41,694 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-03-08 11:26:02,585 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8374, 6.0845, 5.4902, 5.8055, 5.7063, 5.2671, 5.4890, 5.1917], device='cuda:0'), covar=tensor([0.1232, 0.0872, 0.0827, 0.0708, 0.0835, 0.1406, 0.2243, 0.2483], device='cuda:0'), in_proj_covar=tensor([0.0465, 0.0545, 0.0414, 0.0407, 0.0390, 0.0435, 0.0557, 0.0493], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 11:26:03,955 INFO [train2.py:809] (0/4) Epoch 15, batch 600, loss[ctc_loss=0.08836, att_loss=0.2472, loss=0.2154, over 16621.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005648, over 47.00 utterances.], tot_loss[ctc_loss=0.08945, att_loss=0.2438, loss=0.213, over 3121612.54 frames. utt_duration=1270 frames, utt_pad_proportion=0.04743, over 9840.51 utterances.], batch size: 47, lr: 7.32e-03, grad_scale: 8.0 2023-03-08 11:26:54,546 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56404.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:27:23,619 INFO [train2.py:809] (0/4) Epoch 15, batch 650, loss[ctc_loss=0.1148, att_loss=0.2516, loss=0.2242, over 16414.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006845, over 44.00 utterances.], tot_loss[ctc_loss=0.08984, att_loss=0.2435, loss=0.2128, over 3152790.29 frames. utt_duration=1272 frames, utt_pad_proportion=0.04759, over 9925.97 utterances.], batch size: 44, lr: 7.32e-03, grad_scale: 8.0 2023-03-08 11:27:30,070 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4637, 3.6232, 3.0080, 3.0812, 3.8077, 3.3851, 2.3271, 3.7552], device='cuda:0'), covar=tensor([0.1182, 0.0372, 0.0890, 0.0672, 0.0550, 0.0582, 0.1080, 0.0600], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0200, 0.0214, 0.0186, 0.0255, 0.0227, 0.0191, 0.0274], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:27:31,803 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.524e+02 2.098e+02 2.434e+02 3.232e+02 5.405e+02, threshold=4.867e+02, percent-clipped=0.0 2023-03-08 11:28:09,880 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56452.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:28:16,803 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-08 11:28:42,741 INFO [train2.py:809] (0/4) Epoch 15, batch 700, loss[ctc_loss=0.07353, att_loss=0.2469, loss=0.2123, over 16634.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004901, over 47.00 utterances.], tot_loss[ctc_loss=0.09005, att_loss=0.2435, loss=0.2128, over 3180002.85 frames. utt_duration=1268 frames, utt_pad_proportion=0.04892, over 10042.57 utterances.], batch size: 47, lr: 7.31e-03, grad_scale: 8.0 2023-03-08 11:29:23,454 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4367, 5.0190, 4.7683, 4.8617, 5.1297, 4.5125, 3.1313, 4.7967], device='cuda:0'), covar=tensor([0.0137, 0.0099, 0.0130, 0.0091, 0.0071, 0.0124, 0.0840, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0077, 0.0096, 0.0059, 0.0065, 0.0076, 0.0095, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:29:28,137 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56501.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:29:33,081 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56504.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:30:02,714 INFO [train2.py:809] (0/4) Epoch 15, batch 750, loss[ctc_loss=0.09628, att_loss=0.2238, loss=0.1983, over 15646.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008861, over 37.00 utterances.], tot_loss[ctc_loss=0.08962, att_loss=0.2435, loss=0.2127, over 3204139.01 frames. utt_duration=1258 frames, utt_pad_proportion=0.05108, over 10203.07 utterances.], batch size: 37, lr: 7.31e-03, grad_scale: 8.0 2023-03-08 11:30:03,039 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56523.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 11:30:11,027 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 2.381e+02 2.975e+02 3.740e+02 8.505e+02, threshold=5.949e+02, percent-clipped=7.0 2023-03-08 11:30:25,841 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0705, 4.5020, 4.2933, 4.3868, 4.5287, 4.1212, 2.9637, 4.2916], device='cuda:0'), covar=tensor([0.0138, 0.0109, 0.0142, 0.0090, 0.0100, 0.0133, 0.0741, 0.0256], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0077, 0.0096, 0.0059, 0.0065, 0.0076, 0.0094, 0.0097], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:30:30,518 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 11:30:49,606 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56552.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:31:22,460 INFO [train2.py:809] (0/4) Epoch 15, batch 800, loss[ctc_loss=0.1458, att_loss=0.274, loss=0.2484, over 14207.00 frames. utt_duration=393.4 frames, utt_pad_proportion=0.3159, over 145.00 utterances.], tot_loss[ctc_loss=0.08929, att_loss=0.2438, loss=0.2129, over 3226774.19 frames. utt_duration=1249 frames, utt_pad_proportion=0.05198, over 10346.30 utterances.], batch size: 145, lr: 7.31e-03, grad_scale: 8.0 2023-03-08 11:31:22,645 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3106, 4.6548, 4.8950, 4.7662, 4.8441, 5.2077, 4.7762, 5.3069], device='cuda:0'), covar=tensor([0.0782, 0.0798, 0.0783, 0.1310, 0.1730, 0.0926, 0.1250, 0.0683], device='cuda:0'), in_proj_covar=tensor([0.0791, 0.0468, 0.0545, 0.0608, 0.0795, 0.0555, 0.0451, 0.0542], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:31:41,518 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56584.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 11:31:55,577 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.90 vs. limit=5.0 2023-03-08 11:32:44,281 INFO [train2.py:809] (0/4) Epoch 15, batch 850, loss[ctc_loss=0.1004, att_loss=0.257, loss=0.2257, over 16774.00 frames. utt_duration=686.3 frames, utt_pad_proportion=0.14, over 98.00 utterances.], tot_loss[ctc_loss=0.08874, att_loss=0.2433, loss=0.2124, over 3241462.15 frames. utt_duration=1237 frames, utt_pad_proportion=0.05454, over 10490.35 utterances.], batch size: 98, lr: 7.30e-03, grad_scale: 8.0 2023-03-08 11:32:52,932 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.144e+02 2.509e+02 3.065e+02 7.548e+02, threshold=5.017e+02, percent-clipped=1.0 2023-03-08 11:34:04,334 INFO [train2.py:809] (0/4) Epoch 15, batch 900, loss[ctc_loss=0.08595, att_loss=0.2283, loss=0.1998, over 16269.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007889, over 43.00 utterances.], tot_loss[ctc_loss=0.08822, att_loss=0.2428, loss=0.2119, over 3235913.98 frames. utt_duration=1238 frames, utt_pad_proportion=0.05684, over 10468.07 utterances.], batch size: 43, lr: 7.30e-03, grad_scale: 8.0 2023-03-08 11:35:24,397 INFO [train2.py:809] (0/4) Epoch 15, batch 950, loss[ctc_loss=0.1117, att_loss=0.2647, loss=0.2341, over 17056.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008578, over 52.00 utterances.], tot_loss[ctc_loss=0.08953, att_loss=0.2436, loss=0.2128, over 3236740.51 frames. utt_duration=1208 frames, utt_pad_proportion=0.06662, over 10729.92 utterances.], batch size: 52, lr: 7.30e-03, grad_scale: 8.0 2023-03-08 11:35:32,186 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 2.400e+02 2.758e+02 3.423e+02 1.509e+03, threshold=5.516e+02, percent-clipped=3.0 2023-03-08 11:35:35,572 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 11:36:16,641 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8246, 1.8326, 2.3342, 2.0454, 2.7686, 2.3943, 2.4786, 2.6326], device='cuda:0'), covar=tensor([0.1373, 0.3909, 0.2510, 0.2056, 0.1254, 0.1507, 0.2719, 0.1057], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0097, 0.0098, 0.0085, 0.0087, 0.0079, 0.0104, 0.0069], device='cuda:0'), out_proj_covar=tensor([6.3895e-05, 7.0973e-05, 7.3118e-05, 6.2677e-05, 6.2510e-05, 6.1547e-05, 7.3226e-05, 5.4745e-05], device='cuda:0') 2023-03-08 11:36:30,437 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0050, 4.9727, 4.8981, 2.1409, 1.9478, 2.7550, 2.3533, 3.8044], device='cuda:0'), covar=tensor([0.0758, 0.0217, 0.0193, 0.4841, 0.5845, 0.2606, 0.3288, 0.1704], device='cuda:0'), in_proj_covar=tensor([0.0342, 0.0239, 0.0244, 0.0224, 0.0346, 0.0335, 0.0230, 0.0355], device='cuda:0'), out_proj_covar=tensor([1.4946e-04, 8.9555e-05, 1.0551e-04, 9.7585e-05, 1.4716e-04, 1.3269e-04, 9.2022e-05, 1.4716e-04], device='cuda:0') 2023-03-08 11:36:44,601 INFO [train2.py:809] (0/4) Epoch 15, batch 1000, loss[ctc_loss=0.07719, att_loss=0.2218, loss=0.1929, over 15642.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008972, over 37.00 utterances.], tot_loss[ctc_loss=0.0896, att_loss=0.2437, loss=0.2129, over 3239161.14 frames. utt_duration=1209 frames, utt_pad_proportion=0.06759, over 10727.80 utterances.], batch size: 37, lr: 7.29e-03, grad_scale: 8.0 2023-03-08 11:37:17,178 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1604, 5.3746, 5.4540, 5.4032, 5.4840, 5.5062, 5.1719, 4.9855], device='cuda:0'), covar=tensor([0.0976, 0.0558, 0.0232, 0.0406, 0.0265, 0.0272, 0.0334, 0.0312], device='cuda:0'), in_proj_covar=tensor([0.0490, 0.0324, 0.0294, 0.0321, 0.0377, 0.0400, 0.0320, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:37:30,229 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=56801.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:37:33,355 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56803.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:38:05,059 INFO [train2.py:809] (0/4) Epoch 15, batch 1050, loss[ctc_loss=0.08616, att_loss=0.2478, loss=0.2155, over 17343.00 frames. utt_duration=1007 frames, utt_pad_proportion=0.0482, over 69.00 utterances.], tot_loss[ctc_loss=0.08887, att_loss=0.2431, loss=0.2122, over 3243967.70 frames. utt_duration=1213 frames, utt_pad_proportion=0.06655, over 10710.61 utterances.], batch size: 69, lr: 7.29e-03, grad_scale: 8.0 2023-03-08 11:38:12,498 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.111e+02 2.438e+02 3.282e+02 6.926e+02, threshold=4.877e+02, percent-clipped=2.0 2023-03-08 11:38:45,034 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5293, 2.2326, 5.0763, 3.8392, 2.7641, 4.3097, 4.8163, 4.7509], device='cuda:0'), covar=tensor([0.0229, 0.1820, 0.0132, 0.0889, 0.1966, 0.0253, 0.0109, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0239, 0.0152, 0.0301, 0.0262, 0.0189, 0.0138, 0.0167], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:38:46,200 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=56849.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:39:07,047 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9428, 5.2289, 5.5006, 5.3374, 5.3873, 5.9524, 5.1940, 5.9998], device='cuda:0'), covar=tensor([0.0674, 0.0795, 0.0796, 0.1122, 0.1715, 0.0752, 0.0712, 0.0731], device='cuda:0'), in_proj_covar=tensor([0.0795, 0.0466, 0.0547, 0.0606, 0.0796, 0.0554, 0.0453, 0.0543], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:39:10,435 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56864.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:39:24,729 INFO [train2.py:809] (0/4) Epoch 15, batch 1100, loss[ctc_loss=0.08531, att_loss=0.2341, loss=0.2043, over 15767.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.00884, over 38.00 utterances.], tot_loss[ctc_loss=0.08787, att_loss=0.2418, loss=0.211, over 3239696.32 frames. utt_duration=1253 frames, utt_pad_proportion=0.05836, over 10358.10 utterances.], batch size: 38, lr: 7.29e-03, grad_scale: 8.0 2023-03-08 11:39:34,111 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=56879.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 11:40:17,777 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.92 vs. limit=5.0 2023-03-08 11:40:18,390 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-08 11:40:20,529 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56908.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:40:44,820 INFO [train2.py:809] (0/4) Epoch 15, batch 1150, loss[ctc_loss=0.07648, att_loss=0.2393, loss=0.2067, over 16270.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007782, over 43.00 utterances.], tot_loss[ctc_loss=0.08802, att_loss=0.2427, loss=0.2117, over 3255322.42 frames. utt_duration=1265 frames, utt_pad_proportion=0.05151, over 10303.64 utterances.], batch size: 43, lr: 7.28e-03, grad_scale: 8.0 2023-03-08 11:40:52,695 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.206e+02 2.173e+02 2.768e+02 3.173e+02 4.889e+02, threshold=5.536e+02, percent-clipped=1.0 2023-03-08 11:41:58,520 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=56969.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:42:04,418 INFO [train2.py:809] (0/4) Epoch 15, batch 1200, loss[ctc_loss=0.06938, att_loss=0.2269, loss=0.1954, over 15935.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007606, over 41.00 utterances.], tot_loss[ctc_loss=0.08947, att_loss=0.2434, loss=0.2126, over 3256998.61 frames. utt_duration=1233 frames, utt_pad_proportion=0.05803, over 10581.00 utterances.], batch size: 41, lr: 7.28e-03, grad_scale: 8.0 2023-03-08 11:42:36,777 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56993.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:42:39,791 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=56995.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:43:23,348 INFO [train2.py:809] (0/4) Epoch 15, batch 1250, loss[ctc_loss=0.09855, att_loss=0.2538, loss=0.2228, over 17097.00 frames. utt_duration=692.3 frames, utt_pad_proportion=0.1314, over 99.00 utterances.], tot_loss[ctc_loss=0.08898, att_loss=0.2434, loss=0.2125, over 3263950.17 frames. utt_duration=1245 frames, utt_pad_proportion=0.05442, over 10501.52 utterances.], batch size: 99, lr: 7.28e-03, grad_scale: 8.0 2023-03-08 11:43:31,117 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.571e+02 2.221e+02 2.683e+02 3.397e+02 1.061e+03, threshold=5.366e+02, percent-clipped=4.0 2023-03-08 11:43:42,912 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3080, 2.4068, 3.0973, 2.5213, 2.9851, 3.4584, 3.3401, 2.6966], device='cuda:0'), covar=tensor([0.0470, 0.1607, 0.1105, 0.1152, 0.0894, 0.1135, 0.0729, 0.1201], device='cuda:0'), in_proj_covar=tensor([0.0238, 0.0237, 0.0262, 0.0208, 0.0252, 0.0335, 0.0240, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:44:13,634 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57054.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:44:16,967 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57056.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:44:43,935 INFO [train2.py:809] (0/4) Epoch 15, batch 1300, loss[ctc_loss=0.06202, att_loss=0.2268, loss=0.1939, over 16428.00 frames. utt_duration=1495 frames, utt_pad_proportion=0.005984, over 44.00 utterances.], tot_loss[ctc_loss=0.08848, att_loss=0.243, loss=0.2121, over 3263058.80 frames. utt_duration=1239 frames, utt_pad_proportion=0.05622, over 10543.44 utterances.], batch size: 44, lr: 7.27e-03, grad_scale: 8.0 2023-03-08 11:46:03,444 INFO [train2.py:809] (0/4) Epoch 15, batch 1350, loss[ctc_loss=0.08157, att_loss=0.2444, loss=0.2118, over 16537.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005721, over 45.00 utterances.], tot_loss[ctc_loss=0.08785, att_loss=0.2424, loss=0.2115, over 3258955.15 frames. utt_duration=1246 frames, utt_pad_proportion=0.05622, over 10470.82 utterances.], batch size: 45, lr: 7.27e-03, grad_scale: 8.0 2023-03-08 11:46:11,169 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.494e+02 2.073e+02 2.593e+02 3.161e+02 1.501e+03, threshold=5.185e+02, percent-clipped=3.0 2023-03-08 11:46:53,904 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5591, 5.0398, 5.0531, 5.0560, 5.1030, 5.0737, 4.8381, 4.6340], device='cuda:0'), covar=tensor([0.1413, 0.0597, 0.0367, 0.0494, 0.0403, 0.0420, 0.0456, 0.0375], device='cuda:0'), in_proj_covar=tensor([0.0483, 0.0322, 0.0293, 0.0315, 0.0372, 0.0396, 0.0318, 0.0357], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:47:00,472 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57159.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:47:10,939 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-03-08 11:47:22,524 INFO [train2.py:809] (0/4) Epoch 15, batch 1400, loss[ctc_loss=0.08135, att_loss=0.2422, loss=0.21, over 16533.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006557, over 45.00 utterances.], tot_loss[ctc_loss=0.08903, att_loss=0.2439, loss=0.2129, over 3265220.80 frames. utt_duration=1216 frames, utt_pad_proportion=0.06274, over 10752.47 utterances.], batch size: 45, lr: 7.27e-03, grad_scale: 8.0 2023-03-08 11:47:32,842 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57179.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 11:47:44,283 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3456, 2.4246, 3.0857, 2.5143, 2.9680, 3.4739, 3.3365, 2.6363], device='cuda:0'), covar=tensor([0.0547, 0.1804, 0.1184, 0.1322, 0.1139, 0.1146, 0.0755, 0.1407], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0236, 0.0261, 0.0208, 0.0252, 0.0333, 0.0238, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:47:44,313 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57186.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:48:00,808 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57197.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:48:32,262 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5290, 2.5175, 5.0614, 3.8669, 3.0446, 4.5169, 4.9327, 4.7239], device='cuda:0'), covar=tensor([0.0246, 0.1610, 0.0180, 0.0969, 0.1710, 0.0196, 0.0119, 0.0208], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0242, 0.0155, 0.0305, 0.0264, 0.0192, 0.0139, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 11:48:38,116 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4739, 4.9689, 4.7586, 4.9330, 5.1012, 4.6241, 3.4449, 4.8117], device='cuda:0'), covar=tensor([0.0119, 0.0115, 0.0139, 0.0081, 0.0092, 0.0117, 0.0675, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0079, 0.0097, 0.0059, 0.0065, 0.0077, 0.0094, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:48:42,463 INFO [train2.py:809] (0/4) Epoch 15, batch 1450, loss[ctc_loss=0.07955, att_loss=0.2424, loss=0.2098, over 16471.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.007046, over 46.00 utterances.], tot_loss[ctc_loss=0.08837, att_loss=0.2433, loss=0.2123, over 3270248.54 frames. utt_duration=1235 frames, utt_pad_proportion=0.05752, over 10606.74 utterances.], batch size: 46, lr: 7.26e-03, grad_scale: 8.0 2023-03-08 11:48:48,471 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57227.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 11:48:49,835 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.461e+02 2.071e+02 2.451e+02 3.320e+02 6.721e+02, threshold=4.903e+02, percent-clipped=4.0 2023-03-08 11:48:53,793 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2454, 4.4605, 4.6664, 4.9249, 2.8874, 4.7513, 3.1360, 1.9565], device='cuda:0'), covar=tensor([0.0306, 0.0277, 0.0612, 0.0151, 0.1651, 0.0180, 0.1260, 0.1704], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0136, 0.0254, 0.0127, 0.0220, 0.0116, 0.0224, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 11:49:15,872 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57244.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:49:20,428 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57247.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:49:38,298 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57258.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 11:49:48,003 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57264.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:50:02,008 INFO [train2.py:809] (0/4) Epoch 15, batch 1500, loss[ctc_loss=0.07115, att_loss=0.2127, loss=0.1844, over 15479.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.01002, over 36.00 utterances.], tot_loss[ctc_loss=0.08866, att_loss=0.2433, loss=0.2124, over 3269083.75 frames. utt_duration=1237 frames, utt_pad_proportion=0.05756, over 10583.64 utterances.], batch size: 36, lr: 7.26e-03, grad_scale: 16.0 2023-03-08 11:50:53,386 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57305.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:51:17,802 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-08 11:51:20,963 INFO [train2.py:809] (0/4) Epoch 15, batch 1550, loss[ctc_loss=0.1256, att_loss=0.2714, loss=0.2422, over 17377.00 frames. utt_duration=1009 frames, utt_pad_proportion=0.0492, over 69.00 utterances.], tot_loss[ctc_loss=0.08908, att_loss=0.2434, loss=0.2125, over 3274413.40 frames. utt_duration=1240 frames, utt_pad_proportion=0.05599, over 10576.52 utterances.], batch size: 69, lr: 7.26e-03, grad_scale: 16.0 2023-03-08 11:51:29,316 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.286e+02 2.083e+02 2.433e+02 3.048e+02 8.459e+02, threshold=4.866e+02, percent-clipped=4.0 2023-03-08 11:52:02,023 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57349.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:52:03,206 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.55 vs. limit=2.0 2023-03-08 11:52:05,130 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57351.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:52:39,938 INFO [train2.py:809] (0/4) Epoch 15, batch 1600, loss[ctc_loss=0.07968, att_loss=0.2404, loss=0.2083, over 17326.00 frames. utt_duration=878.9 frames, utt_pad_proportion=0.0797, over 79.00 utterances.], tot_loss[ctc_loss=0.08959, att_loss=0.2439, loss=0.2131, over 3274059.08 frames. utt_duration=1223 frames, utt_pad_proportion=0.05987, over 10719.19 utterances.], batch size: 79, lr: 7.26e-03, grad_scale: 16.0 2023-03-08 11:53:20,341 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57398.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:54:00,795 INFO [train2.py:809] (0/4) Epoch 15, batch 1650, loss[ctc_loss=0.08165, att_loss=0.2514, loss=0.2174, over 17052.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009601, over 53.00 utterances.], tot_loss[ctc_loss=0.08891, att_loss=0.2435, loss=0.2125, over 3275491.82 frames. utt_duration=1238 frames, utt_pad_proportion=0.05619, over 10597.64 utterances.], batch size: 53, lr: 7.25e-03, grad_scale: 16.0 2023-03-08 11:54:09,586 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 2.043e+02 2.377e+02 2.961e+02 5.254e+02, threshold=4.753e+02, percent-clipped=2.0 2023-03-08 11:54:42,155 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-08 11:54:59,123 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:54:59,214 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57459.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:55:03,769 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57462.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:55:20,184 INFO [train2.py:809] (0/4) Epoch 15, batch 1700, loss[ctc_loss=0.07662, att_loss=0.2326, loss=0.2014, over 16773.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006436, over 48.00 utterances.], tot_loss[ctc_loss=0.08844, att_loss=0.2437, loss=0.2127, over 3277926.37 frames. utt_duration=1247 frames, utt_pad_proportion=0.05417, over 10523.83 utterances.], batch size: 48, lr: 7.25e-03, grad_scale: 16.0 2023-03-08 11:55:54,904 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57495.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:56:13,672 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57507.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:56:38,717 INFO [train2.py:809] (0/4) Epoch 15, batch 1750, loss[ctc_loss=0.07314, att_loss=0.2408, loss=0.2073, over 17392.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03435, over 63.00 utterances.], tot_loss[ctc_loss=0.08885, att_loss=0.2437, loss=0.2127, over 3279545.64 frames. utt_duration=1255 frames, utt_pad_proportion=0.05084, over 10461.61 utterances.], batch size: 63, lr: 7.25e-03, grad_scale: 16.0 2023-03-08 11:56:39,763 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57523.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:56:46,885 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.541e+02 2.236e+02 2.703e+02 3.413e+02 7.176e+02, threshold=5.406e+02, percent-clipped=5.0 2023-03-08 11:57:08,478 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57542.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:57:26,201 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57553.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 11:57:31,771 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57556.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:57:41,059 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0370, 3.8201, 3.2379, 3.5769, 4.0574, 3.7222, 3.1520, 4.3571], device='cuda:0'), covar=tensor([0.0981, 0.0477, 0.0946, 0.0598, 0.0640, 0.0632, 0.0782, 0.0473], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0197, 0.0211, 0.0184, 0.0254, 0.0223, 0.0189, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:57:44,141 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57564.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:57:58,284 INFO [train2.py:809] (0/4) Epoch 15, batch 1800, loss[ctc_loss=0.05778, att_loss=0.2207, loss=0.1881, over 16015.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006691, over 40.00 utterances.], tot_loss[ctc_loss=0.08817, att_loss=0.2434, loss=0.2124, over 3280811.77 frames. utt_duration=1267 frames, utt_pad_proportion=0.04939, over 10370.97 utterances.], batch size: 40, lr: 7.24e-03, grad_scale: 16.0 2023-03-08 11:58:41,639 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57600.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:58:49,347 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 11:58:56,176 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6430, 3.0123, 3.7897, 4.6603, 4.2172, 4.0326, 2.9865, 2.2710], device='cuda:0'), covar=tensor([0.0633, 0.1760, 0.0729, 0.0476, 0.0667, 0.0421, 0.1426, 0.2111], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0210, 0.0187, 0.0198, 0.0205, 0.0164, 0.0198, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 11:59:00,592 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57612.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 11:59:00,848 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3157, 2.6813, 3.5626, 2.8462, 3.4155, 4.4738, 4.2173, 3.1588], device='cuda:0'), covar=tensor([0.0416, 0.1782, 0.1163, 0.1353, 0.1085, 0.0758, 0.0613, 0.1275], device='cuda:0'), in_proj_covar=tensor([0.0238, 0.0237, 0.0265, 0.0209, 0.0257, 0.0337, 0.0240, 0.0230], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 11:59:19,509 INFO [train2.py:809] (0/4) Epoch 15, batch 1850, loss[ctc_loss=0.1562, att_loss=0.2841, loss=0.2585, over 14344.00 frames. utt_duration=394.7 frames, utt_pad_proportion=0.3136, over 146.00 utterances.], tot_loss[ctc_loss=0.08819, att_loss=0.2434, loss=0.2124, over 3271558.72 frames. utt_duration=1240 frames, utt_pad_proportion=0.05862, over 10562.16 utterances.], batch size: 146, lr: 7.24e-03, grad_scale: 16.0 2023-03-08 11:59:27,209 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.618e+02 2.149e+02 2.578e+02 3.271e+02 8.979e+02, threshold=5.155e+02, percent-clipped=4.0 2023-03-08 11:59:27,505 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8596, 5.1630, 5.1013, 5.0069, 5.1852, 5.1841, 4.8490, 4.6162], device='cuda:0'), covar=tensor([0.0948, 0.0481, 0.0299, 0.0566, 0.0328, 0.0308, 0.0382, 0.0401], device='cuda:0'), in_proj_covar=tensor([0.0491, 0.0324, 0.0294, 0.0324, 0.0380, 0.0402, 0.0323, 0.0362], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 11:59:31,148 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-03-08 12:00:00,081 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57649.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:00:03,757 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57651.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:00:38,107 INFO [train2.py:809] (0/4) Epoch 15, batch 1900, loss[ctc_loss=0.07175, att_loss=0.2464, loss=0.2115, over 16758.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006954, over 48.00 utterances.], tot_loss[ctc_loss=0.08793, att_loss=0.2431, loss=0.2121, over 3269354.64 frames. utt_duration=1256 frames, utt_pad_proportion=0.05509, over 10427.03 utterances.], batch size: 48, lr: 7.24e-03, grad_scale: 16.0 2023-03-08 12:01:04,929 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8120, 5.0910, 5.3928, 5.2197, 5.2590, 5.7841, 5.1618, 5.8963], device='cuda:0'), covar=tensor([0.0780, 0.0885, 0.0788, 0.1164, 0.1813, 0.0900, 0.0746, 0.0625], device='cuda:0'), in_proj_covar=tensor([0.0796, 0.0466, 0.0545, 0.0606, 0.0800, 0.0560, 0.0448, 0.0532], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 12:01:15,663 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57697.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:01:18,924 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57699.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:01:46,547 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 12:01:56,429 INFO [train2.py:809] (0/4) Epoch 15, batch 1950, loss[ctc_loss=0.06199, att_loss=0.2372, loss=0.2022, over 17019.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007744, over 51.00 utterances.], tot_loss[ctc_loss=0.0881, att_loss=0.2431, loss=0.2121, over 3274467.93 frames. utt_duration=1257 frames, utt_pad_proportion=0.05274, over 10432.50 utterances.], batch size: 51, lr: 7.23e-03, grad_scale: 16.0 2023-03-08 12:02:05,439 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 2.187e+02 2.629e+02 3.507e+02 9.887e+02, threshold=5.258e+02, percent-clipped=4.0 2023-03-08 12:02:05,916 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2244, 5.2442, 5.1573, 2.6055, 1.9293, 2.7248, 2.4960, 3.8547], device='cuda:0'), covar=tensor([0.0619, 0.0285, 0.0175, 0.3245, 0.5779, 0.2709, 0.3130, 0.1847], device='cuda:0'), in_proj_covar=tensor([0.0338, 0.0236, 0.0240, 0.0221, 0.0338, 0.0330, 0.0230, 0.0351], device='cuda:0'), out_proj_covar=tensor([1.4743e-04, 8.8964e-05, 1.0370e-04, 9.5789e-05, 1.4413e-04, 1.3086e-04, 9.1898e-05, 1.4527e-04], device='cuda:0') 2023-03-08 12:02:13,618 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=57733.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:02:21,436 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0225, 3.7486, 3.1869, 3.3651, 4.0058, 3.5117, 2.7260, 4.1309], device='cuda:0'), covar=tensor([0.0962, 0.0488, 0.0987, 0.0741, 0.0588, 0.0724, 0.0986, 0.0544], device='cuda:0'), in_proj_covar=tensor([0.0195, 0.0201, 0.0216, 0.0188, 0.0259, 0.0228, 0.0192, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 12:02:47,448 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57754.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:03:17,165 INFO [train2.py:809] (0/4) Epoch 15, batch 2000, loss[ctc_loss=0.08911, att_loss=0.2558, loss=0.2224, over 17356.00 frames. utt_duration=1178 frames, utt_pad_proportion=0.02144, over 59.00 utterances.], tot_loss[ctc_loss=0.08751, att_loss=0.243, loss=0.2119, over 3277910.43 frames. utt_duration=1257 frames, utt_pad_proportion=0.05147, over 10441.55 utterances.], batch size: 59, lr: 7.23e-03, grad_scale: 16.0 2023-03-08 12:03:50,228 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=57794.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:04:28,959 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57818.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:04:37,302 INFO [train2.py:809] (0/4) Epoch 15, batch 2050, loss[ctc_loss=0.07313, att_loss=0.2326, loss=0.2007, over 17096.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01546, over 56.00 utterances.], tot_loss[ctc_loss=0.08822, att_loss=0.2441, loss=0.2129, over 3284125.80 frames. utt_duration=1236 frames, utt_pad_proportion=0.05509, over 10645.36 utterances.], batch size: 56, lr: 7.23e-03, grad_scale: 16.0 2023-03-08 12:04:43,979 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 12:04:45,711 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.062e+02 2.536e+02 3.073e+02 8.062e+02, threshold=5.072e+02, percent-clipped=2.0 2023-03-08 12:05:07,100 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57842.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:05:22,321 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=57851.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:05:25,403 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57853.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 12:05:56,968 INFO [train2.py:809] (0/4) Epoch 15, batch 2100, loss[ctc_loss=0.08679, att_loss=0.257, loss=0.223, over 17021.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007773, over 51.00 utterances.], tot_loss[ctc_loss=0.08835, att_loss=0.2443, loss=0.2131, over 3287714.79 frames. utt_duration=1229 frames, utt_pad_proportion=0.05589, over 10714.07 utterances.], batch size: 51, lr: 7.22e-03, grad_scale: 16.0 2023-03-08 12:06:23,267 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57890.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:06:38,245 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0448, 3.7808, 3.1977, 3.3108, 4.0312, 3.5557, 2.8931, 4.2328], device='cuda:0'), covar=tensor([0.0981, 0.0430, 0.0989, 0.0746, 0.0604, 0.0757, 0.0892, 0.0486], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0199, 0.0215, 0.0188, 0.0256, 0.0228, 0.0190, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 12:06:40,406 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=57900.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:06:41,919 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57901.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:06:56,369 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5150, 3.3845, 3.3230, 3.0469, 3.3299, 3.2867, 3.4740, 2.6074], device='cuda:0'), covar=tensor([0.1000, 0.1464, 0.3408, 0.4500, 0.1795, 0.5352, 0.1540, 0.5604], device='cuda:0'), in_proj_covar=tensor([0.0124, 0.0146, 0.0157, 0.0224, 0.0120, 0.0212, 0.0134, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 12:06:58,639 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 12:07:17,264 INFO [train2.py:809] (0/4) Epoch 15, batch 2150, loss[ctc_loss=0.09461, att_loss=0.2489, loss=0.218, over 16755.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.007087, over 48.00 utterances.], tot_loss[ctc_loss=0.08859, att_loss=0.2445, loss=0.2133, over 3282798.83 frames. utt_duration=1237 frames, utt_pad_proportion=0.05389, over 10627.49 utterances.], batch size: 48, lr: 7.22e-03, grad_scale: 16.0 2023-03-08 12:07:25,214 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.164e+02 2.681e+02 3.169e+02 6.485e+02, threshold=5.361e+02, percent-clipped=3.0 2023-03-08 12:07:57,489 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=57948.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:08:37,645 INFO [train2.py:809] (0/4) Epoch 15, batch 2200, loss[ctc_loss=0.1053, att_loss=0.2529, loss=0.2234, over 16690.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005714, over 46.00 utterances.], tot_loss[ctc_loss=0.08823, att_loss=0.2438, loss=0.2127, over 3285906.65 frames. utt_duration=1264 frames, utt_pad_proportion=0.04686, over 10411.97 utterances.], batch size: 46, lr: 7.22e-03, grad_scale: 16.0 2023-03-08 12:08:40,125 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-08 12:08:57,447 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-08 12:09:20,614 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-58000.pt 2023-03-08 12:10:01,087 INFO [train2.py:809] (0/4) Epoch 15, batch 2250, loss[ctc_loss=0.1193, att_loss=0.2646, loss=0.2355, over 17319.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.0235, over 59.00 utterances.], tot_loss[ctc_loss=0.08793, att_loss=0.2436, loss=0.2125, over 3282881.20 frames. utt_duration=1264 frames, utt_pad_proportion=0.04636, over 10398.01 utterances.], batch size: 59, lr: 7.22e-03, grad_scale: 16.0 2023-03-08 12:10:08,781 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.417e+02 2.180e+02 2.771e+02 3.383e+02 6.703e+02, threshold=5.542e+02, percent-clipped=1.0 2023-03-08 12:10:49,933 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58054.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:11:20,344 INFO [train2.py:809] (0/4) Epoch 15, batch 2300, loss[ctc_loss=0.07546, att_loss=0.2286, loss=0.198, over 16388.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007561, over 44.00 utterances.], tot_loss[ctc_loss=0.08762, att_loss=0.2433, loss=0.2122, over 3284127.68 frames. utt_duration=1252 frames, utt_pad_proportion=0.04918, over 10503.29 utterances.], batch size: 44, lr: 7.21e-03, grad_scale: 8.0 2023-03-08 12:11:45,861 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58089.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:12:06,428 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58102.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:12:33,031 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58118.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:12:40,614 INFO [train2.py:809] (0/4) Epoch 15, batch 2350, loss[ctc_loss=0.09284, att_loss=0.256, loss=0.2233, over 17441.00 frames. utt_duration=1012 frames, utt_pad_proportion=0.04494, over 69.00 utterances.], tot_loss[ctc_loss=0.08809, att_loss=0.244, loss=0.2128, over 3282909.37 frames. utt_duration=1196 frames, utt_pad_proportion=0.06162, over 10991.05 utterances.], batch size: 69, lr: 7.21e-03, grad_scale: 8.0 2023-03-08 12:12:49,822 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.555e+02 2.243e+02 2.592e+02 3.349e+02 5.777e+02, threshold=5.184e+02, percent-clipped=2.0 2023-03-08 12:13:08,864 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58141.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:13:14,168 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58144.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:13:24,611 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58151.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:13:48,218 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58166.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:13:58,916 INFO [train2.py:809] (0/4) Epoch 15, batch 2400, loss[ctc_loss=0.1112, att_loss=0.2768, loss=0.2437, over 17295.00 frames. utt_duration=1004 frames, utt_pad_proportion=0.05092, over 69.00 utterances.], tot_loss[ctc_loss=0.08889, att_loss=0.2448, loss=0.2136, over 3286591.84 frames. utt_duration=1200 frames, utt_pad_proportion=0.05953, over 10970.49 utterances.], batch size: 69, lr: 7.21e-03, grad_scale: 8.0 2023-03-08 12:14:40,623 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58199.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:14:45,743 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3005, 2.6163, 3.3486, 4.4335, 3.8187, 3.9486, 2.8706, 2.0705], device='cuda:0'), covar=tensor([0.0709, 0.2206, 0.0962, 0.0478, 0.0837, 0.0399, 0.1537, 0.2399], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0213, 0.0190, 0.0200, 0.0208, 0.0165, 0.0198, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 12:14:45,819 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58202.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:14:50,478 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58205.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:15:18,769 INFO [train2.py:809] (0/4) Epoch 15, batch 2450, loss[ctc_loss=0.1029, att_loss=0.2579, loss=0.2269, over 17517.00 frames. utt_duration=1017 frames, utt_pad_proportion=0.04177, over 69.00 utterances.], tot_loss[ctc_loss=0.08846, att_loss=0.2443, loss=0.2131, over 3283003.33 frames. utt_duration=1207 frames, utt_pad_proportion=0.05925, over 10896.05 utterances.], batch size: 69, lr: 7.20e-03, grad_scale: 8.0 2023-03-08 12:15:27,700 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 2.141e+02 2.652e+02 3.416e+02 6.108e+02, threshold=5.305e+02, percent-clipped=4.0 2023-03-08 12:16:38,078 INFO [train2.py:809] (0/4) Epoch 15, batch 2500, loss[ctc_loss=0.08365, att_loss=0.253, loss=0.2192, over 17290.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01189, over 55.00 utterances.], tot_loss[ctc_loss=0.08877, att_loss=0.2439, loss=0.2129, over 3281214.34 frames. utt_duration=1209 frames, utt_pad_proportion=0.0599, over 10873.83 utterances.], batch size: 55, lr: 7.20e-03, grad_scale: 8.0 2023-03-08 12:16:53,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9087, 6.1272, 5.5940, 5.8823, 5.8049, 5.3551, 5.5759, 5.2732], device='cuda:0'), covar=tensor([0.1151, 0.0824, 0.0848, 0.0778, 0.0894, 0.1454, 0.2087, 0.2593], device='cuda:0'), in_proj_covar=tensor([0.0476, 0.0547, 0.0422, 0.0420, 0.0399, 0.0441, 0.0570, 0.0505], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 12:17:57,543 INFO [train2.py:809] (0/4) Epoch 15, batch 2550, loss[ctc_loss=0.1091, att_loss=0.2645, loss=0.2334, over 17173.00 frames. utt_duration=871 frames, utt_pad_proportion=0.08797, over 79.00 utterances.], tot_loss[ctc_loss=0.08803, att_loss=0.2428, loss=0.2118, over 3266314.45 frames. utt_duration=1223 frames, utt_pad_proportion=0.06009, over 10692.65 utterances.], batch size: 79, lr: 7.20e-03, grad_scale: 8.0 2023-03-08 12:18:06,732 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.062e+02 2.631e+02 3.242e+02 5.829e+02, threshold=5.263e+02, percent-clipped=2.0 2023-03-08 12:18:26,987 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6298, 4.4621, 4.5913, 4.6016, 5.1343, 4.7489, 4.6031, 2.4122], device='cuda:0'), covar=tensor([0.0179, 0.0311, 0.0275, 0.0289, 0.0884, 0.0170, 0.0280, 0.2046], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0152, 0.0156, 0.0170, 0.0355, 0.0136, 0.0143, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 12:18:48,872 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4504, 2.1954, 5.0109, 3.6992, 2.8704, 4.2114, 4.7705, 4.6901], device='cuda:0'), covar=tensor([0.0217, 0.1938, 0.0148, 0.0992, 0.1872, 0.0251, 0.0110, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0241, 0.0156, 0.0306, 0.0263, 0.0190, 0.0140, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 12:19:17,177 INFO [train2.py:809] (0/4) Epoch 15, batch 2600, loss[ctc_loss=0.09079, att_loss=0.2378, loss=0.2084, over 16399.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007077, over 44.00 utterances.], tot_loss[ctc_loss=0.08731, att_loss=0.2423, loss=0.2113, over 3261625.93 frames. utt_duration=1243 frames, utt_pad_proportion=0.05673, over 10511.49 utterances.], batch size: 44, lr: 7.19e-03, grad_scale: 8.0 2023-03-08 12:19:42,871 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58389.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:19:53,567 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 12:20:37,375 INFO [train2.py:809] (0/4) Epoch 15, batch 2650, loss[ctc_loss=0.08462, att_loss=0.2565, loss=0.2221, over 17311.00 frames. utt_duration=877.8 frames, utt_pad_proportion=0.07894, over 79.00 utterances.], tot_loss[ctc_loss=0.08843, att_loss=0.2434, loss=0.2124, over 3262852.25 frames. utt_duration=1192 frames, utt_pad_proportion=0.06939, over 10965.21 utterances.], batch size: 79, lr: 7.19e-03, grad_scale: 8.0 2023-03-08 12:20:46,673 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.218e+02 2.453e+02 3.355e+02 6.382e+02, threshold=4.906e+02, percent-clipped=2.0 2023-03-08 12:20:59,516 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58437.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:21:11,661 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3385, 4.7802, 4.6254, 4.8115, 4.8759, 4.4484, 3.1232, 4.6891], device='cuda:0'), covar=tensor([0.0140, 0.0135, 0.0154, 0.0092, 0.0118, 0.0146, 0.0809, 0.0288], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0079, 0.0098, 0.0060, 0.0066, 0.0077, 0.0096, 0.0099], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 12:21:57,757 INFO [train2.py:809] (0/4) Epoch 15, batch 2700, loss[ctc_loss=0.08155, att_loss=0.2281, loss=0.1988, over 16411.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006906, over 44.00 utterances.], tot_loss[ctc_loss=0.08835, att_loss=0.2431, loss=0.2121, over 3259550.93 frames. utt_duration=1197 frames, utt_pad_proportion=0.07017, over 10907.55 utterances.], batch size: 44, lr: 7.19e-03, grad_scale: 8.0 2023-03-08 12:22:19,611 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5121, 2.3077, 4.9249, 3.7242, 2.8320, 4.2133, 4.6474, 4.5861], device='cuda:0'), covar=tensor([0.0172, 0.1976, 0.0129, 0.1020, 0.2039, 0.0249, 0.0108, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0240, 0.0155, 0.0305, 0.0262, 0.0190, 0.0139, 0.0169], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 12:22:36,282 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58497.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:22:41,397 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=58500.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:22:44,629 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6174, 3.5800, 3.4985, 3.1981, 3.6227, 3.5784, 3.6174, 2.6413], device='cuda:0'), covar=tensor([0.0946, 0.1221, 0.2737, 0.3527, 0.1317, 0.2981, 0.0857, 0.4867], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0149, 0.0160, 0.0226, 0.0120, 0.0216, 0.0135, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 12:23:17,021 INFO [train2.py:809] (0/4) Epoch 15, batch 2750, loss[ctc_loss=0.1234, att_loss=0.2682, loss=0.2392, over 13868.00 frames. utt_duration=384 frames, utt_pad_proportion=0.3333, over 145.00 utterances.], tot_loss[ctc_loss=0.08902, att_loss=0.2433, loss=0.2124, over 3258357.55 frames. utt_duration=1197 frames, utt_pad_proportion=0.07216, over 10903.27 utterances.], batch size: 145, lr: 7.18e-03, grad_scale: 8.0 2023-03-08 12:23:26,923 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 2.307e+02 2.848e+02 3.393e+02 6.929e+02, threshold=5.697e+02, percent-clipped=6.0 2023-03-08 12:24:35,856 INFO [train2.py:809] (0/4) Epoch 15, batch 2800, loss[ctc_loss=0.06418, att_loss=0.2203, loss=0.1891, over 16025.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006765, over 40.00 utterances.], tot_loss[ctc_loss=0.0886, att_loss=0.2435, loss=0.2125, over 3266044.45 frames. utt_duration=1208 frames, utt_pad_proportion=0.06747, over 10823.78 utterances.], batch size: 40, lr: 7.18e-03, grad_scale: 8.0 2023-03-08 12:25:16,214 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-08 12:25:53,744 INFO [train2.py:809] (0/4) Epoch 15, batch 2850, loss[ctc_loss=0.1207, att_loss=0.2657, loss=0.2367, over 16470.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006267, over 46.00 utterances.], tot_loss[ctc_loss=0.08813, att_loss=0.2428, loss=0.2119, over 3269646.56 frames. utt_duration=1227 frames, utt_pad_proportion=0.06206, over 10675.20 utterances.], batch size: 46, lr: 7.18e-03, grad_scale: 8.0 2023-03-08 12:26:02,953 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.515e+02 2.157e+02 2.674e+02 3.074e+02 6.924e+02, threshold=5.347e+02, percent-clipped=2.0 2023-03-08 12:27:08,290 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-03-08 12:27:12,212 INFO [train2.py:809] (0/4) Epoch 15, batch 2900, loss[ctc_loss=0.09943, att_loss=0.2495, loss=0.2195, over 16542.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005375, over 45.00 utterances.], tot_loss[ctc_loss=0.08857, att_loss=0.2433, loss=0.2124, over 3268805.13 frames. utt_duration=1207 frames, utt_pad_proportion=0.06744, over 10845.77 utterances.], batch size: 45, lr: 7.18e-03, grad_scale: 8.0 2023-03-08 12:27:12,431 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7964, 5.1224, 5.3704, 5.1960, 5.2180, 5.7059, 5.1342, 5.8156], device='cuda:0'), covar=tensor([0.0721, 0.0830, 0.0728, 0.1182, 0.1871, 0.0982, 0.0761, 0.0694], device='cuda:0'), in_proj_covar=tensor([0.0791, 0.0468, 0.0546, 0.0602, 0.0798, 0.0557, 0.0445, 0.0531], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 12:27:26,469 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58681.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:28:31,198 INFO [train2.py:809] (0/4) Epoch 15, batch 2950, loss[ctc_loss=0.0726, att_loss=0.2385, loss=0.2053, over 16390.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.008145, over 44.00 utterances.], tot_loss[ctc_loss=0.08855, att_loss=0.2438, loss=0.2127, over 3276158.58 frames. utt_duration=1235 frames, utt_pad_proportion=0.05843, over 10620.52 utterances.], batch size: 44, lr: 7.17e-03, grad_scale: 8.0 2023-03-08 12:28:41,001 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.392e+02 2.063e+02 2.403e+02 2.955e+02 5.605e+02, threshold=4.806e+02, percent-clipped=1.0 2023-03-08 12:28:51,010 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7574, 5.9993, 5.4094, 5.7527, 5.6249, 5.1942, 5.3542, 5.2486], device='cuda:0'), covar=tensor([0.1128, 0.0833, 0.0818, 0.0712, 0.0926, 0.1440, 0.2276, 0.2115], device='cuda:0'), in_proj_covar=tensor([0.0473, 0.0547, 0.0416, 0.0414, 0.0396, 0.0434, 0.0562, 0.0495], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 12:29:01,917 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58742.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:29:28,964 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5117, 4.5035, 4.4396, 4.3969, 5.1397, 4.5138, 4.4816, 2.4070], device='cuda:0'), covar=tensor([0.0236, 0.0293, 0.0279, 0.0341, 0.0723, 0.0232, 0.0312, 0.2030], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0150, 0.0154, 0.0168, 0.0351, 0.0136, 0.0143, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 12:29:49,918 INFO [train2.py:809] (0/4) Epoch 15, batch 3000, loss[ctc_loss=0.1313, att_loss=0.2647, loss=0.238, over 14033.00 frames. utt_duration=388.5 frames, utt_pad_proportion=0.3244, over 145.00 utterances.], tot_loss[ctc_loss=0.08798, att_loss=0.2433, loss=0.2122, over 3277776.39 frames. utt_duration=1241 frames, utt_pad_proportion=0.05734, over 10580.61 utterances.], batch size: 145, lr: 7.17e-03, grad_scale: 8.0 2023-03-08 12:29:49,920 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 12:30:03,571 INFO [train2.py:843] (0/4) Epoch 15, validation: ctc_loss=0.04336, att_loss=0.2362, loss=0.1976, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 12:30:03,572 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 12:30:07,139 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58775.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:30:07,761 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.14 vs. limit=2.0 2023-03-08 12:30:16,331 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2647, 2.5023, 2.8674, 4.3051, 3.7622, 3.6677, 2.7530, 1.9219], device='cuda:0'), covar=tensor([0.0780, 0.2256, 0.1209, 0.0523, 0.0928, 0.0587, 0.1659, 0.2478], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0211, 0.0187, 0.0198, 0.0206, 0.0168, 0.0196, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 12:30:34,775 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2286, 4.6783, 4.5380, 4.7020, 4.7547, 4.4267, 3.2139, 4.5269], device='cuda:0'), covar=tensor([0.0125, 0.0111, 0.0136, 0.0079, 0.0097, 0.0128, 0.0689, 0.0214], device='cuda:0'), in_proj_covar=tensor([0.0081, 0.0078, 0.0097, 0.0060, 0.0066, 0.0077, 0.0095, 0.0098], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 12:30:42,525 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58797.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:30:47,650 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=58800.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:31:24,434 INFO [train2.py:809] (0/4) Epoch 15, batch 3050, loss[ctc_loss=0.1001, att_loss=0.25, loss=0.22, over 17272.00 frames. utt_duration=876.3 frames, utt_pad_proportion=0.0834, over 79.00 utterances.], tot_loss[ctc_loss=0.08775, att_loss=0.2432, loss=0.2121, over 3283573.23 frames. utt_duration=1241 frames, utt_pad_proportion=0.05458, over 10595.87 utterances.], batch size: 79, lr: 7.17e-03, grad_scale: 8.0 2023-03-08 12:31:24,636 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0376, 5.3051, 5.2838, 5.2704, 5.3017, 5.3195, 5.0116, 4.7672], device='cuda:0'), covar=tensor([0.1038, 0.0533, 0.0227, 0.0443, 0.0303, 0.0296, 0.0327, 0.0342], device='cuda:0'), in_proj_covar=tensor([0.0480, 0.0324, 0.0293, 0.0320, 0.0378, 0.0396, 0.0319, 0.0361], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 12:31:34,016 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.376e+02 2.164e+02 2.584e+02 3.141e+02 6.750e+02, threshold=5.168e+02, percent-clipped=6.0 2023-03-08 12:31:34,187 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7677, 5.9991, 5.4020, 5.7529, 5.6937, 5.1817, 5.3597, 5.1565], device='cuda:0'), covar=tensor([0.1217, 0.0899, 0.0915, 0.0831, 0.1017, 0.1522, 0.2352, 0.2418], device='cuda:0'), in_proj_covar=tensor([0.0473, 0.0551, 0.0418, 0.0416, 0.0398, 0.0434, 0.0564, 0.0500], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 12:31:45,905 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58836.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:31:59,390 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58845.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:32:04,251 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=58848.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:32:26,767 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2482, 3.7509, 3.3138, 3.5211, 4.0067, 3.6366, 2.9831, 4.3044], device='cuda:0'), covar=tensor([0.0853, 0.0478, 0.0970, 0.0652, 0.0669, 0.0651, 0.0821, 0.0455], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0199, 0.0215, 0.0188, 0.0259, 0.0228, 0.0190, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 12:32:43,305 INFO [train2.py:809] (0/4) Epoch 15, batch 3100, loss[ctc_loss=0.08679, att_loss=0.2223, loss=0.1952, over 15768.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008809, over 38.00 utterances.], tot_loss[ctc_loss=0.08841, att_loss=0.2431, loss=0.2122, over 3278407.68 frames. utt_duration=1229 frames, utt_pad_proportion=0.05934, over 10680.38 utterances.], batch size: 38, lr: 7.16e-03, grad_scale: 8.0 2023-03-08 12:33:34,865 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=58906.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:33:52,311 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2023-03-08 12:34:00,654 INFO [train2.py:809] (0/4) Epoch 15, batch 3150, loss[ctc_loss=0.08148, att_loss=0.2323, loss=0.2021, over 16105.00 frames. utt_duration=1535 frames, utt_pad_proportion=0.007526, over 42.00 utterances.], tot_loss[ctc_loss=0.08779, att_loss=0.2425, loss=0.2115, over 3275588.96 frames. utt_duration=1240 frames, utt_pad_proportion=0.05641, over 10577.38 utterances.], batch size: 42, lr: 7.16e-03, grad_scale: 8.0 2023-03-08 12:34:09,777 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 2.265e+02 2.592e+02 3.060e+02 7.077e+02, threshold=5.184e+02, percent-clipped=3.0 2023-03-08 12:35:03,194 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-03-08 12:35:10,500 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=58967.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:35:16,637 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2493, 2.5918, 3.0777, 4.2259, 3.6405, 3.7018, 2.7059, 1.9798], device='cuda:0'), covar=tensor([0.0714, 0.2264, 0.1011, 0.0515, 0.0883, 0.0524, 0.1674, 0.2471], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0211, 0.0186, 0.0198, 0.0206, 0.0167, 0.0197, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 12:35:19,406 INFO [train2.py:809] (0/4) Epoch 15, batch 3200, loss[ctc_loss=0.05889, att_loss=0.2216, loss=0.1891, over 16264.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.008026, over 43.00 utterances.], tot_loss[ctc_loss=0.08647, att_loss=0.2416, loss=0.2106, over 3268011.70 frames. utt_duration=1254 frames, utt_pad_proportion=0.05545, over 10439.30 utterances.], batch size: 43, lr: 7.16e-03, grad_scale: 8.0 2023-03-08 12:36:38,220 INFO [train2.py:809] (0/4) Epoch 15, batch 3250, loss[ctc_loss=0.0644, att_loss=0.2393, loss=0.2043, over 16408.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006684, over 44.00 utterances.], tot_loss[ctc_loss=0.08621, att_loss=0.2415, loss=0.2104, over 3269870.90 frames. utt_duration=1265 frames, utt_pad_proportion=0.05065, over 10354.85 utterances.], batch size: 44, lr: 7.15e-03, grad_scale: 8.0 2023-03-08 12:36:48,043 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 2.216e+02 2.757e+02 3.407e+02 8.183e+02, threshold=5.513e+02, percent-clipped=5.0 2023-03-08 12:37:00,977 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59037.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:37:57,612 INFO [train2.py:809] (0/4) Epoch 15, batch 3300, loss[ctc_loss=0.1136, att_loss=0.2612, loss=0.2317, over 17409.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.0474, over 69.00 utterances.], tot_loss[ctc_loss=0.08674, att_loss=0.2421, loss=0.211, over 3265992.50 frames. utt_duration=1252 frames, utt_pad_proportion=0.05454, over 10445.76 utterances.], batch size: 69, lr: 7.15e-03, grad_scale: 8.0 2023-03-08 12:38:18,399 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-03-08 12:39:16,673 INFO [train2.py:809] (0/4) Epoch 15, batch 3350, loss[ctc_loss=0.1204, att_loss=0.2635, loss=0.2349, over 17209.00 frames. utt_duration=1094 frames, utt_pad_proportion=0.04375, over 63.00 utterances.], tot_loss[ctc_loss=0.08642, att_loss=0.2422, loss=0.211, over 3269015.81 frames. utt_duration=1250 frames, utt_pad_proportion=0.05355, over 10473.02 utterances.], batch size: 63, lr: 7.15e-03, grad_scale: 8.0 2023-03-08 12:39:27,190 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.236e+02 1.917e+02 2.406e+02 2.955e+02 7.524e+02, threshold=4.811e+02, percent-clipped=3.0 2023-03-08 12:39:30,485 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59131.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:40:22,560 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59164.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:40:36,060 INFO [train2.py:809] (0/4) Epoch 15, batch 3400, loss[ctc_loss=0.1326, att_loss=0.2668, loss=0.2399, over 17520.00 frames. utt_duration=1114 frames, utt_pad_proportion=0.02714, over 63.00 utterances.], tot_loss[ctc_loss=0.08671, att_loss=0.2428, loss=0.2116, over 3283000.54 frames. utt_duration=1253 frames, utt_pad_proportion=0.04858, over 10493.44 utterances.], batch size: 63, lr: 7.15e-03, grad_scale: 8.0 2023-03-08 12:41:27,195 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 12:41:56,069 INFO [train2.py:809] (0/4) Epoch 15, batch 3450, loss[ctc_loss=0.0854, att_loss=0.2472, loss=0.2148, over 17326.00 frames. utt_duration=1006 frames, utt_pad_proportion=0.05116, over 69.00 utterances.], tot_loss[ctc_loss=0.08639, att_loss=0.2426, loss=0.2113, over 3287576.93 frames. utt_duration=1261 frames, utt_pad_proportion=0.04496, over 10438.19 utterances.], batch size: 69, lr: 7.14e-03, grad_scale: 8.0 2023-03-08 12:42:00,124 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59225.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:42:06,838 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.536e+02 2.401e+02 2.877e+02 3.602e+02 6.308e+02, threshold=5.755e+02, percent-clipped=6.0 2023-03-08 12:42:58,725 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59262.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:43:16,319 INFO [train2.py:809] (0/4) Epoch 15, batch 3500, loss[ctc_loss=0.09271, att_loss=0.2527, loss=0.2207, over 17013.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.00821, over 51.00 utterances.], tot_loss[ctc_loss=0.08677, att_loss=0.2432, loss=0.2119, over 3287369.04 frames. utt_duration=1255 frames, utt_pad_proportion=0.04745, over 10494.16 utterances.], batch size: 51, lr: 7.14e-03, grad_scale: 8.0 2023-03-08 12:44:35,592 INFO [train2.py:809] (0/4) Epoch 15, batch 3550, loss[ctc_loss=0.1054, att_loss=0.2713, loss=0.2381, over 17364.00 frames. utt_duration=1179 frames, utt_pad_proportion=0.02103, over 59.00 utterances.], tot_loss[ctc_loss=0.08687, att_loss=0.2435, loss=0.2122, over 3285023.49 frames. utt_duration=1255 frames, utt_pad_proportion=0.04945, over 10483.34 utterances.], batch size: 59, lr: 7.14e-03, grad_scale: 8.0 2023-03-08 12:44:45,171 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 2.222e+02 2.629e+02 3.391e+02 5.578e+02, threshold=5.259e+02, percent-clipped=0.0 2023-03-08 12:44:57,751 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59337.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:45:54,580 INFO [train2.py:809] (0/4) Epoch 15, batch 3600, loss[ctc_loss=0.07133, att_loss=0.2399, loss=0.2062, over 17012.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008297, over 51.00 utterances.], tot_loss[ctc_loss=0.08784, att_loss=0.2436, loss=0.2125, over 3279836.97 frames. utt_duration=1232 frames, utt_pad_proportion=0.05715, over 10659.42 utterances.], batch size: 51, lr: 7.13e-03, grad_scale: 8.0 2023-03-08 12:46:13,883 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59385.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:47:15,177 INFO [train2.py:809] (0/4) Epoch 15, batch 3650, loss[ctc_loss=0.1102, att_loss=0.268, loss=0.2364, over 17065.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.008215, over 53.00 utterances.], tot_loss[ctc_loss=0.08778, att_loss=0.2442, loss=0.2129, over 3279009.07 frames. utt_duration=1222 frames, utt_pad_proportion=0.05711, over 10748.39 utterances.], batch size: 53, lr: 7.13e-03, grad_scale: 8.0 2023-03-08 12:47:24,351 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 2.155e+02 2.711e+02 3.466e+02 9.955e+02, threshold=5.422e+02, percent-clipped=4.0 2023-03-08 12:47:27,694 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59431.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:48:08,199 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-03-08 12:48:34,104 INFO [train2.py:809] (0/4) Epoch 15, batch 3700, loss[ctc_loss=0.09874, att_loss=0.249, loss=0.2189, over 17383.00 frames. utt_duration=881.7 frames, utt_pad_proportion=0.07582, over 79.00 utterances.], tot_loss[ctc_loss=0.08757, att_loss=0.2435, loss=0.2123, over 3269095.78 frames. utt_duration=1203 frames, utt_pad_proportion=0.06446, over 10885.39 utterances.], batch size: 79, lr: 7.13e-03, grad_scale: 8.0 2023-03-08 12:48:43,279 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59479.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:49:10,586 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1308, 4.9457, 4.9979, 2.0990, 1.8827, 2.8895, 2.8018, 3.7955], device='cuda:0'), covar=tensor([0.0685, 0.0229, 0.0196, 0.4617, 0.5893, 0.2520, 0.2543, 0.1731], device='cuda:0'), in_proj_covar=tensor([0.0343, 0.0242, 0.0247, 0.0227, 0.0343, 0.0335, 0.0234, 0.0357], device='cuda:0'), out_proj_covar=tensor([1.4932e-04, 9.0375e-05, 1.0656e-04, 9.8423e-05, 1.4590e-04, 1.3287e-04, 9.3634e-05, 1.4743e-04], device='cuda:0') 2023-03-08 12:49:20,184 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59502.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:49:49,037 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59520.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:49:53,482 INFO [train2.py:809] (0/4) Epoch 15, batch 3750, loss[ctc_loss=0.09949, att_loss=0.2558, loss=0.2245, over 17409.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03159, over 63.00 utterances.], tot_loss[ctc_loss=0.08658, att_loss=0.2426, loss=0.2114, over 3269916.61 frames. utt_duration=1218 frames, utt_pad_proportion=0.06058, over 10749.61 utterances.], batch size: 63, lr: 7.12e-03, grad_scale: 8.0 2023-03-08 12:49:54,198 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.57 vs. limit=5.0 2023-03-08 12:49:55,378 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1803, 4.5977, 4.6892, 4.8074, 3.0057, 4.8240, 2.5486, 2.1066], device='cuda:0'), covar=tensor([0.0388, 0.0189, 0.0578, 0.0158, 0.1461, 0.0150, 0.1472, 0.1567], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0134, 0.0245, 0.0125, 0.0218, 0.0117, 0.0221, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 12:49:59,274 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-08 12:50:02,433 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 2.043e+02 2.443e+02 3.044e+02 5.040e+02, threshold=4.885e+02, percent-clipped=0.0 2023-03-08 12:50:54,599 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59562.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:50:56,203 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59563.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:51:12,185 INFO [train2.py:809] (0/4) Epoch 15, batch 3800, loss[ctc_loss=0.06777, att_loss=0.2138, loss=0.1846, over 16301.00 frames. utt_duration=1518 frames, utt_pad_proportion=0.006183, over 43.00 utterances.], tot_loss[ctc_loss=0.08578, att_loss=0.2415, loss=0.2104, over 3272521.49 frames. utt_duration=1233 frames, utt_pad_proportion=0.05529, over 10625.69 utterances.], batch size: 43, lr: 7.12e-03, grad_scale: 8.0 2023-03-08 12:52:09,973 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59610.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:52:31,308 INFO [train2.py:809] (0/4) Epoch 15, batch 3850, loss[ctc_loss=0.07101, att_loss=0.2291, loss=0.1975, over 15952.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.007322, over 41.00 utterances.], tot_loss[ctc_loss=0.08628, att_loss=0.2419, loss=0.2108, over 3267204.06 frames. utt_duration=1214 frames, utt_pad_proportion=0.06293, over 10780.83 utterances.], batch size: 41, lr: 7.12e-03, grad_scale: 8.0 2023-03-08 12:52:40,310 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.434e+02 2.221e+02 2.684e+02 3.228e+02 7.729e+02, threshold=5.369e+02, percent-clipped=7.0 2023-03-08 12:52:54,136 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9055, 3.4296, 4.0696, 3.7450, 4.0946, 4.8740, 4.7189, 3.8767], device='cuda:0'), covar=tensor([0.0236, 0.1313, 0.0822, 0.0898, 0.0696, 0.0682, 0.0409, 0.0860], device='cuda:0'), in_proj_covar=tensor([0.0235, 0.0233, 0.0263, 0.0208, 0.0252, 0.0331, 0.0236, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 12:53:21,708 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5836, 4.6662, 4.5291, 4.5213, 5.3051, 4.7393, 4.6345, 2.4290], device='cuda:0'), covar=tensor([0.0241, 0.0287, 0.0313, 0.0324, 0.0809, 0.0203, 0.0300, 0.1916], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0148, 0.0154, 0.0167, 0.0345, 0.0134, 0.0141, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 12:53:46,841 INFO [train2.py:809] (0/4) Epoch 15, batch 3900, loss[ctc_loss=0.07941, att_loss=0.24, loss=0.2079, over 16178.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006883, over 41.00 utterances.], tot_loss[ctc_loss=0.08671, att_loss=0.2422, loss=0.2111, over 3265526.57 frames. utt_duration=1212 frames, utt_pad_proportion=0.06178, over 10792.44 utterances.], batch size: 41, lr: 7.12e-03, grad_scale: 8.0 2023-03-08 12:54:35,046 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0090, 5.2712, 4.8410, 5.3421, 4.7420, 4.9345, 5.4105, 5.2227], device='cuda:0'), covar=tensor([0.0576, 0.0294, 0.0845, 0.0280, 0.0434, 0.0279, 0.0232, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0366, 0.0291, 0.0344, 0.0306, 0.0297, 0.0225, 0.0277, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 12:55:03,846 INFO [train2.py:809] (0/4) Epoch 15, batch 3950, loss[ctc_loss=0.05609, att_loss=0.1991, loss=0.1705, over 15658.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.00741, over 37.00 utterances.], tot_loss[ctc_loss=0.08672, att_loss=0.2428, loss=0.2116, over 3273951.95 frames. utt_duration=1205 frames, utt_pad_proportion=0.06112, over 10880.52 utterances.], batch size: 37, lr: 7.11e-03, grad_scale: 8.0 2023-03-08 12:55:12,749 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 1.994e+02 2.418e+02 2.996e+02 5.766e+02, threshold=4.836e+02, percent-clipped=1.0 2023-03-08 12:55:54,899 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-15.pt 2023-03-08 12:56:20,781 INFO [train2.py:809] (0/4) Epoch 16, batch 0, loss[ctc_loss=0.08023, att_loss=0.2503, loss=0.2163, over 17102.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.0152, over 56.00 utterances.], tot_loss[ctc_loss=0.08023, att_loss=0.2503, loss=0.2163, over 17102.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.0152, over 56.00 utterances.], batch size: 56, lr: 6.88e-03, grad_scale: 8.0 2023-03-08 12:56:20,783 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 12:56:32,579 INFO [train2.py:843] (0/4) Epoch 16, validation: ctc_loss=0.04399, att_loss=0.2363, loss=0.1978, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 12:56:32,580 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 12:57:52,425 INFO [train2.py:809] (0/4) Epoch 16, batch 50, loss[ctc_loss=0.08105, att_loss=0.2482, loss=0.2148, over 16964.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007845, over 50.00 utterances.], tot_loss[ctc_loss=0.08911, att_loss=0.2433, loss=0.2124, over 728216.31 frames. utt_duration=1199 frames, utt_pad_proportion=0.07623, over 2431.97 utterances.], batch size: 50, lr: 6.88e-03, grad_scale: 8.0 2023-03-08 12:58:13,603 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=59820.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:58:27,785 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.394e+02 2.112e+02 2.420e+02 3.068e+02 7.828e+02, threshold=4.840e+02, percent-clipped=6.0 2023-03-08 12:58:59,249 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59848.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:59:12,663 INFO [train2.py:809] (0/4) Epoch 16, batch 100, loss[ctc_loss=0.07178, att_loss=0.2466, loss=0.2116, over 16991.00 frames. utt_duration=1361 frames, utt_pad_proportion=0.006019, over 50.00 utterances.], tot_loss[ctc_loss=0.08654, att_loss=0.2422, loss=0.2111, over 1287960.80 frames. utt_duration=1210 frames, utt_pad_proportion=0.07258, over 4261.37 utterances.], batch size: 50, lr: 6.88e-03, grad_scale: 8.0 2023-03-08 12:59:14,413 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=59858.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 12:59:30,309 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=59868.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:00:03,275 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=59889.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:00:31,894 INFO [train2.py:809] (0/4) Epoch 16, batch 150, loss[ctc_loss=0.1038, att_loss=0.2554, loss=0.225, over 16477.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006639, over 46.00 utterances.], tot_loss[ctc_loss=0.08578, att_loss=0.2418, loss=0.2106, over 1730467.37 frames. utt_duration=1244 frames, utt_pad_proportion=0.06093, over 5573.13 utterances.], batch size: 46, lr: 6.87e-03, grad_scale: 8.0 2023-03-08 13:00:35,425 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59909.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:01:06,714 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.476e+02 2.211e+02 2.747e+02 3.339e+02 6.489e+02, threshold=5.494e+02, percent-clipped=4.0 2023-03-08 13:01:19,421 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0812, 4.4440, 4.3356, 4.5076, 2.5851, 4.3838, 2.5631, 1.6626], device='cuda:0'), covar=tensor([0.0363, 0.0193, 0.0727, 0.0179, 0.1857, 0.0184, 0.1561, 0.1814], device='cuda:0'), in_proj_covar=tensor([0.0160, 0.0135, 0.0246, 0.0125, 0.0217, 0.0117, 0.0221, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 13:01:40,795 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=59950.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:01:51,216 INFO [train2.py:809] (0/4) Epoch 16, batch 200, loss[ctc_loss=0.09416, att_loss=0.2569, loss=0.2244, over 17318.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.01107, over 55.00 utterances.], tot_loss[ctc_loss=0.08502, att_loss=0.2412, loss=0.21, over 2075654.97 frames. utt_duration=1261 frames, utt_pad_proportion=0.05331, over 6594.56 utterances.], batch size: 55, lr: 6.87e-03, grad_scale: 8.0 2023-03-08 13:03:00,460 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-60000.pt 2023-03-08 13:03:15,166 INFO [train2.py:809] (0/4) Epoch 16, batch 250, loss[ctc_loss=0.1173, att_loss=0.246, loss=0.2203, over 16525.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006366, over 45.00 utterances.], tot_loss[ctc_loss=0.08465, att_loss=0.2411, loss=0.2098, over 2345845.61 frames. utt_duration=1269 frames, utt_pad_proportion=0.04953, over 7401.33 utterances.], batch size: 45, lr: 6.87e-03, grad_scale: 8.0 2023-03-08 13:03:50,522 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.435e+02 2.138e+02 2.534e+02 3.199e+02 6.195e+02, threshold=5.067e+02, percent-clipped=2.0 2023-03-08 13:03:55,631 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6444, 3.6187, 2.9800, 3.2594, 3.8517, 3.4462, 2.6551, 4.0106], device='cuda:0'), covar=tensor([0.1294, 0.0496, 0.1243, 0.0814, 0.0763, 0.0728, 0.1125, 0.0595], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0201, 0.0217, 0.0190, 0.0260, 0.0230, 0.0194, 0.0277], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 13:04:09,032 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60040.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:04:28,356 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60052.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:04:35,587 INFO [train2.py:809] (0/4) Epoch 16, batch 300, loss[ctc_loss=0.09813, att_loss=0.2519, loss=0.2212, over 16629.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005274, over 47.00 utterances.], tot_loss[ctc_loss=0.08528, att_loss=0.2425, loss=0.211, over 2558447.96 frames. utt_duration=1241 frames, utt_pad_proportion=0.05314, over 8258.45 utterances.], batch size: 47, lr: 6.87e-03, grad_scale: 16.0 2023-03-08 13:04:49,137 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60065.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:04:58,080 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8365, 5.1122, 4.6190, 5.1627, 4.5919, 4.8166, 5.2425, 5.0645], device='cuda:0'), covar=tensor([0.0569, 0.0298, 0.0843, 0.0283, 0.0475, 0.0297, 0.0261, 0.0189], device='cuda:0'), in_proj_covar=tensor([0.0366, 0.0290, 0.0342, 0.0303, 0.0298, 0.0224, 0.0277, 0.0259], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 13:05:09,353 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60078.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:05:46,233 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60101.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 13:05:55,886 INFO [train2.py:809] (0/4) Epoch 16, batch 350, loss[ctc_loss=0.119, att_loss=0.2656, loss=0.2363, over 17314.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02302, over 59.00 utterances.], tot_loss[ctc_loss=0.08622, att_loss=0.2429, loss=0.2116, over 2719484.98 frames. utt_duration=1243 frames, utt_pad_proportion=0.05307, over 8764.02 utterances.], batch size: 59, lr: 6.86e-03, grad_scale: 16.0 2023-03-08 13:06:05,683 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60113.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 13:06:26,281 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60126.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:06:30,555 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 2.169e+02 2.724e+02 3.771e+02 9.216e+02, threshold=5.447e+02, percent-clipped=7.0 2023-03-08 13:06:38,909 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9555, 2.1485, 2.6601, 2.6294, 2.7096, 2.7671, 2.4036, 3.3673], device='cuda:0'), covar=tensor([0.1573, 0.3182, 0.2086, 0.1554, 0.2222, 0.1154, 0.2931, 0.0662], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0092, 0.0095, 0.0084, 0.0087, 0.0078, 0.0098, 0.0068], device='cuda:0'), out_proj_covar=tensor([6.3362e-05, 6.9220e-05, 7.2220e-05, 6.2664e-05, 6.2976e-05, 6.1197e-05, 7.1326e-05, 5.4219e-05], device='cuda:0') 2023-03-08 13:06:47,166 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60139.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:06:48,314 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2023-03-08 13:07:14,371 INFO [train2.py:809] (0/4) Epoch 16, batch 400, loss[ctc_loss=0.06802, att_loss=0.2104, loss=0.1819, over 15493.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.008028, over 36.00 utterances.], tot_loss[ctc_loss=0.08638, att_loss=0.2414, loss=0.2104, over 2833839.58 frames. utt_duration=1245 frames, utt_pad_proportion=0.05472, over 9116.74 utterances.], batch size: 36, lr: 6.86e-03, grad_scale: 8.0 2023-03-08 13:07:16,782 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60158.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:07:59,366 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1554, 5.3168, 5.6158, 5.5070, 5.6150, 6.0689, 5.2495, 6.1527], device='cuda:0'), covar=tensor([0.0669, 0.0740, 0.0799, 0.1309, 0.1880, 0.0845, 0.0620, 0.0695], device='cuda:0'), in_proj_covar=tensor([0.0797, 0.0475, 0.0554, 0.0614, 0.0819, 0.0565, 0.0454, 0.0546], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 13:08:19,846 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60198.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:08:29,145 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60204.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:08:32,164 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60206.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:08:33,620 INFO [train2.py:809] (0/4) Epoch 16, batch 450, loss[ctc_loss=0.07624, att_loss=0.2439, loss=0.2103, over 16483.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.00571, over 46.00 utterances.], tot_loss[ctc_loss=0.08623, att_loss=0.2415, loss=0.2104, over 2930277.15 frames. utt_duration=1262 frames, utt_pad_proportion=0.05092, over 9300.47 utterances.], batch size: 46, lr: 6.86e-03, grad_scale: 8.0 2023-03-08 13:08:53,866 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1094, 5.4131, 4.7554, 5.2292, 5.0330, 4.5822, 4.8423, 4.6627], device='cuda:0'), covar=tensor([0.1347, 0.1009, 0.0922, 0.0891, 0.1010, 0.1597, 0.2304, 0.2337], device='cuda:0'), in_proj_covar=tensor([0.0482, 0.0556, 0.0423, 0.0423, 0.0401, 0.0440, 0.0569, 0.0500], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 13:09:10,571 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.357e+02 2.196e+02 2.695e+02 3.219e+02 5.768e+02, threshold=5.391e+02, percent-clipped=1.0 2023-03-08 13:09:20,771 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9208, 5.2223, 4.7953, 5.2959, 4.6623, 4.9336, 5.3885, 5.2136], device='cuda:0'), covar=tensor([0.0587, 0.0348, 0.0803, 0.0279, 0.0459, 0.0245, 0.0226, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0365, 0.0291, 0.0343, 0.0302, 0.0299, 0.0225, 0.0278, 0.0261], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 13:09:34,703 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60245.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:09:54,249 INFO [train2.py:809] (0/4) Epoch 16, batch 500, loss[ctc_loss=0.08023, att_loss=0.2409, loss=0.2088, over 16763.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006836, over 48.00 utterances.], tot_loss[ctc_loss=0.08698, att_loss=0.2429, loss=0.2117, over 3008103.04 frames. utt_duration=1207 frames, utt_pad_proportion=0.06398, over 9978.32 utterances.], batch size: 48, lr: 6.85e-03, grad_scale: 8.0 2023-03-08 13:09:57,636 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60259.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:10:01,500 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-08 13:10:20,079 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 13:11:01,302 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0847, 4.5051, 4.6699, 4.8297, 3.0308, 4.5225, 2.8423, 2.2355], device='cuda:0'), covar=tensor([0.0368, 0.0215, 0.0564, 0.0139, 0.1534, 0.0197, 0.1432, 0.1522], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0137, 0.0251, 0.0127, 0.0218, 0.0120, 0.0223, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 13:11:04,424 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1845, 2.6150, 3.4761, 2.6281, 3.3478, 4.2950, 4.0804, 2.7881], device='cuda:0'), covar=tensor([0.0472, 0.2143, 0.1278, 0.1659, 0.1207, 0.1091, 0.0711, 0.1683], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0236, 0.0264, 0.0209, 0.0253, 0.0336, 0.0240, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 13:11:14,954 INFO [train2.py:809] (0/4) Epoch 16, batch 550, loss[ctc_loss=0.08396, att_loss=0.2335, loss=0.2036, over 15902.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008319, over 39.00 utterances.], tot_loss[ctc_loss=0.08711, att_loss=0.2429, loss=0.2117, over 3069465.39 frames. utt_duration=1211 frames, utt_pad_proportion=0.06185, over 10147.52 utterances.], batch size: 39, lr: 6.85e-03, grad_scale: 8.0 2023-03-08 13:11:15,407 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7359, 2.6643, 5.1364, 4.2581, 3.1265, 4.4181, 4.9110, 4.7207], device='cuda:0'), covar=tensor([0.0244, 0.1579, 0.0194, 0.0807, 0.1664, 0.0208, 0.0119, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0237, 0.0158, 0.0304, 0.0260, 0.0191, 0.0140, 0.0170], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0001, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 13:11:51,582 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.455e+02 1.986e+02 2.421e+02 3.099e+02 6.630e+02, threshold=4.842e+02, percent-clipped=3.0 2023-03-08 13:12:13,516 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2695, 2.6206, 3.0797, 4.3706, 3.8547, 3.7952, 2.7567, 2.1057], device='cuda:0'), covar=tensor([0.0762, 0.2117, 0.1178, 0.0546, 0.0800, 0.0422, 0.1545, 0.2319], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0214, 0.0190, 0.0201, 0.0210, 0.0168, 0.0197, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 13:12:35,535 INFO [train2.py:809] (0/4) Epoch 16, batch 600, loss[ctc_loss=0.05672, att_loss=0.2287, loss=0.1943, over 16391.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007394, over 44.00 utterances.], tot_loss[ctc_loss=0.08674, att_loss=0.2419, loss=0.2109, over 3113590.94 frames. utt_duration=1218 frames, utt_pad_proportion=0.0613, over 10239.66 utterances.], batch size: 44, lr: 6.85e-03, grad_scale: 8.0 2023-03-08 13:12:59,148 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.47 vs. limit=5.0 2023-03-08 13:13:39,228 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60396.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 13:13:57,393 INFO [train2.py:809] (0/4) Epoch 16, batch 650, loss[ctc_loss=0.07958, att_loss=0.2407, loss=0.2085, over 17049.00 frames. utt_duration=1339 frames, utt_pad_proportion=0.006201, over 51.00 utterances.], tot_loss[ctc_loss=0.08673, att_loss=0.242, loss=0.2109, over 3153716.68 frames. utt_duration=1225 frames, utt_pad_proportion=0.05784, over 10306.52 utterances.], batch size: 51, lr: 6.85e-03, grad_scale: 8.0 2023-03-08 13:13:59,124 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60408.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 13:14:20,190 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60421.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:14:33,814 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.189e+02 2.670e+02 3.387e+02 7.027e+02, threshold=5.339e+02, percent-clipped=7.0 2023-03-08 13:14:40,829 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60434.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:15:17,209 INFO [train2.py:809] (0/4) Epoch 16, batch 700, loss[ctc_loss=0.07407, att_loss=0.2338, loss=0.2018, over 16184.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.005998, over 41.00 utterances.], tot_loss[ctc_loss=0.08667, att_loss=0.2422, loss=0.2111, over 3184134.15 frames. utt_duration=1244 frames, utt_pad_proportion=0.05244, over 10254.29 utterances.], batch size: 41, lr: 6.84e-03, grad_scale: 8.0 2023-03-08 13:15:48,717 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1204, 5.4213, 4.8462, 5.2835, 5.0727, 4.7006, 4.9106, 4.7274], device='cuda:0'), covar=tensor([0.1391, 0.0951, 0.0961, 0.0738, 0.0907, 0.1349, 0.2184, 0.2080], device='cuda:0'), in_proj_covar=tensor([0.0479, 0.0554, 0.0420, 0.0419, 0.0397, 0.0438, 0.0568, 0.0503], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 13:16:33,110 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60504.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:16:37,556 INFO [train2.py:809] (0/4) Epoch 16, batch 750, loss[ctc_loss=0.08837, att_loss=0.2409, loss=0.2104, over 16776.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005978, over 48.00 utterances.], tot_loss[ctc_loss=0.08606, att_loss=0.2419, loss=0.2107, over 3203023.08 frames. utt_duration=1252 frames, utt_pad_proportion=0.05243, over 10247.61 utterances.], batch size: 48, lr: 6.84e-03, grad_scale: 8.0 2023-03-08 13:17:08,889 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1885, 5.1634, 5.0922, 2.4466, 2.0914, 3.0355, 2.2580, 3.8212], device='cuda:0'), covar=tensor([0.0641, 0.0244, 0.0224, 0.4624, 0.5553, 0.2299, 0.3554, 0.1787], device='cuda:0'), in_proj_covar=tensor([0.0340, 0.0241, 0.0246, 0.0223, 0.0341, 0.0332, 0.0233, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.4788e-04, 8.9975e-05, 1.0562e-04, 9.6911e-05, 1.4473e-04, 1.3144e-04, 9.3209e-05, 1.4569e-04], device='cuda:0') 2023-03-08 13:17:14,650 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.152e+02 2.643e+02 3.369e+02 1.101e+03, threshold=5.285e+02, percent-clipped=4.0 2023-03-08 13:17:27,133 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9396, 5.1856, 5.1345, 5.0782, 5.1873, 5.1694, 4.8510, 4.7176], device='cuda:0'), covar=tensor([0.0981, 0.0507, 0.0276, 0.0443, 0.0313, 0.0291, 0.0373, 0.0326], device='cuda:0'), in_proj_covar=tensor([0.0493, 0.0332, 0.0301, 0.0325, 0.0385, 0.0407, 0.0326, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 13:17:40,000 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60545.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:17:50,557 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60552.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:17:54,428 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=60554.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:17:58,932 INFO [train2.py:809] (0/4) Epoch 16, batch 800, loss[ctc_loss=0.08761, att_loss=0.2486, loss=0.2164, over 17393.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03338, over 63.00 utterances.], tot_loss[ctc_loss=0.08641, att_loss=0.242, loss=0.2109, over 3217382.81 frames. utt_duration=1257 frames, utt_pad_proportion=0.05085, over 10246.97 utterances.], batch size: 63, lr: 6.84e-03, grad_scale: 8.0 2023-03-08 13:18:26,864 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-08 13:18:56,177 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60593.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:19:18,473 INFO [train2.py:809] (0/4) Epoch 16, batch 850, loss[ctc_loss=0.08001, att_loss=0.2478, loss=0.2142, over 17294.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02402, over 59.00 utterances.], tot_loss[ctc_loss=0.08622, att_loss=0.2415, loss=0.2104, over 3224104.08 frames. utt_duration=1233 frames, utt_pad_proportion=0.05948, over 10473.36 utterances.], batch size: 59, lr: 6.83e-03, grad_scale: 8.0 2023-03-08 13:19:45,541 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.94 vs. limit=5.0 2023-03-08 13:19:55,685 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.256e+02 2.159e+02 2.689e+02 3.192e+02 6.800e+02, threshold=5.378e+02, percent-clipped=2.0 2023-03-08 13:20:39,633 INFO [train2.py:809] (0/4) Epoch 16, batch 900, loss[ctc_loss=0.07684, att_loss=0.2317, loss=0.2007, over 16118.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006926, over 42.00 utterances.], tot_loss[ctc_loss=0.08621, att_loss=0.2416, loss=0.2105, over 3237858.75 frames. utt_duration=1225 frames, utt_pad_proportion=0.06057, over 10587.26 utterances.], batch size: 42, lr: 6.83e-03, grad_scale: 8.0 2023-03-08 13:21:30,002 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 13:21:42,253 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60696.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:21:43,178 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.75 vs. limit=5.0 2023-03-08 13:21:58,818 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7889, 6.0401, 5.4870, 5.8062, 5.6843, 5.2243, 5.3542, 5.2267], device='cuda:0'), covar=tensor([0.1148, 0.0833, 0.0934, 0.0723, 0.0788, 0.1362, 0.2390, 0.2258], device='cuda:0'), in_proj_covar=tensor([0.0480, 0.0553, 0.0424, 0.0423, 0.0399, 0.0440, 0.0571, 0.0506], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 13:22:00,167 INFO [train2.py:809] (0/4) Epoch 16, batch 950, loss[ctc_loss=0.06725, att_loss=0.2429, loss=0.2078, over 16891.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.007091, over 49.00 utterances.], tot_loss[ctc_loss=0.08596, att_loss=0.2427, loss=0.2113, over 3261390.81 frames. utt_duration=1221 frames, utt_pad_proportion=0.05727, over 10693.10 utterances.], batch size: 49, lr: 6.83e-03, grad_scale: 8.0 2023-03-08 13:22:02,050 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60708.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:22:17,993 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=60718.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:22:22,527 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60721.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:22:26,333 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-08 13:22:36,656 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.437e+02 2.156e+02 2.670e+02 3.136e+02 6.539e+02, threshold=5.340e+02, percent-clipped=2.0 2023-03-08 13:22:43,828 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60734.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:22:53,713 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 13:22:59,073 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60744.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:23:04,001 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3307, 4.7323, 4.9338, 4.7977, 4.8051, 5.2144, 4.8182, 5.3035], device='cuda:0'), covar=tensor([0.0791, 0.0745, 0.0819, 0.1173, 0.1863, 0.0990, 0.1039, 0.0816], device='cuda:0'), in_proj_covar=tensor([0.0793, 0.0472, 0.0551, 0.0608, 0.0811, 0.0568, 0.0451, 0.0540], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 13:23:14,891 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-03-08 13:23:19,076 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60756.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:23:20,643 INFO [train2.py:809] (0/4) Epoch 16, batch 1000, loss[ctc_loss=0.07531, att_loss=0.2511, loss=0.216, over 16455.00 frames. utt_duration=1432 frames, utt_pad_proportion=0.007292, over 46.00 utterances.], tot_loss[ctc_loss=0.08595, att_loss=0.2425, loss=0.2112, over 3264282.53 frames. utt_duration=1236 frames, utt_pad_proportion=0.05368, over 10573.64 utterances.], batch size: 46, lr: 6.83e-03, grad_scale: 8.0 2023-03-08 13:23:37,864 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3631, 2.5169, 3.1426, 4.3972, 3.9149, 3.8344, 2.7805, 2.0546], device='cuda:0'), covar=tensor([0.0677, 0.2383, 0.1114, 0.0491, 0.0784, 0.0439, 0.1673, 0.2510], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0214, 0.0189, 0.0202, 0.0211, 0.0167, 0.0197, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 13:23:39,169 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60769.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:23:55,677 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=60779.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:24:00,515 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60782.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:24:40,427 INFO [train2.py:809] (0/4) Epoch 16, batch 1050, loss[ctc_loss=0.06138, att_loss=0.2166, loss=0.1856, over 16287.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006808, over 43.00 utterances.], tot_loss[ctc_loss=0.08611, att_loss=0.2426, loss=0.2113, over 3270178.45 frames. utt_duration=1249 frames, utt_pad_proportion=0.05063, over 10487.66 utterances.], batch size: 43, lr: 6.82e-03, grad_scale: 8.0 2023-03-08 13:25:17,106 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.408e+02 2.141e+02 2.597e+02 3.263e+02 1.129e+03, threshold=5.194e+02, percent-clipped=2.0 2023-03-08 13:25:55,987 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=60854.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:26:00,449 INFO [train2.py:809] (0/4) Epoch 16, batch 1100, loss[ctc_loss=0.09743, att_loss=0.2611, loss=0.2283, over 16474.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006342, over 46.00 utterances.], tot_loss[ctc_loss=0.08672, att_loss=0.2432, loss=0.2119, over 3263210.28 frames. utt_duration=1221 frames, utt_pad_proportion=0.05954, over 10699.28 utterances.], batch size: 46, lr: 6.82e-03, grad_scale: 8.0 2023-03-08 13:27:12,913 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=60902.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:27:21,209 INFO [train2.py:809] (0/4) Epoch 16, batch 1150, loss[ctc_loss=0.08948, att_loss=0.2444, loss=0.2134, over 17285.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01286, over 55.00 utterances.], tot_loss[ctc_loss=0.08571, att_loss=0.2422, loss=0.2109, over 3262513.87 frames. utt_duration=1249 frames, utt_pad_proportion=0.0532, over 10457.53 utterances.], batch size: 55, lr: 6.82e-03, grad_scale: 8.0 2023-03-08 13:27:59,214 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 2.021e+02 2.572e+02 3.109e+02 6.628e+02, threshold=5.143e+02, percent-clipped=1.0 2023-03-08 13:28:04,993 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=2.00 vs. limit=2.0 2023-03-08 13:28:42,078 INFO [train2.py:809] (0/4) Epoch 16, batch 1200, loss[ctc_loss=0.0827, att_loss=0.2249, loss=0.1965, over 16406.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006517, over 44.00 utterances.], tot_loss[ctc_loss=0.08426, att_loss=0.2406, loss=0.2093, over 3264420.41 frames. utt_duration=1275 frames, utt_pad_proportion=0.0474, over 10257.09 utterances.], batch size: 44, lr: 6.81e-03, grad_scale: 8.0 2023-03-08 13:28:49,194 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-03-08 13:30:02,896 INFO [train2.py:809] (0/4) Epoch 16, batch 1250, loss[ctc_loss=0.06558, att_loss=0.2312, loss=0.1981, over 16158.00 frames. utt_duration=1578 frames, utt_pad_proportion=0.005563, over 41.00 utterances.], tot_loss[ctc_loss=0.08449, att_loss=0.2408, loss=0.2095, over 3256781.28 frames. utt_duration=1248 frames, utt_pad_proportion=0.05684, over 10446.73 utterances.], batch size: 41, lr: 6.81e-03, grad_scale: 8.0 2023-03-08 13:30:08,208 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9967, 5.1034, 4.9813, 2.3527, 1.9699, 2.7250, 2.3160, 3.7900], device='cuda:0'), covar=tensor([0.0769, 0.0260, 0.0225, 0.4422, 0.5842, 0.2784, 0.3392, 0.1829], device='cuda:0'), in_proj_covar=tensor([0.0340, 0.0242, 0.0246, 0.0226, 0.0341, 0.0334, 0.0234, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.4845e-04, 9.0466e-05, 1.0553e-04, 9.8016e-05, 1.4475e-04, 1.3235e-04, 9.4090e-05, 1.4584e-04], device='cuda:0') 2023-03-08 13:30:39,966 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 1.974e+02 2.401e+02 2.960e+02 6.292e+02, threshold=4.803e+02, percent-clipped=4.0 2023-03-08 13:31:22,897 INFO [train2.py:809] (0/4) Epoch 16, batch 1300, loss[ctc_loss=0.09816, att_loss=0.2622, loss=0.2294, over 17061.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008316, over 52.00 utterances.], tot_loss[ctc_loss=0.085, att_loss=0.2415, loss=0.2102, over 3265089.17 frames. utt_duration=1239 frames, utt_pad_proportion=0.05691, over 10557.61 utterances.], batch size: 52, lr: 6.81e-03, grad_scale: 8.0 2023-03-08 13:31:29,557 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61061.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:31:30,978 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9674, 5.2822, 4.7521, 5.3521, 4.6159, 5.0224, 5.4030, 5.1421], device='cuda:0'), covar=tensor([0.0598, 0.0301, 0.0896, 0.0299, 0.0473, 0.0225, 0.0235, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0365, 0.0290, 0.0344, 0.0303, 0.0298, 0.0223, 0.0275, 0.0260], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 13:31:34,251 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3709, 5.1851, 5.0529, 2.8728, 2.5185, 3.4810, 2.8868, 4.0355], device='cuda:0'), covar=tensor([0.0573, 0.0335, 0.0260, 0.4151, 0.5045, 0.2109, 0.2876, 0.1616], device='cuda:0'), in_proj_covar=tensor([0.0342, 0.0243, 0.0247, 0.0227, 0.0342, 0.0336, 0.0236, 0.0355], device='cuda:0'), out_proj_covar=tensor([1.4917e-04, 9.0737e-05, 1.0613e-04, 9.8615e-05, 1.4532e-04, 1.3286e-04, 9.4490e-05, 1.4628e-04], device='cuda:0') 2023-03-08 13:31:50,220 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61074.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:32:00,865 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-03-08 13:32:42,647 INFO [train2.py:809] (0/4) Epoch 16, batch 1350, loss[ctc_loss=0.1038, att_loss=0.259, loss=0.2279, over 16623.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005424, over 47.00 utterances.], tot_loss[ctc_loss=0.08559, att_loss=0.242, loss=0.2107, over 3264261.79 frames. utt_duration=1207 frames, utt_pad_proportion=0.06545, over 10833.17 utterances.], batch size: 47, lr: 6.81e-03, grad_scale: 8.0 2023-03-08 13:33:06,269 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61122.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:33:18,945 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.675e+02 2.149e+02 2.520e+02 3.082e+02 5.998e+02, threshold=5.039e+02, percent-clipped=2.0 2023-03-08 13:34:02,071 INFO [train2.py:809] (0/4) Epoch 16, batch 1400, loss[ctc_loss=0.08165, att_loss=0.2463, loss=0.2134, over 16321.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006551, over 45.00 utterances.], tot_loss[ctc_loss=0.08549, att_loss=0.2412, loss=0.21, over 3257880.05 frames. utt_duration=1226 frames, utt_pad_proportion=0.06329, over 10640.01 utterances.], batch size: 45, lr: 6.80e-03, grad_scale: 8.0 2023-03-08 13:34:11,330 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4703, 4.9323, 4.8189, 4.7849, 5.0784, 4.6497, 3.4196, 4.8302], device='cuda:0'), covar=tensor([0.0125, 0.0095, 0.0120, 0.0111, 0.0082, 0.0110, 0.0667, 0.0227], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0080, 0.0099, 0.0063, 0.0067, 0.0079, 0.0098, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 13:35:21,831 INFO [train2.py:809] (0/4) Epoch 16, batch 1450, loss[ctc_loss=0.07716, att_loss=0.2547, loss=0.2192, over 17050.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.00976, over 53.00 utterances.], tot_loss[ctc_loss=0.08574, att_loss=0.2417, loss=0.2105, over 3258214.25 frames. utt_duration=1195 frames, utt_pad_proportion=0.07155, over 10918.30 utterances.], batch size: 53, lr: 6.80e-03, grad_scale: 8.0 2023-03-08 13:35:58,473 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.469e+02 2.227e+02 2.591e+02 3.129e+02 8.266e+02, threshold=5.182e+02, percent-clipped=3.0 2023-03-08 13:36:40,870 INFO [train2.py:809] (0/4) Epoch 16, batch 1500, loss[ctc_loss=0.08218, att_loss=0.2234, loss=0.1951, over 15891.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008942, over 39.00 utterances.], tot_loss[ctc_loss=0.0861, att_loss=0.2415, loss=0.2104, over 3261286.28 frames. utt_duration=1217 frames, utt_pad_proportion=0.06534, over 10735.65 utterances.], batch size: 39, lr: 6.80e-03, grad_scale: 8.0 2023-03-08 13:37:59,312 INFO [train2.py:809] (0/4) Epoch 16, batch 1550, loss[ctc_loss=0.06616, att_loss=0.2105, loss=0.1816, over 15498.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.008945, over 36.00 utterances.], tot_loss[ctc_loss=0.08557, att_loss=0.2412, loss=0.21, over 3264276.58 frames. utt_duration=1231 frames, utt_pad_proportion=0.06195, over 10616.09 utterances.], batch size: 36, lr: 6.80e-03, grad_scale: 8.0 2023-03-08 13:38:35,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.223e+02 2.094e+02 2.601e+02 3.579e+02 7.093e+02, threshold=5.201e+02, percent-clipped=2.0 2023-03-08 13:39:02,978 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9293, 5.3361, 4.8812, 5.3547, 4.7691, 4.9492, 5.4617, 5.2062], device='cuda:0'), covar=tensor([0.0601, 0.0271, 0.0815, 0.0295, 0.0440, 0.0262, 0.0193, 0.0193], device='cuda:0'), in_proj_covar=tensor([0.0372, 0.0293, 0.0348, 0.0306, 0.0301, 0.0226, 0.0280, 0.0265], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 13:39:18,918 INFO [train2.py:809] (0/4) Epoch 16, batch 1600, loss[ctc_loss=0.06844, att_loss=0.2221, loss=0.1914, over 15647.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008138, over 37.00 utterances.], tot_loss[ctc_loss=0.08495, att_loss=0.2407, loss=0.2096, over 3264533.69 frames. utt_duration=1253 frames, utt_pad_proportion=0.05651, over 10431.29 utterances.], batch size: 37, lr: 6.79e-03, grad_scale: 8.0 2023-03-08 13:39:46,382 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61374.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:40:38,372 INFO [train2.py:809] (0/4) Epoch 16, batch 1650, loss[ctc_loss=0.1505, att_loss=0.282, loss=0.2557, over 13635.00 frames. utt_duration=372.6 frames, utt_pad_proportion=0.3463, over 147.00 utterances.], tot_loss[ctc_loss=0.08526, att_loss=0.2409, loss=0.2098, over 3259945.96 frames. utt_duration=1263 frames, utt_pad_proportion=0.05647, over 10338.70 utterances.], batch size: 147, lr: 6.79e-03, grad_scale: 8.0 2023-03-08 13:40:54,172 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=61417.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:41:02,478 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61422.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:41:14,484 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.484e+02 2.105e+02 2.549e+02 2.944e+02 7.207e+02, threshold=5.098e+02, percent-clipped=3.0 2023-03-08 13:41:56,930 INFO [train2.py:809] (0/4) Epoch 16, batch 1700, loss[ctc_loss=0.05479, att_loss=0.2055, loss=0.1753, over 14084.00 frames. utt_duration=1819 frames, utt_pad_proportion=0.05127, over 31.00 utterances.], tot_loss[ctc_loss=0.08506, att_loss=0.2402, loss=0.2092, over 3258451.17 frames. utt_duration=1268 frames, utt_pad_proportion=0.05517, over 10295.12 utterances.], batch size: 31, lr: 6.79e-03, grad_scale: 8.0 2023-03-08 13:42:53,774 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1714, 5.1394, 4.9536, 2.9196, 4.8918, 4.7962, 4.5280, 2.8693], device='cuda:0'), covar=tensor([0.0119, 0.0080, 0.0238, 0.1035, 0.0087, 0.0172, 0.0257, 0.1334], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0094, 0.0093, 0.0109, 0.0079, 0.0106, 0.0097, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 13:43:16,828 INFO [train2.py:809] (0/4) Epoch 16, batch 1750, loss[ctc_loss=0.07793, att_loss=0.2558, loss=0.2202, over 17414.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.04546, over 69.00 utterances.], tot_loss[ctc_loss=0.08556, att_loss=0.2414, loss=0.2102, over 3273912.03 frames. utt_duration=1273 frames, utt_pad_proportion=0.04857, over 10300.09 utterances.], batch size: 69, lr: 6.78e-03, grad_scale: 8.0 2023-03-08 13:43:53,593 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 1.998e+02 2.470e+02 3.033e+02 6.251e+02, threshold=4.940e+02, percent-clipped=1.0 2023-03-08 13:44:36,285 INFO [train2.py:809] (0/4) Epoch 16, batch 1800, loss[ctc_loss=0.1047, att_loss=0.2581, loss=0.2275, over 17393.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03341, over 63.00 utterances.], tot_loss[ctc_loss=0.0842, att_loss=0.2397, loss=0.2086, over 3262300.72 frames. utt_duration=1290 frames, utt_pad_proportion=0.04569, over 10125.04 utterances.], batch size: 63, lr: 6.78e-03, grad_scale: 8.0 2023-03-08 13:45:22,018 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1534, 5.1335, 4.9556, 3.0384, 4.9137, 4.7121, 4.5304, 2.8389], device='cuda:0'), covar=tensor([0.0140, 0.0090, 0.0268, 0.0978, 0.0095, 0.0178, 0.0263, 0.1307], device='cuda:0'), in_proj_covar=tensor([0.0068, 0.0094, 0.0093, 0.0109, 0.0079, 0.0106, 0.0097, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 13:45:55,240 INFO [train2.py:809] (0/4) Epoch 16, batch 1850, loss[ctc_loss=0.1152, att_loss=0.2683, loss=0.2377, over 17362.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03506, over 63.00 utterances.], tot_loss[ctc_loss=0.08439, att_loss=0.2395, loss=0.2085, over 3259693.38 frames. utt_duration=1267 frames, utt_pad_proportion=0.05391, over 10306.42 utterances.], batch size: 63, lr: 6.78e-03, grad_scale: 8.0 2023-03-08 13:46:31,938 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.537e+02 2.233e+02 2.629e+02 3.133e+02 5.185e+02, threshold=5.258e+02, percent-clipped=1.0 2023-03-08 13:47:14,832 INFO [train2.py:809] (0/4) Epoch 16, batch 1900, loss[ctc_loss=0.07774, att_loss=0.2427, loss=0.2097, over 16468.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006629, over 46.00 utterances.], tot_loss[ctc_loss=0.0847, att_loss=0.2402, loss=0.2091, over 3266433.45 frames. utt_duration=1267 frames, utt_pad_proportion=0.05119, over 10328.37 utterances.], batch size: 46, lr: 6.78e-03, grad_scale: 8.0 2023-03-08 13:47:16,588 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7880, 5.0911, 5.3968, 5.2980, 5.2777, 5.7750, 5.1038, 5.8649], device='cuda:0'), covar=tensor([0.0693, 0.0807, 0.0719, 0.1131, 0.1761, 0.0849, 0.0715, 0.0606], device='cuda:0'), in_proj_covar=tensor([0.0806, 0.0473, 0.0559, 0.0619, 0.0817, 0.0572, 0.0452, 0.0542], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 13:47:28,996 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4872, 3.3084, 3.1941, 2.9108, 3.3059, 3.3408, 3.3320, 2.4756], device='cuda:0'), covar=tensor([0.1275, 0.2202, 0.3418, 0.4202, 0.2565, 0.3821, 0.1467, 0.5102], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0156, 0.0169, 0.0231, 0.0131, 0.0226, 0.0143, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 13:48:33,455 INFO [train2.py:809] (0/4) Epoch 16, batch 1950, loss[ctc_loss=0.0741, att_loss=0.2417, loss=0.2081, over 16266.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.008178, over 43.00 utterances.], tot_loss[ctc_loss=0.08429, att_loss=0.24, loss=0.2088, over 3259212.29 frames. utt_duration=1278 frames, utt_pad_proportion=0.04981, over 10210.34 utterances.], batch size: 43, lr: 6.77e-03, grad_scale: 8.0 2023-03-08 13:48:49,373 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=61717.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:49:09,854 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.327e+02 2.102e+02 2.537e+02 2.891e+02 6.099e+02, threshold=5.073e+02, percent-clipped=2.0 2023-03-08 13:49:13,240 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61732.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:49:30,454 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61743.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:49:31,929 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1584, 5.2485, 4.8374, 2.6765, 4.9612, 4.8133, 4.3349, 2.3989], device='cuda:0'), covar=tensor([0.0176, 0.0107, 0.0321, 0.1406, 0.0105, 0.0190, 0.0422, 0.2090], device='cuda:0'), in_proj_covar=tensor([0.0069, 0.0094, 0.0093, 0.0109, 0.0079, 0.0106, 0.0097, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 13:49:53,040 INFO [train2.py:809] (0/4) Epoch 16, batch 2000, loss[ctc_loss=0.07324, att_loss=0.2255, loss=0.195, over 15886.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009175, over 39.00 utterances.], tot_loss[ctc_loss=0.0836, att_loss=0.2392, loss=0.2081, over 3257126.20 frames. utt_duration=1295 frames, utt_pad_proportion=0.04555, over 10071.24 utterances.], batch size: 39, lr: 6.77e-03, grad_scale: 8.0 2023-03-08 13:49:59,175 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7727, 5.0936, 5.3934, 5.2718, 5.2287, 5.7970, 5.0496, 5.8435], device='cuda:0'), covar=tensor([0.0687, 0.0723, 0.0736, 0.1205, 0.1792, 0.0783, 0.0705, 0.0612], device='cuda:0'), in_proj_covar=tensor([0.0810, 0.0475, 0.0562, 0.0621, 0.0820, 0.0574, 0.0455, 0.0547], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 13:50:05,405 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=61765.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:50:49,911 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61793.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:51:08,245 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61804.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:51:12,560 INFO [train2.py:809] (0/4) Epoch 16, batch 2050, loss[ctc_loss=0.08826, att_loss=0.2233, loss=0.1963, over 15767.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008761, over 38.00 utterances.], tot_loss[ctc_loss=0.08454, att_loss=0.2408, loss=0.2096, over 3274969.02 frames. utt_duration=1292 frames, utt_pad_proportion=0.04113, over 10153.55 utterances.], batch size: 38, lr: 6.77e-03, grad_scale: 8.0 2023-03-08 13:51:48,273 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 13:51:48,692 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.602e+02 2.229e+02 2.633e+02 3.391e+02 8.027e+02, threshold=5.266e+02, percent-clipped=7.0 2023-03-08 13:51:58,240 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0777, 5.3134, 5.2471, 5.2700, 5.3871, 5.3362, 4.9968, 4.8305], device='cuda:0'), covar=tensor([0.0955, 0.0478, 0.0296, 0.0522, 0.0271, 0.0302, 0.0370, 0.0343], device='cuda:0'), in_proj_covar=tensor([0.0491, 0.0329, 0.0301, 0.0325, 0.0383, 0.0402, 0.0326, 0.0362], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 13:52:32,058 INFO [train2.py:809] (0/4) Epoch 16, batch 2100, loss[ctc_loss=0.09054, att_loss=0.2585, loss=0.2249, over 17010.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008268, over 51.00 utterances.], tot_loss[ctc_loss=0.08504, att_loss=0.2413, loss=0.2101, over 3277989.82 frames. utt_duration=1289 frames, utt_pad_proportion=0.04104, over 10186.41 utterances.], batch size: 51, lr: 6.77e-03, grad_scale: 8.0 2023-03-08 13:53:51,104 INFO [train2.py:809] (0/4) Epoch 16, batch 2150, loss[ctc_loss=0.08954, att_loss=0.2409, loss=0.2106, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.00731, over 43.00 utterances.], tot_loss[ctc_loss=0.08518, att_loss=0.2414, loss=0.2101, over 3279654.54 frames. utt_duration=1291 frames, utt_pad_proportion=0.04081, over 10171.41 utterances.], batch size: 43, lr: 6.76e-03, grad_scale: 8.0 2023-03-08 13:54:03,271 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=61914.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:54:27,519 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 2.191e+02 2.731e+02 3.267e+02 7.309e+02, threshold=5.461e+02, percent-clipped=1.0 2023-03-08 13:55:11,211 INFO [train2.py:809] (0/4) Epoch 16, batch 2200, loss[ctc_loss=0.0756, att_loss=0.2482, loss=0.2136, over 16487.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005288, over 46.00 utterances.], tot_loss[ctc_loss=0.08538, att_loss=0.2416, loss=0.2103, over 3271378.01 frames. utt_duration=1258 frames, utt_pad_proportion=0.05235, over 10416.70 utterances.], batch size: 46, lr: 6.76e-03, grad_scale: 8.0 2023-03-08 13:55:40,188 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=61975.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:55:46,415 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5272, 2.8296, 3.3879, 4.4736, 4.0813, 4.0046, 2.9577, 2.0453], device='cuda:0'), covar=tensor([0.0595, 0.2046, 0.0997, 0.0504, 0.0715, 0.0405, 0.1436, 0.2420], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0211, 0.0189, 0.0202, 0.0213, 0.0168, 0.0196, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 13:56:20,421 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-62000.pt 2023-03-08 13:56:35,308 INFO [train2.py:809] (0/4) Epoch 16, batch 2250, loss[ctc_loss=0.08363, att_loss=0.217, loss=0.1903, over 15768.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008699, over 38.00 utterances.], tot_loss[ctc_loss=0.08553, att_loss=0.2417, loss=0.2105, over 3265843.20 frames. utt_duration=1252 frames, utt_pad_proportion=0.05406, over 10445.05 utterances.], batch size: 38, lr: 6.76e-03, grad_scale: 8.0 2023-03-08 13:57:11,173 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.510e+02 2.098e+02 2.625e+02 3.264e+02 7.065e+02, threshold=5.250e+02, percent-clipped=5.0 2023-03-08 13:57:11,543 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6328, 2.9892, 3.7422, 3.1779, 3.6237, 4.6779, 4.5143, 3.4542], device='cuda:0'), covar=tensor([0.0356, 0.1781, 0.1059, 0.1358, 0.1017, 0.1087, 0.0637, 0.1218], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0236, 0.0264, 0.0208, 0.0250, 0.0338, 0.0240, 0.0230], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 13:57:54,975 INFO [train2.py:809] (0/4) Epoch 16, batch 2300, loss[ctc_loss=0.1004, att_loss=0.2354, loss=0.2084, over 15883.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.007522, over 39.00 utterances.], tot_loss[ctc_loss=0.08519, att_loss=0.2418, loss=0.2104, over 3267252.01 frames. utt_duration=1274 frames, utt_pad_proportion=0.0475, over 10268.47 utterances.], batch size: 39, lr: 6.75e-03, grad_scale: 8.0 2023-03-08 13:58:43,789 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62088.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:58:47,574 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7663, 3.5381, 3.5271, 2.9985, 3.6256, 3.5905, 3.6241, 2.5077], device='cuda:0'), covar=tensor([0.0978, 0.1076, 0.1838, 0.3944, 0.0919, 0.2788, 0.0805, 0.5149], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0158, 0.0168, 0.0231, 0.0132, 0.0224, 0.0144, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 13:59:02,013 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62099.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:59:14,698 INFO [train2.py:809] (0/4) Epoch 16, batch 2350, loss[ctc_loss=0.09171, att_loss=0.2485, loss=0.2172, over 16273.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007798, over 43.00 utterances.], tot_loss[ctc_loss=0.08556, att_loss=0.2419, loss=0.2106, over 3267833.55 frames. utt_duration=1262 frames, utt_pad_proportion=0.0507, over 10368.71 utterances.], batch size: 43, lr: 6.75e-03, grad_scale: 8.0 2023-03-08 13:59:38,203 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62122.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 13:59:49,925 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.396e+02 2.132e+02 2.401e+02 3.401e+02 9.519e+02, threshold=4.802e+02, percent-clipped=3.0 2023-03-08 14:00:34,127 INFO [train2.py:809] (0/4) Epoch 16, batch 2400, loss[ctc_loss=0.08089, att_loss=0.2525, loss=0.2182, over 17293.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01098, over 55.00 utterances.], tot_loss[ctc_loss=0.08624, att_loss=0.2427, loss=0.2114, over 3274167.35 frames. utt_duration=1227 frames, utt_pad_proportion=0.0586, over 10690.65 utterances.], batch size: 55, lr: 6.75e-03, grad_scale: 16.0 2023-03-08 14:01:15,272 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62183.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:01:48,089 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.86 vs. limit=2.0 2023-03-08 14:01:53,485 INFO [train2.py:809] (0/4) Epoch 16, batch 2450, loss[ctc_loss=0.07401, att_loss=0.2456, loss=0.2113, over 16449.00 frames. utt_duration=1432 frames, utt_pad_proportion=0.007578, over 46.00 utterances.], tot_loss[ctc_loss=0.08637, att_loss=0.2431, loss=0.2117, over 3260560.88 frames. utt_duration=1213 frames, utt_pad_proportion=0.06488, over 10768.70 utterances.], batch size: 46, lr: 6.75e-03, grad_scale: 16.0 2023-03-08 14:02:29,358 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 2.234e+02 2.691e+02 3.084e+02 9.437e+02, threshold=5.381e+02, percent-clipped=5.0 2023-03-08 14:02:50,388 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4419, 4.9161, 4.6788, 4.8514, 4.9212, 4.6390, 3.2026, 4.7976], device='cuda:0'), covar=tensor([0.0102, 0.0090, 0.0121, 0.0089, 0.0095, 0.0082, 0.0744, 0.0179], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0081, 0.0101, 0.0064, 0.0069, 0.0080, 0.0099, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 14:02:55,569 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.23 vs. limit=5.0 2023-03-08 14:03:09,633 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 14:03:12,441 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.20 vs. limit=5.0 2023-03-08 14:03:13,139 INFO [train2.py:809] (0/4) Epoch 16, batch 2500, loss[ctc_loss=0.1005, att_loss=0.2636, loss=0.231, over 17300.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.0113, over 55.00 utterances.], tot_loss[ctc_loss=0.0871, att_loss=0.2434, loss=0.2122, over 3259202.48 frames. utt_duration=1172 frames, utt_pad_proportion=0.076, over 11137.83 utterances.], batch size: 55, lr: 6.74e-03, grad_scale: 16.0 2023-03-08 14:03:33,434 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62270.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:03:38,168 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62273.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:04:21,660 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7764, 3.7561, 3.0909, 3.3151, 3.9437, 3.6427, 3.0429, 4.2210], device='cuda:0'), covar=tensor([0.1122, 0.0497, 0.1136, 0.0680, 0.0574, 0.0665, 0.0812, 0.0400], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0205, 0.0217, 0.0190, 0.0262, 0.0229, 0.0194, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 14:04:32,513 INFO [train2.py:809] (0/4) Epoch 16, batch 2550, loss[ctc_loss=0.1022, att_loss=0.2595, loss=0.2281, over 17065.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008374, over 52.00 utterances.], tot_loss[ctc_loss=0.08673, att_loss=0.2431, loss=0.2118, over 3259065.86 frames. utt_duration=1179 frames, utt_pad_proportion=0.07494, over 11069.71 utterances.], batch size: 52, lr: 6.74e-03, grad_scale: 8.0 2023-03-08 14:05:09,995 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.185e+02 2.191e+02 2.595e+02 3.145e+02 7.901e+02, threshold=5.189e+02, percent-clipped=4.0 2023-03-08 14:05:15,168 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62334.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:05:52,042 INFO [train2.py:809] (0/4) Epoch 16, batch 2600, loss[ctc_loss=0.1258, att_loss=0.2659, loss=0.2379, over 17468.00 frames. utt_duration=1014 frames, utt_pad_proportion=0.04357, over 69.00 utterances.], tot_loss[ctc_loss=0.0854, att_loss=0.2422, loss=0.2108, over 3266964.84 frames. utt_duration=1218 frames, utt_pad_proportion=0.06402, over 10744.74 utterances.], batch size: 69, lr: 6.74e-03, grad_scale: 8.0 2023-03-08 14:06:41,241 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62388.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:06:45,750 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62391.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:06:58,810 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62399.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:07:11,494 INFO [train2.py:809] (0/4) Epoch 16, batch 2650, loss[ctc_loss=0.0683, att_loss=0.2289, loss=0.1968, over 16016.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006738, over 40.00 utterances.], tot_loss[ctc_loss=0.08531, att_loss=0.2422, loss=0.2108, over 3273181.01 frames. utt_duration=1235 frames, utt_pad_proportion=0.0583, over 10613.55 utterances.], batch size: 40, lr: 6.74e-03, grad_scale: 8.0 2023-03-08 14:07:49,380 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.560e+02 2.130e+02 2.553e+02 3.253e+02 7.281e+02, threshold=5.105e+02, percent-clipped=6.0 2023-03-08 14:07:57,887 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62436.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:08:15,309 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62447.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:08:24,070 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62452.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:08:31,463 INFO [train2.py:809] (0/4) Epoch 16, batch 2700, loss[ctc_loss=0.09381, att_loss=0.2674, loss=0.2326, over 17130.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01362, over 56.00 utterances.], tot_loss[ctc_loss=0.08587, att_loss=0.2432, loss=0.2117, over 3287788.51 frames. utt_duration=1241 frames, utt_pad_proportion=0.05208, over 10613.57 utterances.], batch size: 56, lr: 6.73e-03, grad_scale: 8.0 2023-03-08 14:09:05,211 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62478.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:09:51,786 INFO [train2.py:809] (0/4) Epoch 16, batch 2750, loss[ctc_loss=0.08454, att_loss=0.2301, loss=0.201, over 15748.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.00874, over 38.00 utterances.], tot_loss[ctc_loss=0.0856, att_loss=0.2421, loss=0.2108, over 3273113.18 frames. utt_duration=1230 frames, utt_pad_proportion=0.05884, over 10660.19 utterances.], batch size: 38, lr: 6.73e-03, grad_scale: 8.0 2023-03-08 14:10:23,398 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6611, 3.0582, 3.8066, 4.6218, 3.9842, 4.1437, 2.9706, 2.4804], device='cuda:0'), covar=tensor([0.0608, 0.1958, 0.0833, 0.0476, 0.0789, 0.0380, 0.1399, 0.2158], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0215, 0.0191, 0.0203, 0.0213, 0.0171, 0.0197, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 14:10:29,271 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.352e+02 2.080e+02 2.636e+02 3.250e+02 8.951e+02, threshold=5.273e+02, percent-clipped=4.0 2023-03-08 14:10:52,901 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62545.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:11:10,990 INFO [train2.py:809] (0/4) Epoch 16, batch 2800, loss[ctc_loss=0.0824, att_loss=0.2455, loss=0.2129, over 17127.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01458, over 56.00 utterances.], tot_loss[ctc_loss=0.08488, att_loss=0.2414, loss=0.2101, over 3269908.13 frames. utt_duration=1232 frames, utt_pad_proportion=0.05919, over 10627.50 utterances.], batch size: 56, lr: 6.73e-03, grad_scale: 8.0 2023-03-08 14:11:31,243 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62570.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:11:55,448 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62585.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:12:29,628 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62606.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:12:30,823 INFO [train2.py:809] (0/4) Epoch 16, batch 2850, loss[ctc_loss=0.09258, att_loss=0.2505, loss=0.2189, over 17271.00 frames. utt_duration=1257 frames, utt_pad_proportion=0.01314, over 55.00 utterances.], tot_loss[ctc_loss=0.08484, att_loss=0.2412, loss=0.21, over 3267023.14 frames. utt_duration=1227 frames, utt_pad_proportion=0.06163, over 10662.79 utterances.], batch size: 55, lr: 6.72e-03, grad_scale: 8.0 2023-03-08 14:12:48,183 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62618.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:13:01,661 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6316, 3.7268, 3.4606, 3.8144, 2.7047, 3.7543, 2.6124, 2.0770], device='cuda:0'), covar=tensor([0.0401, 0.0260, 0.0781, 0.0221, 0.1436, 0.0229, 0.1371, 0.1430], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0141, 0.0258, 0.0133, 0.0220, 0.0122, 0.0229, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 14:13:05,992 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62629.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:13:09,362 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 1.902e+02 2.514e+02 2.955e+02 4.757e+02, threshold=5.028e+02, percent-clipped=0.0 2023-03-08 14:13:19,156 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4497, 2.6266, 3.3144, 4.4516, 3.9139, 4.0089, 2.8640, 2.3521], device='cuda:0'), covar=tensor([0.0748, 0.2354, 0.1106, 0.0552, 0.0874, 0.0441, 0.1646, 0.2125], device='cuda:0'), in_proj_covar=tensor([0.0179, 0.0216, 0.0192, 0.0205, 0.0214, 0.0172, 0.0198, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 14:13:34,649 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62646.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:13:51,155 INFO [train2.py:809] (0/4) Epoch 16, batch 2900, loss[ctc_loss=0.1171, att_loss=0.2747, loss=0.2432, over 17302.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02437, over 59.00 utterances.], tot_loss[ctc_loss=0.08464, att_loss=0.2413, loss=0.21, over 3266865.55 frames. utt_duration=1224 frames, utt_pad_proportion=0.06152, over 10689.25 utterances.], batch size: 59, lr: 6.72e-03, grad_scale: 8.0 2023-03-08 14:14:03,412 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-03-08 14:14:05,851 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62666.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:15:12,445 INFO [train2.py:809] (0/4) Epoch 16, batch 2950, loss[ctc_loss=0.1257, att_loss=0.2684, loss=0.2399, over 17035.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.009782, over 53.00 utterances.], tot_loss[ctc_loss=0.08506, att_loss=0.2414, loss=0.2101, over 3269478.79 frames. utt_duration=1223 frames, utt_pad_proportion=0.06102, over 10704.64 utterances.], batch size: 53, lr: 6.72e-03, grad_scale: 8.0 2023-03-08 14:15:44,543 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62727.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:15:50,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 2.017e+02 2.387e+02 2.901e+02 6.735e+02, threshold=4.774e+02, percent-clipped=4.0 2023-03-08 14:15:59,997 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6135, 3.0072, 3.6508, 3.0601, 3.6212, 4.6748, 4.5085, 3.2493], device='cuda:0'), covar=tensor([0.0358, 0.1707, 0.1242, 0.1323, 0.1041, 0.0909, 0.0532, 0.1205], device='cuda:0'), in_proj_covar=tensor([0.0235, 0.0235, 0.0264, 0.0207, 0.0248, 0.0337, 0.0238, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 14:16:17,597 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62747.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:16:32,913 INFO [train2.py:809] (0/4) Epoch 16, batch 3000, loss[ctc_loss=0.09951, att_loss=0.2641, loss=0.2312, over 17288.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01269, over 55.00 utterances.], tot_loss[ctc_loss=0.08421, att_loss=0.2412, loss=0.2098, over 3276194.03 frames. utt_duration=1239 frames, utt_pad_proportion=0.05611, over 10592.02 utterances.], batch size: 55, lr: 6.72e-03, grad_scale: 8.0 2023-03-08 14:16:32,916 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 14:16:46,719 INFO [train2.py:843] (0/4) Epoch 16, validation: ctc_loss=0.0433, att_loss=0.235, loss=0.1967, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 14:16:46,720 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 14:16:59,458 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7756, 4.9829, 5.4929, 4.9129, 4.8486, 5.6165, 5.0646, 5.6321], device='cuda:0'), covar=tensor([0.1251, 0.1448, 0.1066, 0.2258, 0.3434, 0.1493, 0.1066, 0.1271], device='cuda:0'), in_proj_covar=tensor([0.0804, 0.0472, 0.0557, 0.0615, 0.0811, 0.0566, 0.0448, 0.0543], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 14:17:11,755 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=62773.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:17:19,173 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62778.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:18:06,025 INFO [train2.py:809] (0/4) Epoch 16, batch 3050, loss[ctc_loss=0.06677, att_loss=0.2169, loss=0.1869, over 15512.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.007576, over 36.00 utterances.], tot_loss[ctc_loss=0.08412, att_loss=0.2419, loss=0.2103, over 3285776.33 frames. utt_duration=1243 frames, utt_pad_proportion=0.05256, over 10583.19 utterances.], batch size: 36, lr: 6.71e-03, grad_scale: 8.0 2023-03-08 14:18:36,324 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62826.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:18:43,970 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.353e+02 2.145e+02 2.502e+02 3.037e+02 6.032e+02, threshold=5.003e+02, percent-clipped=3.0 2023-03-08 14:18:48,928 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=62834.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:19:28,200 INFO [train2.py:809] (0/4) Epoch 16, batch 3100, loss[ctc_loss=0.07021, att_loss=0.2232, loss=0.1926, over 15896.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008521, over 39.00 utterances.], tot_loss[ctc_loss=0.08337, att_loss=0.241, loss=0.2095, over 3281298.69 frames. utt_duration=1271 frames, utt_pad_proportion=0.04671, over 10336.17 utterances.], batch size: 39, lr: 6.71e-03, grad_scale: 8.0 2023-03-08 14:19:43,487 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 14:20:42,344 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62901.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:20:45,882 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6550, 3.6671, 3.8756, 2.5433, 2.5016, 2.8088, 2.7407, 3.4687], device='cuda:0'), covar=tensor([0.0660, 0.0408, 0.0334, 0.3402, 0.3749, 0.2115, 0.1941, 0.1224], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0249, 0.0252, 0.0232, 0.0345, 0.0335, 0.0241, 0.0361], device='cuda:0'), out_proj_covar=tensor([1.5147e-04, 9.3673e-05, 1.0903e-04, 1.0104e-04, 1.4665e-04, 1.3265e-04, 9.6134e-05, 1.4882e-04], device='cuda:0') 2023-03-08 14:20:51,879 INFO [train2.py:809] (0/4) Epoch 16, batch 3150, loss[ctc_loss=0.0852, att_loss=0.2286, loss=0.1999, over 15869.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01003, over 39.00 utterances.], tot_loss[ctc_loss=0.08372, att_loss=0.2412, loss=0.2097, over 3283488.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.04968, over 10417.26 utterances.], batch size: 39, lr: 6.71e-03, grad_scale: 8.0 2023-03-08 14:20:56,381 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-08 14:21:28,144 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=62929.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:21:31,247 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.078e+02 2.532e+02 3.325e+02 7.691e+02, threshold=5.063e+02, percent-clipped=6.0 2023-03-08 14:21:49,251 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=62941.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:22:15,345 INFO [train2.py:809] (0/4) Epoch 16, batch 3200, loss[ctc_loss=0.07603, att_loss=0.2186, loss=0.1901, over 15760.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009264, over 38.00 utterances.], tot_loss[ctc_loss=0.08414, att_loss=0.241, loss=0.2097, over 3272125.15 frames. utt_duration=1268 frames, utt_pad_proportion=0.05099, over 10331.68 utterances.], batch size: 38, lr: 6.71e-03, grad_scale: 8.0 2023-03-08 14:22:47,673 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=62977.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:23:19,176 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7489, 5.0515, 4.3563, 5.2024, 4.6215, 4.8691, 5.1963, 4.9235], device='cuda:0'), covar=tensor([0.0666, 0.0281, 0.1122, 0.0289, 0.0390, 0.0286, 0.0273, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0372, 0.0294, 0.0347, 0.0312, 0.0299, 0.0224, 0.0281, 0.0262], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 14:23:37,495 INFO [train2.py:809] (0/4) Epoch 16, batch 3250, loss[ctc_loss=0.1102, att_loss=0.2579, loss=0.2284, over 17319.00 frames. utt_duration=1101 frames, utt_pad_proportion=0.0376, over 63.00 utterances.], tot_loss[ctc_loss=0.085, att_loss=0.2417, loss=0.2103, over 3275826.05 frames. utt_duration=1253 frames, utt_pad_proportion=0.05343, over 10473.08 utterances.], batch size: 63, lr: 6.70e-03, grad_scale: 8.0 2023-03-08 14:24:03,069 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63022.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:24:17,455 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.315e+02 2.062e+02 2.466e+02 2.972e+02 8.137e+02, threshold=4.932e+02, percent-clipped=1.0 2023-03-08 14:24:45,790 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63047.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:25:01,619 INFO [train2.py:809] (0/4) Epoch 16, batch 3300, loss[ctc_loss=0.0818, att_loss=0.2442, loss=0.2118, over 16324.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006718, over 45.00 utterances.], tot_loss[ctc_loss=0.0834, att_loss=0.2411, loss=0.2095, over 3281336.52 frames. utt_duration=1278 frames, utt_pad_proportion=0.04619, over 10286.11 utterances.], batch size: 45, lr: 6.70e-03, grad_scale: 8.0 2023-03-08 14:26:04,462 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63095.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:26:24,098 INFO [train2.py:809] (0/4) Epoch 16, batch 3350, loss[ctc_loss=0.1024, att_loss=0.2634, loss=0.2312, over 17106.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01482, over 56.00 utterances.], tot_loss[ctc_loss=0.08352, att_loss=0.2414, loss=0.2098, over 3285956.72 frames. utt_duration=1283 frames, utt_pad_proportion=0.04255, over 10254.52 utterances.], batch size: 56, lr: 6.70e-03, grad_scale: 8.0 2023-03-08 14:27:00,459 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63129.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:27:03,357 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.304e+02 2.090e+02 2.534e+02 3.183e+02 5.519e+02, threshold=5.068e+02, percent-clipped=6.0 2023-03-08 14:27:47,072 INFO [train2.py:809] (0/4) Epoch 16, batch 3400, loss[ctc_loss=0.08897, att_loss=0.2402, loss=0.2099, over 16259.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.008605, over 43.00 utterances.], tot_loss[ctc_loss=0.08329, att_loss=0.2416, loss=0.2099, over 3288084.58 frames. utt_duration=1275 frames, utt_pad_proportion=0.04326, over 10327.94 utterances.], batch size: 43, lr: 6.70e-03, grad_scale: 8.0 2023-03-08 14:28:08,163 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.22 vs. limit=2.0 2023-03-08 14:28:47,195 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-03-08 14:29:01,667 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63201.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:29:11,055 INFO [train2.py:809] (0/4) Epoch 16, batch 3450, loss[ctc_loss=0.09181, att_loss=0.2558, loss=0.223, over 16884.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.007547, over 49.00 utterances.], tot_loss[ctc_loss=0.08369, att_loss=0.2411, loss=0.2097, over 3268177.30 frames. utt_duration=1246 frames, utt_pad_proportion=0.05309, over 10500.70 utterances.], batch size: 49, lr: 6.69e-03, grad_scale: 8.0 2023-03-08 14:29:34,438 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63221.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:29:49,778 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 2.022e+02 2.458e+02 2.979e+02 7.741e+02, threshold=4.916e+02, percent-clipped=5.0 2023-03-08 14:30:06,787 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63241.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:30:20,166 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63249.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:30:32,955 INFO [train2.py:809] (0/4) Epoch 16, batch 3500, loss[ctc_loss=0.09112, att_loss=0.2592, loss=0.2256, over 16775.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.005345, over 48.00 utterances.], tot_loss[ctc_loss=0.08403, att_loss=0.2412, loss=0.2098, over 3270874.35 frames. utt_duration=1260 frames, utt_pad_proportion=0.05005, over 10399.94 utterances.], batch size: 48, lr: 6.69e-03, grad_scale: 8.0 2023-03-08 14:30:49,391 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1377, 5.4109, 5.3670, 5.2953, 5.4997, 5.4320, 5.2411, 4.8845], device='cuda:0'), covar=tensor([0.0895, 0.0447, 0.0232, 0.0466, 0.0225, 0.0265, 0.0268, 0.0298], device='cuda:0'), in_proj_covar=tensor([0.0485, 0.0328, 0.0303, 0.0320, 0.0378, 0.0401, 0.0325, 0.0359], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 14:31:14,087 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63282.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:31:25,073 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63289.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:31:56,005 INFO [train2.py:809] (0/4) Epoch 16, batch 3550, loss[ctc_loss=0.08339, att_loss=0.2376, loss=0.2068, over 16261.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.007681, over 43.00 utterances.], tot_loss[ctc_loss=0.08492, att_loss=0.2416, loss=0.2102, over 3269046.70 frames. utt_duration=1221 frames, utt_pad_proportion=0.06, over 10724.92 utterances.], batch size: 43, lr: 6.69e-03, grad_scale: 8.0 2023-03-08 14:32:20,773 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63322.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:32:36,185 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.051e+02 2.467e+02 3.018e+02 7.226e+02, threshold=4.933e+02, percent-clipped=6.0 2023-03-08 14:33:19,238 INFO [train2.py:809] (0/4) Epoch 16, batch 3600, loss[ctc_loss=0.07789, att_loss=0.2441, loss=0.2109, over 16477.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.005981, over 46.00 utterances.], tot_loss[ctc_loss=0.08469, att_loss=0.2419, loss=0.2105, over 3266203.69 frames. utt_duration=1212 frames, utt_pad_proportion=0.06163, over 10793.60 utterances.], batch size: 46, lr: 6.69e-03, grad_scale: 8.0 2023-03-08 14:33:39,693 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63370.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:34:42,626 INFO [train2.py:809] (0/4) Epoch 16, batch 3650, loss[ctc_loss=0.07356, att_loss=0.2234, loss=0.1934, over 15905.00 frames. utt_duration=1633 frames, utt_pad_proportion=0.007945, over 39.00 utterances.], tot_loss[ctc_loss=0.08495, att_loss=0.2419, loss=0.2105, over 3262841.78 frames. utt_duration=1206 frames, utt_pad_proportion=0.06452, over 10839.49 utterances.], batch size: 39, lr: 6.68e-03, grad_scale: 8.0 2023-03-08 14:35:03,583 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63420.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:35:06,809 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9551, 3.7022, 3.0301, 3.2686, 3.8010, 3.4808, 2.6223, 4.1668], device='cuda:0'), covar=tensor([0.0995, 0.0459, 0.1064, 0.0729, 0.0686, 0.0681, 0.1035, 0.0491], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0205, 0.0218, 0.0190, 0.0263, 0.0229, 0.0194, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 14:35:06,820 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63422.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:35:16,354 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5491, 5.0312, 4.8188, 5.0336, 5.0812, 4.6854, 3.4611, 4.9697], device='cuda:0'), covar=tensor([0.0112, 0.0099, 0.0126, 0.0078, 0.0083, 0.0114, 0.0690, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0083, 0.0080, 0.0099, 0.0063, 0.0067, 0.0079, 0.0098, 0.0100], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002, 0.0003, 0.0003], device='cuda:0') 2023-03-08 14:35:19,549 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63429.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:35:22,398 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.142e+02 2.479e+02 3.214e+02 4.969e+02, threshold=4.958e+02, percent-clipped=1.0 2023-03-08 14:36:06,127 INFO [train2.py:809] (0/4) Epoch 16, batch 3700, loss[ctc_loss=0.07233, att_loss=0.233, loss=0.2009, over 16557.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005387, over 45.00 utterances.], tot_loss[ctc_loss=0.08502, att_loss=0.2419, loss=0.2105, over 3267576.08 frames. utt_duration=1226 frames, utt_pad_proportion=0.05942, over 10675.20 utterances.], batch size: 45, lr: 6.68e-03, grad_scale: 8.0 2023-03-08 14:36:26,366 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4788, 2.1308, 4.9194, 3.6174, 2.8961, 4.1456, 4.7768, 4.5470], device='cuda:0'), covar=tensor([0.0229, 0.2025, 0.0150, 0.1143, 0.1944, 0.0274, 0.0109, 0.0216], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0244, 0.0163, 0.0312, 0.0265, 0.0198, 0.0144, 0.0174], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 14:36:40,541 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63477.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:36:47,195 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63481.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:36:50,294 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63483.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:36:59,522 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-03-08 14:37:07,616 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.60 vs. limit=2.0 2023-03-08 14:37:29,305 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4775, 4.3940, 4.3724, 4.3340, 4.9097, 4.5187, 4.3045, 2.3538], device='cuda:0'), covar=tensor([0.0230, 0.0291, 0.0337, 0.0257, 0.0882, 0.0214, 0.0318, 0.2077], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0155, 0.0163, 0.0173, 0.0350, 0.0137, 0.0147, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 14:37:30,409 INFO [train2.py:809] (0/4) Epoch 16, batch 3750, loss[ctc_loss=0.09319, att_loss=0.253, loss=0.221, over 17111.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01552, over 56.00 utterances.], tot_loss[ctc_loss=0.08462, att_loss=0.2417, loss=0.2103, over 3268504.10 frames. utt_duration=1222 frames, utt_pad_proportion=0.05961, over 10707.66 utterances.], batch size: 56, lr: 6.68e-03, grad_scale: 8.0 2023-03-08 14:38:09,923 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 2.179e+02 2.509e+02 3.051e+02 5.574e+02, threshold=5.018e+02, percent-clipped=2.0 2023-03-08 14:38:52,443 INFO [train2.py:809] (0/4) Epoch 16, batch 3800, loss[ctc_loss=0.0697, att_loss=0.233, loss=0.2003, over 16547.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005912, over 45.00 utterances.], tot_loss[ctc_loss=0.08499, att_loss=0.2423, loss=0.2109, over 3270023.16 frames. utt_duration=1217 frames, utt_pad_proportion=0.06255, over 10765.39 utterances.], batch size: 45, lr: 6.67e-03, grad_scale: 8.0 2023-03-08 14:38:58,038 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-03-08 14:39:11,858 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63569.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:39:17,664 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 14:39:25,939 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63577.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:39:40,633 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63586.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:40:07,589 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4961, 4.5804, 4.4837, 4.4908, 5.1803, 4.5713, 4.5840, 2.4771], device='cuda:0'), covar=tensor([0.0223, 0.0285, 0.0314, 0.0271, 0.0754, 0.0188, 0.0260, 0.1902], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0155, 0.0163, 0.0173, 0.0350, 0.0137, 0.0146, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 14:40:10,190 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=5.18 vs. limit=5.0 2023-03-08 14:40:15,130 INFO [train2.py:809] (0/4) Epoch 16, batch 3850, loss[ctc_loss=0.07749, att_loss=0.2197, loss=0.1913, over 15869.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01008, over 39.00 utterances.], tot_loss[ctc_loss=0.08428, att_loss=0.2421, loss=0.2106, over 3276442.83 frames. utt_duration=1228 frames, utt_pad_proportion=0.05976, over 10687.91 utterances.], batch size: 39, lr: 6.67e-03, grad_scale: 8.0 2023-03-08 14:40:52,486 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63630.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:40:53,568 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 2.062e+02 2.555e+02 3.342e+02 5.623e+02, threshold=5.109e+02, percent-clipped=2.0 2023-03-08 14:41:10,756 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-08 14:41:19,867 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63647.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:41:33,921 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6711, 3.3437, 3.3820, 2.8432, 3.3315, 3.3982, 3.3959, 2.3181], device='cuda:0'), covar=tensor([0.1109, 0.1397, 0.1793, 0.4693, 0.2257, 0.1900, 0.1015, 0.5518], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0158, 0.0170, 0.0233, 0.0132, 0.0229, 0.0146, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 14:41:35,251 INFO [train2.py:809] (0/4) Epoch 16, batch 3900, loss[ctc_loss=0.1049, att_loss=0.2575, loss=0.227, over 17314.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.024, over 59.00 utterances.], tot_loss[ctc_loss=0.08406, att_loss=0.2418, loss=0.2103, over 3271487.05 frames. utt_duration=1201 frames, utt_pad_proportion=0.06717, over 10909.19 utterances.], batch size: 59, lr: 6.67e-03, grad_scale: 8.0 2023-03-08 14:42:14,133 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0741, 3.8010, 3.7622, 3.1997, 3.7507, 3.8113, 3.7667, 2.7450], device='cuda:0'), covar=tensor([0.0842, 0.1230, 0.1812, 0.3825, 0.1879, 0.2282, 0.0778, 0.4909], device='cuda:0'), in_proj_covar=tensor([0.0137, 0.0158, 0.0170, 0.0234, 0.0132, 0.0229, 0.0146, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 14:42:54,346 INFO [train2.py:809] (0/4) Epoch 16, batch 3950, loss[ctc_loss=0.07288, att_loss=0.2425, loss=0.2086, over 16873.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007958, over 49.00 utterances.], tot_loss[ctc_loss=0.08387, att_loss=0.2413, loss=0.2098, over 3274029.32 frames. utt_duration=1205 frames, utt_pad_proportion=0.06535, over 10880.76 utterances.], batch size: 49, lr: 6.67e-03, grad_scale: 8.0 2023-03-08 14:43:31,754 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.446e+02 2.047e+02 2.429e+02 3.327e+02 6.424e+02, threshold=4.857e+02, percent-clipped=3.0 2023-03-08 14:43:47,149 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-16.pt 2023-03-08 14:44:08,713 INFO [train2.py:809] (0/4) Epoch 17, batch 0, loss[ctc_loss=0.1067, att_loss=0.2593, loss=0.2288, over 16961.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007655, over 50.00 utterances.], tot_loss[ctc_loss=0.1067, att_loss=0.2593, loss=0.2288, over 16961.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007655, over 50.00 utterances.], batch size: 50, lr: 6.46e-03, grad_scale: 8.0 2023-03-08 14:44:08,715 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 14:44:15,138 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1819, 5.2689, 4.9705, 2.8130, 2.2012, 3.2575, 2.5394, 3.9467], device='cuda:0'), covar=tensor([0.0629, 0.0237, 0.0302, 0.4524, 0.5946, 0.2361, 0.3827, 0.1522], device='cuda:0'), in_proj_covar=tensor([0.0340, 0.0244, 0.0249, 0.0225, 0.0338, 0.0329, 0.0239, 0.0352], device='cuda:0'), out_proj_covar=tensor([1.4724e-04, 9.0898e-05, 1.0734e-04, 9.7254e-05, 1.4338e-04, 1.2993e-04, 9.5658e-05, 1.4504e-04], device='cuda:0') 2023-03-08 14:44:15,522 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8761, 4.5146, 4.4576, 4.3375, 4.3749, 4.8237, 4.6839, 4.8785], device='cuda:0'), covar=tensor([0.0840, 0.0728, 0.0892, 0.1239, 0.2271, 0.0882, 0.0464, 0.0679], device='cuda:0'), in_proj_covar=tensor([0.0799, 0.0469, 0.0555, 0.0615, 0.0814, 0.0568, 0.0449, 0.0546], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 14:44:18,564 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6174, 2.7472, 2.6518, 2.4365, 2.7411, 2.5276, 2.8006, 1.9316], device='cuda:0'), covar=tensor([0.0814, 0.1809, 0.2149, 0.4655, 0.1280, 0.2529, 0.1399, 0.5709], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0158, 0.0170, 0.0232, 0.0131, 0.0227, 0.0145, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 14:44:21,757 INFO [train2.py:843] (0/4) Epoch 17, validation: ctc_loss=0.04327, att_loss=0.2362, loss=0.1976, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 14:44:21,758 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 14:45:20,142 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63776.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:45:23,452 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63778.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:45:44,536 INFO [train2.py:809] (0/4) Epoch 17, batch 50, loss[ctc_loss=0.06368, att_loss=0.2378, loss=0.203, over 17296.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01145, over 55.00 utterances.], tot_loss[ctc_loss=0.08727, att_loss=0.247, loss=0.2151, over 745790.99 frames. utt_duration=1151 frames, utt_pad_proportion=0.07226, over 2594.49 utterances.], batch size: 55, lr: 6.46e-03, grad_scale: 8.0 2023-03-08 14:46:16,969 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63810.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:46:27,064 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0788, 6.2776, 5.7333, 6.0554, 5.9994, 5.5099, 5.7550, 5.5323], device='cuda:0'), covar=tensor([0.1237, 0.0882, 0.0928, 0.0845, 0.0929, 0.1467, 0.2121, 0.2513], device='cuda:0'), in_proj_covar=tensor([0.0489, 0.0560, 0.0428, 0.0430, 0.0405, 0.0450, 0.0578, 0.0502], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 14:46:50,924 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.085e+02 2.041e+02 2.465e+02 3.095e+02 5.834e+02, threshold=4.929e+02, percent-clipped=2.0 2023-03-08 14:47:08,059 INFO [train2.py:809] (0/4) Epoch 17, batch 100, loss[ctc_loss=0.08756, att_loss=0.2518, loss=0.219, over 16886.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006568, over 49.00 utterances.], tot_loss[ctc_loss=0.08495, att_loss=0.2424, loss=0.2109, over 1298966.90 frames. utt_duration=1206 frames, utt_pad_proportion=0.06659, over 4314.64 utterances.], batch size: 49, lr: 6.46e-03, grad_scale: 8.0 2023-03-08 14:47:16,173 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63845.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:47:58,779 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63871.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:48:08,249 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=63877.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:48:31,339 INFO [train2.py:809] (0/4) Epoch 17, batch 150, loss[ctc_loss=0.09534, att_loss=0.2448, loss=0.2149, over 17101.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01595, over 56.00 utterances.], tot_loss[ctc_loss=0.0856, att_loss=0.2426, loss=0.2112, over 1738170.73 frames. utt_duration=1163 frames, utt_pad_proportion=0.07598, over 5986.63 utterances.], batch size: 56, lr: 6.46e-03, grad_scale: 4.0 2023-03-08 14:48:56,205 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63906.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:49:12,676 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=63916.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:49:26,793 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:49:26,899 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63925.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:49:37,542 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 1.983e+02 2.552e+02 2.961e+02 5.419e+02, threshold=5.104e+02, percent-clipped=2.0 2023-03-08 14:49:42,791 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0911, 4.5191, 4.2435, 4.6080, 2.8311, 4.5659, 2.6662, 1.9650], device='cuda:0'), covar=tensor([0.0384, 0.0182, 0.0709, 0.0176, 0.1595, 0.0135, 0.1463, 0.1599], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0139, 0.0254, 0.0134, 0.0219, 0.0122, 0.0228, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 14:49:52,616 INFO [train2.py:809] (0/4) Epoch 17, batch 200, loss[ctc_loss=0.05981, att_loss=0.2096, loss=0.1796, over 15630.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.01, over 37.00 utterances.], tot_loss[ctc_loss=0.08452, att_loss=0.2419, loss=0.2104, over 2086689.64 frames. utt_duration=1201 frames, utt_pad_proportion=0.06343, over 6961.17 utterances.], batch size: 37, lr: 6.45e-03, grad_scale: 4.0 2023-03-08 14:49:55,022 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=63942.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:50:23,873 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2764, 5.2099, 5.0128, 3.3064, 5.0265, 4.8584, 4.6179, 3.2763], device='cuda:0'), covar=tensor([0.0128, 0.0085, 0.0291, 0.0800, 0.0082, 0.0160, 0.0250, 0.1022], device='cuda:0'), in_proj_covar=tensor([0.0070, 0.0096, 0.0094, 0.0109, 0.0079, 0.0106, 0.0098, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 14:50:52,518 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=63977.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:51:15,257 INFO [train2.py:809] (0/4) Epoch 17, batch 250, loss[ctc_loss=0.09349, att_loss=0.2432, loss=0.2133, over 16629.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005215, over 47.00 utterances.], tot_loss[ctc_loss=0.08273, att_loss=0.2402, loss=0.2087, over 2352054.50 frames. utt_duration=1246 frames, utt_pad_proportion=0.05199, over 7557.84 utterances.], batch size: 47, lr: 6.45e-03, grad_scale: 4.0 2023-03-08 14:51:32,027 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-64000.pt 2023-03-08 14:52:14,224 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64023.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:52:28,263 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.064e+02 2.460e+02 3.052e+02 5.985e+02, threshold=4.919e+02, percent-clipped=1.0 2023-03-08 14:52:43,727 INFO [train2.py:809] (0/4) Epoch 17, batch 300, loss[ctc_loss=0.0643, att_loss=0.2275, loss=0.1949, over 16181.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006807, over 41.00 utterances.], tot_loss[ctc_loss=0.08341, att_loss=0.2412, loss=0.2096, over 2555436.24 frames. utt_duration=1199 frames, utt_pad_proportion=0.06413, over 8537.41 utterances.], batch size: 41, lr: 6.45e-03, grad_scale: 4.0 2023-03-08 14:53:39,693 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64076.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:53:42,872 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64078.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:53:52,331 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64084.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:54:04,130 INFO [train2.py:809] (0/4) Epoch 17, batch 350, loss[ctc_loss=0.08181, att_loss=0.2284, loss=0.1991, over 16158.00 frames. utt_duration=1578 frames, utt_pad_proportion=0.007685, over 41.00 utterances.], tot_loss[ctc_loss=0.08363, att_loss=0.2409, loss=0.2094, over 2705690.37 frames. utt_duration=1196 frames, utt_pad_proportion=0.06928, over 9062.70 utterances.], batch size: 41, lr: 6.45e-03, grad_scale: 4.0 2023-03-08 14:54:50,093 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 14:54:56,963 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64124.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:55:00,015 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64126.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:55:09,305 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 2.060e+02 2.368e+02 3.330e+02 9.339e+02, threshold=4.736e+02, percent-clipped=6.0 2023-03-08 14:55:25,122 INFO [train2.py:809] (0/4) Epoch 17, batch 400, loss[ctc_loss=0.06881, att_loss=0.236, loss=0.2026, over 17553.00 frames. utt_duration=890.3 frames, utt_pad_proportion=0.06873, over 79.00 utterances.], tot_loss[ctc_loss=0.08297, att_loss=0.2404, loss=0.2089, over 2830744.00 frames. utt_duration=1208 frames, utt_pad_proportion=0.06461, over 9388.63 utterances.], batch size: 79, lr: 6.44e-03, grad_scale: 8.0 2023-03-08 14:56:02,636 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9099, 2.2671, 2.3943, 2.3684, 2.7468, 2.6418, 1.9877, 2.8676], device='cuda:0'), covar=tensor([0.1952, 0.3473, 0.2914, 0.2038, 0.1858, 0.1545, 0.3517, 0.1251], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0101, 0.0106, 0.0095, 0.0098, 0.0087, 0.0108, 0.0077], device='cuda:0'), out_proj_covar=tensor([6.9952e-05, 7.6646e-05, 8.0651e-05, 7.0692e-05, 7.1024e-05, 6.8552e-05, 7.9045e-05, 6.1784e-05], device='cuda:0') 2023-03-08 14:56:06,656 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-03-08 14:56:07,165 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64166.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:56:36,062 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2067, 4.6819, 4.8134, 4.6967, 2.8682, 4.7568, 3.1796, 2.0717], device='cuda:0'), covar=tensor([0.0398, 0.0189, 0.0547, 0.0185, 0.1515, 0.0153, 0.1218, 0.1630], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0141, 0.0256, 0.0135, 0.0218, 0.0123, 0.0228, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 14:56:47,838 INFO [train2.py:809] (0/4) Epoch 17, batch 450, loss[ctc_loss=0.08102, att_loss=0.2342, loss=0.2036, over 16276.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007463, over 43.00 utterances.], tot_loss[ctc_loss=0.08274, att_loss=0.2401, loss=0.2086, over 2923206.98 frames. utt_duration=1210 frames, utt_pad_proportion=0.06655, over 9675.51 utterances.], batch size: 43, lr: 6.44e-03, grad_scale: 8.0 2023-03-08 14:57:04,187 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64201.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:57:42,924 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64225.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:57:53,581 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.502e+02 2.140e+02 2.597e+02 3.193e+02 5.967e+02, threshold=5.195e+02, percent-clipped=4.0 2023-03-08 14:57:53,972 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64232.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:58:09,210 INFO [train2.py:809] (0/4) Epoch 17, batch 500, loss[ctc_loss=0.08175, att_loss=0.2505, loss=0.2168, over 16797.00 frames. utt_duration=687.1 frames, utt_pad_proportion=0.139, over 98.00 utterances.], tot_loss[ctc_loss=0.08245, att_loss=0.2404, loss=0.2088, over 3009995.32 frames. utt_duration=1239 frames, utt_pad_proportion=0.05615, over 9731.97 utterances.], batch size: 98, lr: 6.44e-03, grad_scale: 8.0 2023-03-08 14:58:11,147 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64242.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:58:58,165 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64272.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:58:59,753 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64273.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:59:00,871 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 14:59:27,309 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64290.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 14:59:29,440 INFO [train2.py:809] (0/4) Epoch 17, batch 550, loss[ctc_loss=0.1068, att_loss=0.2585, loss=0.2281, over 16983.00 frames. utt_duration=687.6 frames, utt_pad_proportion=0.1351, over 99.00 utterances.], tot_loss[ctc_loss=0.08318, att_loss=0.241, loss=0.2094, over 3069427.85 frames. utt_duration=1232 frames, utt_pad_proportion=0.05731, over 9979.02 utterances.], batch size: 99, lr: 6.44e-03, grad_scale: 8.0 2023-03-08 14:59:33,720 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64293.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:00:16,528 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5224, 4.9767, 4.8014, 5.0166, 5.0033, 4.6713, 3.4195, 4.8903], device='cuda:0'), covar=tensor([0.0127, 0.0103, 0.0125, 0.0066, 0.0107, 0.0106, 0.0698, 0.0206], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0082, 0.0102, 0.0064, 0.0068, 0.0080, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 15:00:26,241 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-08 15:00:34,620 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 1.879e+02 2.342e+02 2.961e+02 5.931e+02, threshold=4.683e+02, percent-clipped=3.0 2023-03-08 15:00:50,374 INFO [train2.py:809] (0/4) Epoch 17, batch 600, loss[ctc_loss=0.1078, att_loss=0.2576, loss=0.2276, over 17092.00 frames. utt_duration=1222 frames, utt_pad_proportion=0.01659, over 56.00 utterances.], tot_loss[ctc_loss=0.08314, att_loss=0.2403, loss=0.2088, over 3105586.39 frames. utt_duration=1237 frames, utt_pad_proportion=0.05816, over 10050.87 utterances.], batch size: 56, lr: 6.43e-03, grad_scale: 8.0 2023-03-08 15:00:57,626 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64345.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:01:51,586 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64379.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:02:12,268 INFO [train2.py:809] (0/4) Epoch 17, batch 650, loss[ctc_loss=0.1062, att_loss=0.2394, loss=0.2127, over 15946.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.006242, over 41.00 utterances.], tot_loss[ctc_loss=0.08374, att_loss=0.2406, loss=0.2092, over 3142804.74 frames. utt_duration=1221 frames, utt_pad_proportion=0.06182, over 10305.17 utterances.], batch size: 41, lr: 6.43e-03, grad_scale: 8.0 2023-03-08 15:02:36,707 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64406.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:03:17,365 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.002e+02 2.471e+02 3.095e+02 7.929e+02, threshold=4.943e+02, percent-clipped=8.0 2023-03-08 15:03:29,027 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 15:03:32,609 INFO [train2.py:809] (0/4) Epoch 17, batch 700, loss[ctc_loss=0.09298, att_loss=0.2489, loss=0.2177, over 17284.00 frames. utt_duration=1173 frames, utt_pad_proportion=0.02403, over 59.00 utterances.], tot_loss[ctc_loss=0.08411, att_loss=0.2406, loss=0.2093, over 3170056.34 frames. utt_duration=1234 frames, utt_pad_proportion=0.05971, over 10290.24 utterances.], batch size: 59, lr: 6.43e-03, grad_scale: 8.0 2023-03-08 15:04:12,644 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64465.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:04:14,086 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64466.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:04:55,320 INFO [train2.py:809] (0/4) Epoch 17, batch 750, loss[ctc_loss=0.05538, att_loss=0.2166, loss=0.1844, over 16000.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.008297, over 40.00 utterances.], tot_loss[ctc_loss=0.08435, att_loss=0.2407, loss=0.2094, over 3193403.62 frames. utt_duration=1203 frames, utt_pad_proportion=0.06688, over 10631.03 utterances.], batch size: 40, lr: 6.43e-03, grad_scale: 8.0 2023-03-08 15:05:11,186 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64501.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:05:31,705 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64514.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:05:31,808 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9041, 5.1598, 5.3866, 5.2711, 5.3389, 5.8100, 5.1229, 5.9285], device='cuda:0'), covar=tensor([0.0586, 0.0732, 0.0751, 0.1255, 0.1770, 0.0857, 0.0659, 0.0637], device='cuda:0'), in_proj_covar=tensor([0.0786, 0.0467, 0.0553, 0.0608, 0.0808, 0.0565, 0.0444, 0.0550], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 15:05:41,765 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5616, 2.4651, 5.0664, 4.0297, 3.0213, 4.2466, 4.9655, 4.5854], device='cuda:0'), covar=tensor([0.0272, 0.1761, 0.0225, 0.0922, 0.1874, 0.0244, 0.0115, 0.0274], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0239, 0.0164, 0.0304, 0.0261, 0.0195, 0.0144, 0.0173], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 15:05:51,402 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64526.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:06:01,020 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.298e+02 2.805e+02 3.347e+02 6.423e+02, threshold=5.609e+02, percent-clipped=6.0 2023-03-08 15:06:16,298 INFO [train2.py:809] (0/4) Epoch 17, batch 800, loss[ctc_loss=0.06753, att_loss=0.2203, loss=0.1898, over 16121.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006695, over 42.00 utterances.], tot_loss[ctc_loss=0.08398, att_loss=0.2407, loss=0.2093, over 3211984.50 frames. utt_duration=1222 frames, utt_pad_proportion=0.06159, over 10530.19 utterances.], batch size: 42, lr: 6.42e-03, grad_scale: 8.0 2023-03-08 15:06:28,759 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64549.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:07:05,761 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64572.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:07:32,187 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64588.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:07:33,842 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64589.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:07:36,533 INFO [train2.py:809] (0/4) Epoch 17, batch 850, loss[ctc_loss=0.1096, att_loss=0.2571, loss=0.2276, over 17142.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01376, over 56.00 utterances.], tot_loss[ctc_loss=0.08429, att_loss=0.2411, loss=0.2098, over 3229701.91 frames. utt_duration=1223 frames, utt_pad_proportion=0.05923, over 10573.39 utterances.], batch size: 56, lr: 6.42e-03, grad_scale: 8.0 2023-03-08 15:07:45,220 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5112, 4.6183, 4.6301, 4.5128, 5.2223, 4.6753, 4.5796, 2.4108], device='cuda:0'), covar=tensor([0.0240, 0.0285, 0.0264, 0.0269, 0.0664, 0.0194, 0.0276, 0.1954], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0158, 0.0164, 0.0178, 0.0357, 0.0138, 0.0149, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 15:08:11,605 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7328, 5.0408, 4.5779, 5.0464, 4.4680, 4.7079, 5.0959, 4.9070], device='cuda:0'), covar=tensor([0.0577, 0.0237, 0.0762, 0.0292, 0.0446, 0.0321, 0.0219, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0371, 0.0294, 0.0349, 0.0314, 0.0302, 0.0223, 0.0282, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 15:08:24,156 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64620.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:08:43,542 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.557e+02 2.027e+02 2.349e+02 3.075e+02 7.936e+02, threshold=4.698e+02, percent-clipped=3.0 2023-03-08 15:08:58,461 INFO [train2.py:809] (0/4) Epoch 17, batch 900, loss[ctc_loss=0.05185, att_loss=0.2078, loss=0.1766, over 15653.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.008466, over 37.00 utterances.], tot_loss[ctc_loss=0.08427, att_loss=0.2409, loss=0.2096, over 3237908.03 frames. utt_duration=1239 frames, utt_pad_proportion=0.05678, over 10465.43 utterances.], batch size: 37, lr: 6.42e-03, grad_scale: 8.0 2023-03-08 15:09:10,699 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64648.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:09:14,084 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64650.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:10:01,386 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64679.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:10:21,891 INFO [train2.py:809] (0/4) Epoch 17, batch 950, loss[ctc_loss=0.06625, att_loss=0.2368, loss=0.2027, over 17371.00 frames. utt_duration=1179 frames, utt_pad_proportion=0.01964, over 59.00 utterances.], tot_loss[ctc_loss=0.08356, att_loss=0.2402, loss=0.2088, over 3232190.85 frames. utt_duration=1216 frames, utt_pad_proportion=0.06737, over 10643.44 utterances.], batch size: 59, lr: 6.42e-03, grad_scale: 8.0 2023-03-08 15:10:38,217 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64701.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:10:42,270 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64703.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:10:51,691 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64709.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:11:19,675 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64727.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:11:28,017 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.133e+02 2.001e+02 2.338e+02 2.858e+02 5.487e+02, threshold=4.676e+02, percent-clipped=1.0 2023-03-08 15:11:36,662 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9851, 5.2407, 5.1935, 5.2086, 5.2399, 5.2221, 4.9765, 4.6587], device='cuda:0'), covar=tensor([0.0943, 0.0481, 0.0257, 0.0409, 0.0298, 0.0305, 0.0364, 0.0366], device='cuda:0'), in_proj_covar=tensor([0.0496, 0.0335, 0.0311, 0.0328, 0.0391, 0.0406, 0.0332, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 15:11:42,851 INFO [train2.py:809] (0/4) Epoch 17, batch 1000, loss[ctc_loss=0.08058, att_loss=0.2499, loss=0.216, over 16773.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.005493, over 48.00 utterances.], tot_loss[ctc_loss=0.08349, att_loss=0.2403, loss=0.2089, over 3242455.75 frames. utt_duration=1228 frames, utt_pad_proportion=0.06417, over 10577.24 utterances.], batch size: 48, lr: 6.41e-03, grad_scale: 8.0 2023-03-08 15:11:49,878 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64745.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:12:01,460 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64752.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:12:20,054 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64764.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:13:02,951 INFO [train2.py:809] (0/4) Epoch 17, batch 1050, loss[ctc_loss=0.06606, att_loss=0.2161, loss=0.1861, over 15628.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009827, over 37.00 utterances.], tot_loss[ctc_loss=0.08308, att_loss=0.2401, loss=0.2087, over 3249839.75 frames. utt_duration=1247 frames, utt_pad_proportion=0.05861, over 10436.73 utterances.], batch size: 37, lr: 6.41e-03, grad_scale: 8.0 2023-03-08 15:13:12,100 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-08 15:13:27,784 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64806.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:13:38,361 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=64813.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:13:50,481 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64821.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:14:02,866 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.4138, 5.6990, 5.1867, 5.5161, 5.3218, 5.0029, 5.1414, 4.9578], device='cuda:0'), covar=tensor([0.1271, 0.0879, 0.0861, 0.0804, 0.0961, 0.1227, 0.2158, 0.2183], device='cuda:0'), in_proj_covar=tensor([0.0496, 0.0568, 0.0428, 0.0428, 0.0413, 0.0451, 0.0582, 0.0508], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 15:14:08,045 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.316e+02 2.188e+02 2.644e+02 3.168e+02 6.507e+02, threshold=5.288e+02, percent-clipped=8.0 2023-03-08 15:14:22,624 INFO [train2.py:809] (0/4) Epoch 17, batch 1100, loss[ctc_loss=0.06813, att_loss=0.2422, loss=0.2074, over 17398.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03311, over 63.00 utterances.], tot_loss[ctc_loss=0.08324, att_loss=0.2403, loss=0.2089, over 3245094.11 frames. utt_duration=1230 frames, utt_pad_proportion=0.0627, over 10564.43 utterances.], batch size: 63, lr: 6.41e-03, grad_scale: 8.0 2023-03-08 15:15:37,896 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=64888.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:15:42,102 INFO [train2.py:809] (0/4) Epoch 17, batch 1150, loss[ctc_loss=0.09168, att_loss=0.2566, loss=0.2236, over 17292.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01241, over 55.00 utterances.], tot_loss[ctc_loss=0.08321, att_loss=0.2401, loss=0.2087, over 3252045.74 frames. utt_duration=1247 frames, utt_pad_proportion=0.05825, over 10443.48 utterances.], batch size: 55, lr: 6.41e-03, grad_scale: 8.0 2023-03-08 15:16:40,338 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3171, 4.4380, 4.4219, 4.2835, 4.9894, 4.4935, 4.3879, 2.4437], device='cuda:0'), covar=tensor([0.0282, 0.0311, 0.0306, 0.0280, 0.0775, 0.0210, 0.0292, 0.1835], device='cuda:0'), in_proj_covar=tensor([0.0140, 0.0157, 0.0164, 0.0177, 0.0356, 0.0137, 0.0148, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-08 15:16:47,808 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.183e+02 2.059e+02 2.511e+02 2.874e+02 5.628e+02, threshold=5.021e+02, percent-clipped=1.0 2023-03-08 15:16:54,651 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=64936.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:17:02,368 INFO [train2.py:809] (0/4) Epoch 17, batch 1200, loss[ctc_loss=0.064, att_loss=0.2269, loss=0.1943, over 15958.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006284, over 41.00 utterances.], tot_loss[ctc_loss=0.08262, att_loss=0.2397, loss=0.2083, over 3253633.74 frames. utt_duration=1262 frames, utt_pad_proportion=0.05488, over 10327.16 utterances.], batch size: 41, lr: 6.40e-03, grad_scale: 8.0 2023-03-08 15:17:09,404 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=64945.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:17:58,147 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=64976.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:18:22,345 INFO [train2.py:809] (0/4) Epoch 17, batch 1250, loss[ctc_loss=0.0688, att_loss=0.2252, loss=0.1939, over 16391.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.007066, over 44.00 utterances.], tot_loss[ctc_loss=0.08274, att_loss=0.24, loss=0.2085, over 3262679.41 frames. utt_duration=1258 frames, utt_pad_proportion=0.05294, over 10387.97 utterances.], batch size: 44, lr: 6.40e-03, grad_scale: 8.0 2023-03-08 15:18:39,100 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65001.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:18:44,323 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65004.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:19:25,183 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-03-08 15:19:27,892 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.065e+02 2.436e+02 3.098e+02 5.553e+02, threshold=4.871e+02, percent-clipped=2.0 2023-03-08 15:19:36,252 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65037.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 15:19:41,382 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-03-08 15:19:41,905 INFO [train2.py:809] (0/4) Epoch 17, batch 1300, loss[ctc_loss=0.06422, att_loss=0.222, loss=0.1905, over 15952.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.005806, over 41.00 utterances.], tot_loss[ctc_loss=0.08304, att_loss=0.2404, loss=0.2089, over 3257186.83 frames. utt_duration=1243 frames, utt_pad_proportion=0.05795, over 10493.80 utterances.], batch size: 41, lr: 6.40e-03, grad_scale: 8.0 2023-03-08 15:19:54,988 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65049.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:20:05,123 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65055.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:20:11,143 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65059.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:21:01,330 INFO [train2.py:809] (0/4) Epoch 17, batch 1350, loss[ctc_loss=0.07555, att_loss=0.2429, loss=0.2094, over 17329.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.0106, over 55.00 utterances.], tot_loss[ctc_loss=0.08342, att_loss=0.2408, loss=0.2093, over 3266938.41 frames. utt_duration=1233 frames, utt_pad_proportion=0.05797, over 10610.26 utterances.], batch size: 55, lr: 6.40e-03, grad_scale: 8.0 2023-03-08 15:21:18,511 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65101.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:21:29,239 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65108.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:21:41,820 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65116.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:21:46,172 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8016, 6.0385, 5.4389, 5.8341, 5.7146, 5.2685, 5.4483, 5.2494], device='cuda:0'), covar=tensor([0.1200, 0.0924, 0.0948, 0.0762, 0.0902, 0.1397, 0.2238, 0.2259], device='cuda:0'), in_proj_covar=tensor([0.0499, 0.0573, 0.0433, 0.0433, 0.0416, 0.0457, 0.0588, 0.0513], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 15:21:49,413 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65121.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:22:07,568 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 2.057e+02 2.538e+02 3.006e+02 4.389e+02, threshold=5.076e+02, percent-clipped=0.0 2023-03-08 15:22:22,265 INFO [train2.py:809] (0/4) Epoch 17, batch 1400, loss[ctc_loss=0.07813, att_loss=0.2342, loss=0.203, over 16008.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007141, over 40.00 utterances.], tot_loss[ctc_loss=0.08313, att_loss=0.2406, loss=0.2091, over 3272608.96 frames. utt_duration=1235 frames, utt_pad_proportion=0.05553, over 10614.75 utterances.], batch size: 40, lr: 6.39e-03, grad_scale: 8.0 2023-03-08 15:23:06,259 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65169.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:23:41,172 INFO [train2.py:809] (0/4) Epoch 17, batch 1450, loss[ctc_loss=0.07067, att_loss=0.2111, loss=0.183, over 15639.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009162, over 37.00 utterances.], tot_loss[ctc_loss=0.08307, att_loss=0.2402, loss=0.2088, over 3268577.10 frames. utt_duration=1267 frames, utt_pad_proportion=0.04912, over 10334.82 utterances.], batch size: 37, lr: 6.39e-03, grad_scale: 8.0 2023-03-08 15:23:46,643 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1369, 4.3193, 4.1387, 4.7844, 2.5462, 4.6507, 2.4071, 1.9411], device='cuda:0'), covar=tensor([0.0380, 0.0255, 0.0769, 0.0137, 0.1759, 0.0151, 0.1729, 0.1778], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0144, 0.0260, 0.0137, 0.0221, 0.0125, 0.0232, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 15:24:46,654 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.514e+02 2.067e+02 2.518e+02 3.214e+02 7.164e+02, threshold=5.035e+02, percent-clipped=4.0 2023-03-08 15:24:58,506 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2158, 4.5087, 4.4744, 4.9051, 2.6600, 4.8006, 2.5708, 2.4618], device='cuda:0'), covar=tensor([0.0401, 0.0251, 0.0681, 0.0146, 0.1796, 0.0152, 0.1762, 0.1556], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0145, 0.0261, 0.0138, 0.0221, 0.0125, 0.0233, 0.0207], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 15:25:01,178 INFO [train2.py:809] (0/4) Epoch 17, batch 1500, loss[ctc_loss=0.07607, att_loss=0.2359, loss=0.2039, over 16290.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.005959, over 43.00 utterances.], tot_loss[ctc_loss=0.08264, att_loss=0.2403, loss=0.2088, over 3272102.89 frames. utt_duration=1263 frames, utt_pad_proportion=0.05057, over 10375.73 utterances.], batch size: 43, lr: 6.39e-03, grad_scale: 8.0 2023-03-08 15:25:08,252 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65245.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:25:34,933 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.45 vs. limit=5.0 2023-03-08 15:25:37,498 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8907, 5.1724, 5.4509, 5.3213, 5.3311, 5.8270, 5.1330, 5.9567], device='cuda:0'), covar=tensor([0.0701, 0.0770, 0.0867, 0.1137, 0.1893, 0.0889, 0.0694, 0.0656], device='cuda:0'), in_proj_covar=tensor([0.0798, 0.0470, 0.0557, 0.0615, 0.0814, 0.0568, 0.0452, 0.0555], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 15:26:11,694 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65285.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:26:21,177 INFO [train2.py:809] (0/4) Epoch 17, batch 1550, loss[ctc_loss=0.0664, att_loss=0.2263, loss=0.1943, over 15952.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.00726, over 41.00 utterances.], tot_loss[ctc_loss=0.08263, att_loss=0.2406, loss=0.209, over 3274328.26 frames. utt_duration=1272 frames, utt_pad_proportion=0.04793, over 10308.32 utterances.], batch size: 41, lr: 6.39e-03, grad_scale: 8.0 2023-03-08 15:26:24,397 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65293.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:26:42,496 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65304.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:27:27,594 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.558e+02 2.050e+02 2.369e+02 2.748e+02 8.749e+02, threshold=4.738e+02, percent-clipped=3.0 2023-03-08 15:27:27,824 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65332.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:27:41,713 INFO [train2.py:809] (0/4) Epoch 17, batch 1600, loss[ctc_loss=0.08457, att_loss=0.2496, loss=0.2166, over 17257.00 frames. utt_duration=875 frames, utt_pad_proportion=0.08284, over 79.00 utterances.], tot_loss[ctc_loss=0.08247, att_loss=0.24, loss=0.2085, over 3268420.06 frames. utt_duration=1276 frames, utt_pad_proportion=0.04909, over 10254.91 utterances.], batch size: 79, lr: 6.38e-03, grad_scale: 8.0 2023-03-08 15:27:50,407 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65346.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:27:54,285 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3447, 4.4749, 4.5545, 4.4074, 5.0319, 4.4358, 4.5202, 2.3680], device='cuda:0'), covar=tensor([0.0275, 0.0269, 0.0251, 0.0261, 0.0840, 0.0244, 0.0269, 0.2114], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0160, 0.0167, 0.0181, 0.0362, 0.0140, 0.0151, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 15:28:00,023 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65352.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:28:10,890 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65359.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:28:33,724 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4525, 2.5790, 4.9015, 3.7568, 2.8929, 4.1988, 4.7395, 4.5122], device='cuda:0'), covar=tensor([0.0225, 0.1702, 0.0199, 0.1010, 0.1798, 0.0239, 0.0136, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0243, 0.0169, 0.0314, 0.0266, 0.0200, 0.0150, 0.0178], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 15:29:02,426 INFO [train2.py:809] (0/4) Epoch 17, batch 1650, loss[ctc_loss=0.0666, att_loss=0.2327, loss=0.1995, over 16521.00 frames. utt_duration=1470 frames, utt_pad_proportion=0.006562, over 45.00 utterances.], tot_loss[ctc_loss=0.08183, att_loss=0.2393, loss=0.2078, over 3267597.97 frames. utt_duration=1281 frames, utt_pad_proportion=0.04791, over 10218.85 utterances.], batch size: 45, lr: 6.38e-03, grad_scale: 8.0 2023-03-08 15:29:20,656 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65401.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:29:29,998 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65407.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:29:31,711 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65408.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:29:36,199 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65411.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:30:06,595 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2982, 4.6844, 4.9656, 4.7510, 4.7804, 5.2319, 4.7822, 5.3413], device='cuda:0'), covar=tensor([0.0763, 0.0784, 0.0733, 0.1177, 0.1977, 0.0893, 0.1029, 0.0643], device='cuda:0'), in_proj_covar=tensor([0.0807, 0.0474, 0.0565, 0.0622, 0.0827, 0.0575, 0.0461, 0.0557], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 15:30:09,496 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.074e+02 2.531e+02 3.179e+02 8.424e+02, threshold=5.062e+02, percent-clipped=4.0 2023-03-08 15:30:23,564 INFO [train2.py:809] (0/4) Epoch 17, batch 1700, loss[ctc_loss=0.1034, att_loss=0.2497, loss=0.2204, over 16881.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.006891, over 49.00 utterances.], tot_loss[ctc_loss=0.08203, att_loss=0.2393, loss=0.2078, over 3269612.54 frames. utt_duration=1258 frames, utt_pad_proportion=0.0521, over 10404.83 utterances.], batch size: 49, lr: 6.38e-03, grad_scale: 8.0 2023-03-08 15:30:37,474 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65449.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:30:48,515 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65456.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:31:44,387 INFO [train2.py:809] (0/4) Epoch 17, batch 1750, loss[ctc_loss=0.08625, att_loss=0.2484, loss=0.216, over 16956.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.00802, over 50.00 utterances.], tot_loss[ctc_loss=0.08266, att_loss=0.2397, loss=0.2083, over 3269469.76 frames. utt_duration=1258 frames, utt_pad_proportion=0.05277, over 10410.32 utterances.], batch size: 50, lr: 6.38e-03, grad_scale: 8.0 2023-03-08 15:32:40,963 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9403, 2.3450, 2.5640, 2.7115, 2.6195, 2.9209, 2.5297, 2.8903], device='cuda:0'), covar=tensor([0.1115, 0.3082, 0.2248, 0.1329, 0.1472, 0.1037, 0.2096, 0.1048], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0105, 0.0109, 0.0094, 0.0101, 0.0088, 0.0108, 0.0078], device='cuda:0'), out_proj_covar=tensor([7.1475e-05, 7.9506e-05, 8.2890e-05, 7.0911e-05, 7.3632e-05, 6.9915e-05, 7.9262e-05, 6.2688e-05], device='cuda:0') 2023-03-08 15:32:49,897 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.165e+02 2.005e+02 2.493e+02 3.105e+02 6.516e+02, threshold=4.986e+02, percent-clipped=3.0 2023-03-08 15:32:50,238 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65532.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 15:33:05,036 INFO [train2.py:809] (0/4) Epoch 17, batch 1800, loss[ctc_loss=0.09527, att_loss=0.2596, loss=0.2267, over 17316.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02275, over 59.00 utterances.], tot_loss[ctc_loss=0.08218, att_loss=0.2393, loss=0.2079, over 3269264.17 frames. utt_duration=1255 frames, utt_pad_proportion=0.05405, over 10430.92 utterances.], batch size: 59, lr: 6.37e-03, grad_scale: 8.0 2023-03-08 15:34:26,297 INFO [train2.py:809] (0/4) Epoch 17, batch 1850, loss[ctc_loss=0.07167, att_loss=0.2437, loss=0.2093, over 16623.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005469, over 47.00 utterances.], tot_loss[ctc_loss=0.08314, att_loss=0.2403, loss=0.2089, over 3273025.37 frames. utt_duration=1254 frames, utt_pad_proportion=0.05381, over 10453.59 utterances.], batch size: 47, lr: 6.37e-03, grad_scale: 8.0 2023-03-08 15:34:29,906 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65593.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 15:34:45,688 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5675, 4.8636, 4.4212, 4.9430, 4.3396, 4.5455, 4.9536, 4.7828], device='cuda:0'), covar=tensor([0.0598, 0.0317, 0.0819, 0.0288, 0.0450, 0.0383, 0.0249, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0370, 0.0296, 0.0347, 0.0311, 0.0300, 0.0224, 0.0279, 0.0264], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0005, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 15:35:31,591 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.506e+02 1.987e+02 2.376e+02 2.824e+02 5.455e+02, threshold=4.753e+02, percent-clipped=1.0 2023-03-08 15:35:31,921 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65632.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:35:46,839 INFO [train2.py:809] (0/4) Epoch 17, batch 1900, loss[ctc_loss=0.09005, att_loss=0.2475, loss=0.216, over 17398.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03304, over 63.00 utterances.], tot_loss[ctc_loss=0.08447, att_loss=0.2415, loss=0.2101, over 3261455.42 frames. utt_duration=1172 frames, utt_pad_proportion=0.07831, over 11145.80 utterances.], batch size: 63, lr: 6.37e-03, grad_scale: 8.0 2023-03-08 15:35:47,056 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65641.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:35:55,263 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-03-08 15:36:48,030 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65680.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:37:05,991 INFO [train2.py:809] (0/4) Epoch 17, batch 1950, loss[ctc_loss=0.09466, att_loss=0.253, loss=0.2213, over 16523.00 frames. utt_duration=1470 frames, utt_pad_proportion=0.005935, over 45.00 utterances.], tot_loss[ctc_loss=0.08396, att_loss=0.241, loss=0.2096, over 3267994.00 frames. utt_duration=1187 frames, utt_pad_proportion=0.07204, over 11022.34 utterances.], batch size: 45, lr: 6.37e-03, grad_scale: 8.0 2023-03-08 15:37:36,732 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65711.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:38:10,052 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.350e+02 2.136e+02 2.531e+02 3.248e+02 6.819e+02, threshold=5.061e+02, percent-clipped=5.0 2023-03-08 15:38:25,204 INFO [train2.py:809] (0/4) Epoch 17, batch 2000, loss[ctc_loss=0.1032, att_loss=0.2625, loss=0.2306, over 16321.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006642, over 45.00 utterances.], tot_loss[ctc_loss=0.08334, att_loss=0.2405, loss=0.209, over 3258281.57 frames. utt_duration=1206 frames, utt_pad_proportion=0.06833, over 10816.06 utterances.], batch size: 45, lr: 6.37e-03, grad_scale: 8.0 2023-03-08 15:38:53,011 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65759.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:39:04,630 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1392, 4.5994, 4.4024, 4.6193, 2.5274, 4.5584, 2.3242, 1.6252], device='cuda:0'), covar=tensor([0.0343, 0.0168, 0.0737, 0.0162, 0.1948, 0.0160, 0.1827, 0.1945], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0145, 0.0262, 0.0139, 0.0223, 0.0126, 0.0233, 0.0208], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 15:39:44,926 INFO [train2.py:809] (0/4) Epoch 17, batch 2050, loss[ctc_loss=0.09776, att_loss=0.259, loss=0.2268, over 14333.00 frames. utt_duration=394.1 frames, utt_pad_proportion=0.3121, over 146.00 utterances.], tot_loss[ctc_loss=0.08278, att_loss=0.2399, loss=0.2085, over 3260385.32 frames. utt_duration=1219 frames, utt_pad_proportion=0.06424, over 10712.95 utterances.], batch size: 146, lr: 6.36e-03, grad_scale: 8.0 2023-03-08 15:40:51,675 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.094e+02 1.921e+02 2.500e+02 3.054e+02 5.987e+02, threshold=5.000e+02, percent-clipped=2.0 2023-03-08 15:41:01,794 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=65838.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:41:06,586 INFO [train2.py:809] (0/4) Epoch 17, batch 2100, loss[ctc_loss=0.06254, att_loss=0.2444, loss=0.208, over 16472.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.00697, over 46.00 utterances.], tot_loss[ctc_loss=0.08304, att_loss=0.24, loss=0.2086, over 3261688.94 frames. utt_duration=1235 frames, utt_pad_proportion=0.06045, over 10581.08 utterances.], batch size: 46, lr: 6.36e-03, grad_scale: 8.0 2023-03-08 15:41:35,075 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0061, 5.1597, 5.1601, 5.1889, 5.2590, 5.2237, 4.9636, 4.7348], device='cuda:0'), covar=tensor([0.0970, 0.0585, 0.0290, 0.0438, 0.0324, 0.0338, 0.0348, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0496, 0.0334, 0.0312, 0.0329, 0.0390, 0.0405, 0.0328, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 15:42:06,868 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3629, 2.6821, 3.1392, 4.4970, 4.1587, 4.0578, 3.0333, 2.3667], device='cuda:0'), covar=tensor([0.0696, 0.2166, 0.1088, 0.0556, 0.0646, 0.0389, 0.1366, 0.2030], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0215, 0.0187, 0.0206, 0.0211, 0.0171, 0.0197, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 15:42:21,782 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=65888.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 15:42:26,208 INFO [train2.py:809] (0/4) Epoch 17, batch 2150, loss[ctc_loss=0.08885, att_loss=0.248, loss=0.2162, over 16601.00 frames. utt_duration=1414 frames, utt_pad_proportion=0.006858, over 47.00 utterances.], tot_loss[ctc_loss=0.08348, att_loss=0.2405, loss=0.2091, over 3263517.92 frames. utt_duration=1234 frames, utt_pad_proportion=0.06006, over 10591.09 utterances.], batch size: 47, lr: 6.36e-03, grad_scale: 16.0 2023-03-08 15:42:38,855 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=65899.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:42:57,750 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1313, 5.3704, 5.3537, 5.3374, 5.4579, 5.3782, 5.1247, 4.8806], device='cuda:0'), covar=tensor([0.0903, 0.0522, 0.0238, 0.0433, 0.0254, 0.0305, 0.0304, 0.0312], device='cuda:0'), in_proj_covar=tensor([0.0495, 0.0334, 0.0311, 0.0329, 0.0389, 0.0404, 0.0328, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 15:43:31,156 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 2.187e+02 2.638e+02 3.352e+02 9.129e+02, threshold=5.276e+02, percent-clipped=7.0 2023-03-08 15:43:45,838 INFO [train2.py:809] (0/4) Epoch 17, batch 2200, loss[ctc_loss=0.05912, att_loss=0.2244, loss=0.1914, over 16626.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005349, over 47.00 utterances.], tot_loss[ctc_loss=0.0836, att_loss=0.2405, loss=0.2092, over 3268607.56 frames. utt_duration=1221 frames, utt_pad_proportion=0.06131, over 10725.41 utterances.], batch size: 47, lr: 6.36e-03, grad_scale: 16.0 2023-03-08 15:43:46,127 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=65941.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:43:47,649 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6860, 4.9294, 4.4953, 4.9995, 4.3601, 4.6309, 5.0508, 4.8666], device='cuda:0'), covar=tensor([0.0586, 0.0357, 0.0813, 0.0353, 0.0499, 0.0306, 0.0245, 0.0222], device='cuda:0'), in_proj_covar=tensor([0.0380, 0.0301, 0.0353, 0.0317, 0.0309, 0.0228, 0.0286, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 15:45:01,902 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=65989.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:45:04,874 INFO [train2.py:809] (0/4) Epoch 17, batch 2250, loss[ctc_loss=0.07241, att_loss=0.232, loss=0.2001, over 16266.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007391, over 43.00 utterances.], tot_loss[ctc_loss=0.0827, att_loss=0.24, loss=0.2086, over 3263989.27 frames. utt_duration=1238 frames, utt_pad_proportion=0.05876, over 10555.25 utterances.], batch size: 43, lr: 6.35e-03, grad_scale: 16.0 2023-03-08 15:45:18,590 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-66000.pt 2023-03-08 15:46:12,738 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.416e+02 2.093e+02 2.407e+02 3.008e+02 8.516e+02, threshold=4.813e+02, percent-clipped=4.0 2023-03-08 15:46:27,300 INFO [train2.py:809] (0/4) Epoch 17, batch 2300, loss[ctc_loss=0.1159, att_loss=0.2682, loss=0.2377, over 16684.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006041, over 46.00 utterances.], tot_loss[ctc_loss=0.08287, att_loss=0.2397, loss=0.2083, over 3259953.34 frames. utt_duration=1267 frames, utt_pad_proportion=0.05326, over 10305.42 utterances.], batch size: 46, lr: 6.35e-03, grad_scale: 8.0 2023-03-08 15:47:48,103 INFO [train2.py:809] (0/4) Epoch 17, batch 2350, loss[ctc_loss=0.08979, att_loss=0.252, loss=0.2196, over 16614.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005902, over 47.00 utterances.], tot_loss[ctc_loss=0.08383, att_loss=0.2401, loss=0.2089, over 3252010.26 frames. utt_duration=1256 frames, utt_pad_proportion=0.05768, over 10366.08 utterances.], batch size: 47, lr: 6.35e-03, grad_scale: 8.0 2023-03-08 15:48:54,077 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-08 15:48:55,989 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.531e+02 2.214e+02 2.683e+02 3.352e+02 8.000e+02, threshold=5.366e+02, percent-clipped=7.0 2023-03-08 15:49:06,335 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3105, 4.7682, 4.5584, 4.7172, 4.8493, 4.4451, 3.2790, 4.6983], device='cuda:0'), covar=tensor([0.0125, 0.0108, 0.0141, 0.0090, 0.0078, 0.0124, 0.0702, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0085, 0.0082, 0.0103, 0.0064, 0.0068, 0.0080, 0.0098, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 15:49:09,126 INFO [train2.py:809] (0/4) Epoch 17, batch 2400, loss[ctc_loss=0.05838, att_loss=0.2154, loss=0.184, over 16000.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007207, over 40.00 utterances.], tot_loss[ctc_loss=0.08324, att_loss=0.2403, loss=0.2089, over 3265603.52 frames. utt_duration=1271 frames, utt_pad_proportion=0.05158, over 10293.33 utterances.], batch size: 40, lr: 6.35e-03, grad_scale: 8.0 2023-03-08 15:49:14,417 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66144.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:49:15,009 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.65 vs. limit=2.0 2023-03-08 15:49:16,490 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-03-08 15:49:39,822 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66160.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:50:10,936 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-08 15:50:20,189 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9173, 2.1212, 2.2361, 2.5537, 2.5965, 2.4737, 2.4890, 2.8721], device='cuda:0'), covar=tensor([0.1291, 0.3160, 0.2932, 0.1665, 0.1608, 0.1274, 0.2312, 0.0851], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0104, 0.0108, 0.0093, 0.0100, 0.0087, 0.0106, 0.0078], device='cuda:0'), out_proj_covar=tensor([7.0582e-05, 7.8562e-05, 8.2237e-05, 7.0126e-05, 7.3290e-05, 6.9083e-05, 7.8381e-05, 6.2099e-05], device='cuda:0') 2023-03-08 15:50:24,677 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66188.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 15:50:29,109 INFO [train2.py:809] (0/4) Epoch 17, batch 2450, loss[ctc_loss=0.1187, att_loss=0.2573, loss=0.2295, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007843, over 44.00 utterances.], tot_loss[ctc_loss=0.08252, att_loss=0.2399, loss=0.2084, over 3261492.12 frames. utt_duration=1290 frames, utt_pad_proportion=0.04817, over 10121.59 utterances.], batch size: 44, lr: 6.34e-03, grad_scale: 8.0 2023-03-08 15:50:33,872 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66194.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:50:45,259 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0743, 4.9741, 4.9302, 2.3512, 2.0768, 2.8939, 2.0918, 3.8797], device='cuda:0'), covar=tensor([0.0644, 0.0247, 0.0259, 0.4756, 0.5396, 0.2476, 0.3555, 0.1566], device='cuda:0'), in_proj_covar=tensor([0.0343, 0.0253, 0.0251, 0.0232, 0.0341, 0.0332, 0.0240, 0.0357], device='cuda:0'), out_proj_covar=tensor([1.4791e-04, 9.3222e-05, 1.0805e-04, 9.9955e-05, 1.4399e-04, 1.3090e-04, 9.6026e-05, 1.4691e-04], device='cuda:0') 2023-03-08 15:50:50,531 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-08 15:50:51,180 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66205.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 15:51:04,683 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0589, 4.8644, 4.8649, 2.2767, 1.9192, 2.8567, 2.1400, 3.8331], device='cuda:0'), covar=tensor([0.0652, 0.0251, 0.0245, 0.4678, 0.5614, 0.2606, 0.3517, 0.1638], device='cuda:0'), in_proj_covar=tensor([0.0345, 0.0255, 0.0253, 0.0233, 0.0342, 0.0334, 0.0241, 0.0359], device='cuda:0'), out_proj_covar=tensor([1.4863e-04, 9.3847e-05, 1.0862e-04, 1.0048e-04, 1.4467e-04, 1.3159e-04, 9.6516e-05, 1.4754e-04], device='cuda:0') 2023-03-08 15:51:16,819 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66221.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 15:51:35,451 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.379e+02 1.994e+02 2.398e+02 2.990e+02 4.880e+02, threshold=4.796e+02, percent-clipped=0.0 2023-03-08 15:51:40,276 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66236.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:51:48,521 INFO [train2.py:809] (0/4) Epoch 17, batch 2500, loss[ctc_loss=0.06756, att_loss=0.2133, loss=0.1842, over 15359.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01205, over 35.00 utterances.], tot_loss[ctc_loss=0.08209, att_loss=0.2389, loss=0.2075, over 3258659.40 frames. utt_duration=1315 frames, utt_pad_proportion=0.04201, over 9926.27 utterances.], batch size: 35, lr: 6.34e-03, grad_scale: 8.0 2023-03-08 15:52:16,100 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66258.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:52:57,223 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1940, 2.7190, 3.5210, 2.6516, 3.3448, 4.3459, 4.1598, 2.9574], device='cuda:0'), covar=tensor([0.0390, 0.1826, 0.1139, 0.1597, 0.1174, 0.0894, 0.0669, 0.1432], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0241, 0.0271, 0.0215, 0.0259, 0.0354, 0.0251, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 15:53:03,918 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3334, 5.3294, 4.8894, 2.6797, 5.0396, 4.9140, 4.4803, 2.7667], device='cuda:0'), covar=tensor([0.0117, 0.0111, 0.0315, 0.1395, 0.0107, 0.0194, 0.0377, 0.1846], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0097, 0.0095, 0.0110, 0.0081, 0.0107, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 15:53:08,204 INFO [train2.py:809] (0/4) Epoch 17, batch 2550, loss[ctc_loss=0.08612, att_loss=0.2379, loss=0.2075, over 16430.00 frames. utt_duration=1496 frames, utt_pad_proportion=0.004914, over 44.00 utterances.], tot_loss[ctc_loss=0.08282, att_loss=0.2394, loss=0.2081, over 3247788.13 frames. utt_duration=1282 frames, utt_pad_proportion=0.05155, over 10146.50 utterances.], batch size: 44, lr: 6.34e-03, grad_scale: 8.0 2023-03-08 15:53:11,570 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1919, 5.1137, 4.8853, 3.0136, 4.8760, 4.6937, 4.3140, 2.9254], device='cuda:0'), covar=tensor([0.0090, 0.0111, 0.0258, 0.1007, 0.0096, 0.0204, 0.0343, 0.1266], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0097, 0.0095, 0.0110, 0.0081, 0.0107, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 15:53:52,721 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66319.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 15:54:14,465 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.777e+02 2.284e+02 2.590e+02 3.284e+02 7.249e+02, threshold=5.180e+02, percent-clipped=6.0 2023-03-08 15:54:25,275 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 15:54:27,435 INFO [train2.py:809] (0/4) Epoch 17, batch 2600, loss[ctc_loss=0.06736, att_loss=0.2382, loss=0.204, over 16680.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006235, over 46.00 utterances.], tot_loss[ctc_loss=0.08289, att_loss=0.2395, loss=0.2082, over 3250145.00 frames. utt_duration=1269 frames, utt_pad_proportion=0.05443, over 10256.15 utterances.], batch size: 46, lr: 6.34e-03, grad_scale: 8.0 2023-03-08 15:55:06,092 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5854, 3.5113, 3.4353, 3.0514, 3.5943, 3.5475, 3.5271, 2.5978], device='cuda:0'), covar=tensor([0.1112, 0.1271, 0.2033, 0.4397, 0.0876, 0.2146, 0.0958, 0.4268], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0162, 0.0175, 0.0235, 0.0138, 0.0231, 0.0152, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 15:55:47,240 INFO [train2.py:809] (0/4) Epoch 17, batch 2650, loss[ctc_loss=0.07875, att_loss=0.2634, loss=0.2265, over 17030.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007278, over 51.00 utterances.], tot_loss[ctc_loss=0.08282, att_loss=0.2397, loss=0.2083, over 3255123.04 frames. utt_duration=1267 frames, utt_pad_proportion=0.05448, over 10290.85 utterances.], batch size: 51, lr: 6.33e-03, grad_scale: 8.0 2023-03-08 15:56:19,026 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-03-08 15:56:53,519 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.144e+02 2.549e+02 3.292e+02 5.806e+02, threshold=5.098e+02, percent-clipped=4.0 2023-03-08 15:57:06,279 INFO [train2.py:809] (0/4) Epoch 17, batch 2700, loss[ctc_loss=0.08192, att_loss=0.2315, loss=0.2016, over 16379.00 frames. utt_duration=1490 frames, utt_pad_proportion=0.007717, over 44.00 utterances.], tot_loss[ctc_loss=0.0841, att_loss=0.2402, loss=0.209, over 3247392.40 frames. utt_duration=1209 frames, utt_pad_proportion=0.07167, over 10754.63 utterances.], batch size: 44, lr: 6.33e-03, grad_scale: 8.0 2023-03-08 15:57:55,904 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5594, 3.2190, 3.1551, 2.7488, 3.2198, 3.1785, 3.2858, 2.2018], device='cuda:0'), covar=tensor([0.1197, 0.1568, 0.2545, 0.4732, 0.1564, 0.3819, 0.1098, 0.5678], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0163, 0.0175, 0.0234, 0.0138, 0.0232, 0.0153, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 15:58:25,421 INFO [train2.py:809] (0/4) Epoch 17, batch 2750, loss[ctc_loss=0.0735, att_loss=0.2136, loss=0.1856, over 15879.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.008672, over 39.00 utterances.], tot_loss[ctc_loss=0.08367, att_loss=0.2409, loss=0.2095, over 3264388.61 frames. utt_duration=1209 frames, utt_pad_proportion=0.06592, over 10817.74 utterances.], batch size: 39, lr: 6.33e-03, grad_scale: 8.0 2023-03-08 15:58:30,319 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66494.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 15:58:40,069 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66500.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:58:58,532 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5950, 3.4662, 3.3248, 3.0097, 3.4147, 3.4700, 3.4212, 2.3995], device='cuda:0'), covar=tensor([0.1166, 0.1381, 0.2690, 0.4458, 0.2535, 0.2084, 0.1305, 0.5702], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0163, 0.0175, 0.0235, 0.0138, 0.0233, 0.0153, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 15:59:05,205 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66516.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 15:59:12,739 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66521.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 15:59:31,456 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.407e+02 2.169e+02 2.491e+02 3.348e+02 1.363e+03, threshold=4.982e+02, percent-clipped=6.0 2023-03-08 15:59:43,569 INFO [train2.py:809] (0/4) Epoch 17, batch 2800, loss[ctc_loss=0.088, att_loss=0.2393, loss=0.209, over 16135.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005679, over 42.00 utterances.], tot_loss[ctc_loss=0.08295, att_loss=0.2404, loss=0.2089, over 3270719.93 frames. utt_duration=1231 frames, utt_pad_proportion=0.05885, over 10645.00 utterances.], batch size: 42, lr: 6.33e-03, grad_scale: 8.0 2023-03-08 15:59:45,178 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66542.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 16:00:14,215 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6963, 2.2649, 5.1854, 4.2121, 3.2012, 4.4876, 4.9931, 4.8446], device='cuda:0'), covar=tensor([0.0202, 0.1681, 0.0147, 0.0760, 0.1658, 0.0177, 0.0094, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0242, 0.0168, 0.0310, 0.0266, 0.0199, 0.0151, 0.0179], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:00:48,727 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66582.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 16:01:02,311 INFO [train2.py:809] (0/4) Epoch 17, batch 2850, loss[ctc_loss=0.07973, att_loss=0.2434, loss=0.2107, over 16549.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005747, over 45.00 utterances.], tot_loss[ctc_loss=0.08254, att_loss=0.2395, loss=0.2081, over 3263925.58 frames. utt_duration=1235 frames, utt_pad_proportion=0.06096, over 10584.30 utterances.], batch size: 45, lr: 6.32e-03, grad_scale: 8.0 2023-03-08 16:01:39,564 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66614.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:01:57,651 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-03-08 16:02:09,211 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.066e+02 2.507e+02 3.083e+02 5.054e+02, threshold=5.014e+02, percent-clipped=1.0 2023-03-08 16:02:17,980 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-03-08 16:02:21,163 INFO [train2.py:809] (0/4) Epoch 17, batch 2900, loss[ctc_loss=0.09632, att_loss=0.2578, loss=0.2255, over 16866.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.008398, over 49.00 utterances.], tot_loss[ctc_loss=0.08296, att_loss=0.2399, loss=0.2085, over 3269340.81 frames. utt_duration=1226 frames, utt_pad_proportion=0.06185, over 10681.19 utterances.], batch size: 49, lr: 6.32e-03, grad_scale: 8.0 2023-03-08 16:02:34,935 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0183, 4.9416, 4.6852, 2.9341, 4.7196, 4.5827, 4.3408, 2.7107], device='cuda:0'), covar=tensor([0.0113, 0.0089, 0.0307, 0.0999, 0.0108, 0.0192, 0.0273, 0.1323], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0096, 0.0096, 0.0109, 0.0080, 0.0107, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 16:03:41,154 INFO [train2.py:809] (0/4) Epoch 17, batch 2950, loss[ctc_loss=0.1038, att_loss=0.2488, loss=0.2198, over 16468.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006312, over 46.00 utterances.], tot_loss[ctc_loss=0.08241, att_loss=0.24, loss=0.2085, over 3275058.70 frames. utt_duration=1243 frames, utt_pad_proportion=0.05596, over 10549.94 utterances.], batch size: 46, lr: 6.32e-03, grad_scale: 8.0 2023-03-08 16:03:58,752 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2363, 5.2413, 5.0442, 3.3196, 5.0428, 4.8475, 4.6642, 3.0825], device='cuda:0'), covar=tensor([0.0126, 0.0092, 0.0226, 0.0821, 0.0090, 0.0173, 0.0239, 0.1120], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0097, 0.0096, 0.0109, 0.0081, 0.0107, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 16:04:37,904 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3613, 2.6667, 3.0540, 4.4156, 3.8928, 4.0086, 2.8595, 2.1254], device='cuda:0'), covar=tensor([0.0723, 0.2377, 0.1076, 0.0501, 0.0833, 0.0416, 0.1659, 0.2469], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0215, 0.0189, 0.0208, 0.0213, 0.0172, 0.0198, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 16:04:39,545 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4807, 2.3787, 5.0636, 3.9402, 3.0461, 4.2764, 4.8508, 4.7166], device='cuda:0'), covar=tensor([0.0285, 0.1680, 0.0169, 0.0904, 0.1686, 0.0254, 0.0114, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0245, 0.0169, 0.0313, 0.0268, 0.0202, 0.0153, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:04:44,840 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66731.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 16:04:47,517 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 2.044e+02 2.450e+02 3.054e+02 6.191e+02, threshold=4.900e+02, percent-clipped=3.0 2023-03-08 16:05:00,733 INFO [train2.py:809] (0/4) Epoch 17, batch 3000, loss[ctc_loss=0.07361, att_loss=0.2334, loss=0.2015, over 16115.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.00581, over 42.00 utterances.], tot_loss[ctc_loss=0.08178, att_loss=0.2393, loss=0.2078, over 3272895.13 frames. utt_duration=1244 frames, utt_pad_proportion=0.0568, over 10536.56 utterances.], batch size: 42, lr: 6.32e-03, grad_scale: 8.0 2023-03-08 16:05:00,735 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 16:05:14,969 INFO [train2.py:843] (0/4) Epoch 17, validation: ctc_loss=0.04199, att_loss=0.2349, loss=0.1964, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 16:05:14,970 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 16:06:34,013 INFO [train2.py:809] (0/4) Epoch 17, batch 3050, loss[ctc_loss=0.109, att_loss=0.2448, loss=0.2176, over 16379.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.008105, over 44.00 utterances.], tot_loss[ctc_loss=0.08274, att_loss=0.2399, loss=0.2085, over 3279702.75 frames. utt_duration=1247 frames, utt_pad_proportion=0.05248, over 10531.09 utterances.], batch size: 44, lr: 6.32e-03, grad_scale: 4.0 2023-03-08 16:06:35,919 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=66792.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 16:06:49,684 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66800.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 16:07:14,794 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66816.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 16:07:15,709 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.73 vs. limit=5.0 2023-03-08 16:07:42,433 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 2.123e+02 2.532e+02 3.472e+02 1.749e+03, threshold=5.065e+02, percent-clipped=12.0 2023-03-08 16:07:53,809 INFO [train2.py:809] (0/4) Epoch 17, batch 3100, loss[ctc_loss=0.119, att_loss=0.2643, loss=0.2352, over 16476.00 frames. utt_duration=674.2 frames, utt_pad_proportion=0.1552, over 98.00 utterances.], tot_loss[ctc_loss=0.083, att_loss=0.2399, loss=0.2086, over 3270156.08 frames. utt_duration=1272 frames, utt_pad_proportion=0.04748, over 10291.93 utterances.], batch size: 98, lr: 6.31e-03, grad_scale: 4.0 2023-03-08 16:08:05,847 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66848.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:08:30,860 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66864.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 16:08:50,683 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=66877.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 16:09:12,470 INFO [train2.py:809] (0/4) Epoch 17, batch 3150, loss[ctc_loss=0.09613, att_loss=0.2513, loss=0.2203, over 16620.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.004755, over 47.00 utterances.], tot_loss[ctc_loss=0.08324, att_loss=0.2404, loss=0.2089, over 3277563.49 frames. utt_duration=1257 frames, utt_pad_proportion=0.04832, over 10443.21 utterances.], batch size: 47, lr: 6.31e-03, grad_scale: 4.0 2023-03-08 16:09:39,924 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.69 vs. limit=2.0 2023-03-08 16:09:50,593 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=66914.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:10:21,243 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 2.012e+02 2.406e+02 2.883e+02 7.010e+02, threshold=4.812e+02, percent-clipped=2.0 2023-03-08 16:10:32,787 INFO [train2.py:809] (0/4) Epoch 17, batch 3200, loss[ctc_loss=0.08319, att_loss=0.2511, loss=0.2176, over 17042.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01031, over 53.00 utterances.], tot_loss[ctc_loss=0.08313, att_loss=0.2404, loss=0.209, over 3270661.64 frames. utt_duration=1252 frames, utt_pad_proportion=0.05079, over 10465.76 utterances.], batch size: 53, lr: 6.31e-03, grad_scale: 8.0 2023-03-08 16:10:58,899 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=66957.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:11:06,327 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=66962.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:11:12,486 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8075, 6.1156, 5.5703, 5.7940, 5.7549, 5.3422, 5.4640, 5.3817], device='cuda:0'), covar=tensor([0.1430, 0.0939, 0.1005, 0.0884, 0.0956, 0.1506, 0.2452, 0.2255], device='cuda:0'), in_proj_covar=tensor([0.0508, 0.0580, 0.0447, 0.0446, 0.0423, 0.0457, 0.0598, 0.0518], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 16:11:34,474 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.38 vs. limit=5.0 2023-03-08 16:11:51,563 INFO [train2.py:809] (0/4) Epoch 17, batch 3250, loss[ctc_loss=0.07185, att_loss=0.2286, loss=0.1973, over 16109.00 frames. utt_duration=1535 frames, utt_pad_proportion=0.006869, over 42.00 utterances.], tot_loss[ctc_loss=0.08271, att_loss=0.2404, loss=0.2089, over 3268082.42 frames. utt_duration=1239 frames, utt_pad_proportion=0.05415, over 10564.08 utterances.], batch size: 42, lr: 6.31e-03, grad_scale: 8.0 2023-03-08 16:12:34,795 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67018.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 16:12:58,913 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 2.115e+02 2.445e+02 2.942e+02 7.309e+02, threshold=4.890e+02, percent-clipped=3.0 2023-03-08 16:13:11,485 INFO [train2.py:809] (0/4) Epoch 17, batch 3300, loss[ctc_loss=0.1125, att_loss=0.2286, loss=0.2054, over 15531.00 frames. utt_duration=1727 frames, utt_pad_proportion=0.006757, over 36.00 utterances.], tot_loss[ctc_loss=0.08207, att_loss=0.2395, loss=0.208, over 3263870.14 frames. utt_duration=1261 frames, utt_pad_proportion=0.05101, over 10366.22 utterances.], batch size: 36, lr: 6.30e-03, grad_scale: 8.0 2023-03-08 16:14:17,406 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67083.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:14:24,016 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67087.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 16:14:31,292 INFO [train2.py:809] (0/4) Epoch 17, batch 3350, loss[ctc_loss=0.06016, att_loss=0.2127, loss=0.1822, over 15631.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009573, over 37.00 utterances.], tot_loss[ctc_loss=0.08199, att_loss=0.2396, loss=0.2081, over 3265304.15 frames. utt_duration=1246 frames, utt_pad_proportion=0.05606, over 10498.54 utterances.], batch size: 37, lr: 6.30e-03, grad_scale: 8.0 2023-03-08 16:14:52,203 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-08 16:15:00,750 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0783, 5.2872, 5.2708, 5.1944, 5.3340, 5.3246, 5.0550, 4.7886], device='cuda:0'), covar=tensor([0.1023, 0.0515, 0.0264, 0.0510, 0.0292, 0.0315, 0.0341, 0.0365], device='cuda:0'), in_proj_covar=tensor([0.0498, 0.0336, 0.0315, 0.0328, 0.0396, 0.0409, 0.0335, 0.0372], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 16:15:37,319 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.585e+02 2.015e+02 2.419e+02 3.041e+02 5.956e+02, threshold=4.838e+02, percent-clipped=4.0 2023-03-08 16:15:49,839 INFO [train2.py:809] (0/4) Epoch 17, batch 3400, loss[ctc_loss=0.1168, att_loss=0.2577, loss=0.2295, over 16890.00 frames. utt_duration=683.9 frames, utt_pad_proportion=0.1419, over 99.00 utterances.], tot_loss[ctc_loss=0.0827, att_loss=0.2398, loss=0.2084, over 3269931.86 frames. utt_duration=1233 frames, utt_pad_proportion=0.05723, over 10619.64 utterances.], batch size: 99, lr: 6.30e-03, grad_scale: 8.0 2023-03-08 16:15:54,783 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67144.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:16:45,539 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67177.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 16:17:08,458 INFO [train2.py:809] (0/4) Epoch 17, batch 3450, loss[ctc_loss=0.09649, att_loss=0.2536, loss=0.2222, over 16884.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006656, over 49.00 utterances.], tot_loss[ctc_loss=0.08276, att_loss=0.2402, loss=0.2087, over 3274680.35 frames. utt_duration=1239 frames, utt_pad_proportion=0.0558, over 10585.93 utterances.], batch size: 49, lr: 6.30e-03, grad_scale: 8.0 2023-03-08 16:17:26,723 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.24 vs. limit=5.0 2023-03-08 16:17:34,643 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-08 16:17:38,293 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1589, 4.5240, 4.4573, 4.5889, 4.6428, 4.3304, 3.2362, 4.4944], device='cuda:0'), covar=tensor([0.0138, 0.0136, 0.0150, 0.0095, 0.0105, 0.0131, 0.0734, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0084, 0.0106, 0.0066, 0.0070, 0.0082, 0.0101, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:18:01,706 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67225.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 16:18:10,961 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6325, 5.0159, 4.8716, 4.9740, 5.1073, 4.7618, 3.4814, 4.9917], device='cuda:0'), covar=tensor([0.0112, 0.0119, 0.0125, 0.0084, 0.0079, 0.0117, 0.0762, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0085, 0.0106, 0.0066, 0.0070, 0.0083, 0.0101, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:18:15,121 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.178e+02 2.510e+02 3.172e+02 6.489e+02, threshold=5.020e+02, percent-clipped=3.0 2023-03-08 16:18:27,084 INFO [train2.py:809] (0/4) Epoch 17, batch 3500, loss[ctc_loss=0.06754, att_loss=0.2206, loss=0.19, over 15637.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.008724, over 37.00 utterances.], tot_loss[ctc_loss=0.08248, att_loss=0.24, loss=0.2085, over 3266416.44 frames. utt_duration=1238 frames, utt_pad_proportion=0.0576, over 10563.94 utterances.], batch size: 37, lr: 6.29e-03, grad_scale: 8.0 2023-03-08 16:19:46,679 INFO [train2.py:809] (0/4) Epoch 17, batch 3550, loss[ctc_loss=0.09408, att_loss=0.2623, loss=0.2287, over 17343.00 frames. utt_duration=1177 frames, utt_pad_proportion=0.0221, over 59.00 utterances.], tot_loss[ctc_loss=0.08266, att_loss=0.2407, loss=0.2091, over 3276632.78 frames. utt_duration=1254 frames, utt_pad_proportion=0.05223, over 10463.92 utterances.], batch size: 59, lr: 6.29e-03, grad_scale: 8.0 2023-03-08 16:19:49,106 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.99 vs. limit=2.0 2023-03-08 16:20:21,430 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67313.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 16:20:35,419 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9058, 5.1707, 5.1727, 5.0711, 5.1760, 5.1777, 4.8816, 4.6467], device='cuda:0'), covar=tensor([0.1076, 0.0564, 0.0288, 0.0520, 0.0330, 0.0302, 0.0354, 0.0383], device='cuda:0'), in_proj_covar=tensor([0.0502, 0.0337, 0.0316, 0.0332, 0.0395, 0.0409, 0.0335, 0.0371], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 16:20:54,012 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.300e+02 1.995e+02 2.311e+02 2.839e+02 6.877e+02, threshold=4.621e+02, percent-clipped=1.0 2023-03-08 16:21:06,785 INFO [train2.py:809] (0/4) Epoch 17, batch 3600, loss[ctc_loss=0.0852, att_loss=0.2488, loss=0.2161, over 17368.00 frames. utt_duration=880.9 frames, utt_pad_proportion=0.07858, over 79.00 utterances.], tot_loss[ctc_loss=0.08216, att_loss=0.2403, loss=0.2087, over 3278119.19 frames. utt_duration=1254 frames, utt_pad_proportion=0.0509, over 10467.75 utterances.], batch size: 79, lr: 6.29e-03, grad_scale: 8.0 2023-03-08 16:21:12,352 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-08 16:22:01,501 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1556, 5.4243, 5.6819, 5.4394, 5.5953, 6.1203, 5.3421, 6.1984], device='cuda:0'), covar=tensor([0.0646, 0.0641, 0.0769, 0.1168, 0.1805, 0.0784, 0.0561, 0.0602], device='cuda:0'), in_proj_covar=tensor([0.0820, 0.0480, 0.0569, 0.0634, 0.0828, 0.0581, 0.0462, 0.0575], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:22:19,248 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67387.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 16:22:26,310 INFO [train2.py:809] (0/4) Epoch 17, batch 3650, loss[ctc_loss=0.1071, att_loss=0.2563, loss=0.2265, over 17088.00 frames. utt_duration=691.8 frames, utt_pad_proportion=0.1309, over 99.00 utterances.], tot_loss[ctc_loss=0.08229, att_loss=0.2402, loss=0.2086, over 3262385.92 frames. utt_duration=1226 frames, utt_pad_proportion=0.06221, over 10654.83 utterances.], batch size: 99, lr: 6.29e-03, grad_scale: 8.0 2023-03-08 16:22:32,995 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4624, 2.5571, 5.0398, 3.9175, 3.1717, 4.2105, 4.7934, 4.6350], device='cuda:0'), covar=tensor([0.0255, 0.1735, 0.0156, 0.0929, 0.1675, 0.0263, 0.0130, 0.0221], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0238, 0.0167, 0.0306, 0.0262, 0.0198, 0.0151, 0.0180], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:23:21,902 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1363, 2.5342, 3.1274, 4.3434, 3.7823, 3.7446, 2.7757, 2.0986], device='cuda:0'), covar=tensor([0.0745, 0.2181, 0.0946, 0.0508, 0.0753, 0.0485, 0.1536, 0.2223], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0211, 0.0186, 0.0207, 0.0210, 0.0171, 0.0198, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 16:23:34,279 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 2.053e+02 2.468e+02 2.850e+02 9.204e+02, threshold=4.935e+02, percent-clipped=2.0 2023-03-08 16:23:35,847 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 16:23:36,614 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67435.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 16:23:44,077 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67439.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:23:47,236 INFO [train2.py:809] (0/4) Epoch 17, batch 3700, loss[ctc_loss=0.05303, att_loss=0.203, loss=0.173, over 15362.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01184, over 35.00 utterances.], tot_loss[ctc_loss=0.08219, att_loss=0.2404, loss=0.2087, over 3270512.71 frames. utt_duration=1244 frames, utt_pad_proportion=0.05506, over 10531.08 utterances.], batch size: 35, lr: 6.28e-03, grad_scale: 8.0 2023-03-08 16:24:39,158 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-08 16:24:59,097 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0057, 5.1501, 5.0914, 2.2137, 1.8513, 2.6159, 2.6001, 3.9494], device='cuda:0'), covar=tensor([0.0711, 0.0247, 0.0186, 0.4607, 0.6337, 0.2924, 0.3040, 0.1569], device='cuda:0'), in_proj_covar=tensor([0.0346, 0.0255, 0.0253, 0.0232, 0.0343, 0.0333, 0.0241, 0.0357], device='cuda:0'), out_proj_covar=tensor([1.4839e-04, 9.4056e-05, 1.0889e-04, 9.9904e-05, 1.4448e-04, 1.3120e-04, 9.6297e-05, 1.4659e-04], device='cuda:0') 2023-03-08 16:25:07,046 INFO [train2.py:809] (0/4) Epoch 17, batch 3750, loss[ctc_loss=0.07218, att_loss=0.2258, loss=0.1951, over 16286.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006899, over 43.00 utterances.], tot_loss[ctc_loss=0.08258, att_loss=0.2401, loss=0.2086, over 3265631.30 frames. utt_duration=1244 frames, utt_pad_proportion=0.0557, over 10509.35 utterances.], batch size: 43, lr: 6.28e-03, grad_scale: 8.0 2023-03-08 16:25:10,595 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2362, 3.9284, 3.3707, 3.6140, 4.1079, 3.8007, 3.2998, 4.4734], device='cuda:0'), covar=tensor([0.0966, 0.0448, 0.1067, 0.0657, 0.0665, 0.0596, 0.0790, 0.0454], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0209, 0.0220, 0.0193, 0.0266, 0.0233, 0.0196, 0.0276], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 16:25:22,425 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6859, 2.5592, 5.1592, 4.1069, 3.0293, 4.4383, 4.8468, 4.8726], device='cuda:0'), covar=tensor([0.0172, 0.1473, 0.0111, 0.0783, 0.1697, 0.0178, 0.0094, 0.0143], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0240, 0.0168, 0.0308, 0.0265, 0.0200, 0.0152, 0.0182], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:25:42,126 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67513.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:26:14,091 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.055e+02 2.519e+02 3.069e+02 5.925e+02, threshold=5.038e+02, percent-clipped=2.0 2023-03-08 16:26:25,906 INFO [train2.py:809] (0/4) Epoch 17, batch 3800, loss[ctc_loss=0.06285, att_loss=0.2267, loss=0.194, over 16326.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006597, over 45.00 utterances.], tot_loss[ctc_loss=0.08231, att_loss=0.2403, loss=0.2087, over 3274049.56 frames. utt_duration=1229 frames, utt_pad_proportion=0.05572, over 10667.31 utterances.], batch size: 45, lr: 6.28e-03, grad_scale: 8.0 2023-03-08 16:27:18,143 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67574.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 16:27:30,720 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1290, 2.7472, 3.5663, 2.7764, 3.4775, 4.3334, 4.1913, 3.0517], device='cuda:0'), covar=tensor([0.0478, 0.1811, 0.1113, 0.1410, 0.0961, 0.0925, 0.0549, 0.1331], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0240, 0.0270, 0.0212, 0.0257, 0.0352, 0.0250, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:27:30,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3740, 2.4834, 4.8401, 3.7738, 2.8549, 4.1478, 4.3390, 4.5232], device='cuda:0'), covar=tensor([0.0213, 0.1775, 0.0116, 0.0878, 0.1795, 0.0238, 0.0171, 0.0213], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0239, 0.0168, 0.0307, 0.0264, 0.0200, 0.0151, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:27:46,044 INFO [train2.py:809] (0/4) Epoch 17, batch 3850, loss[ctc_loss=0.08369, att_loss=0.2577, loss=0.2229, over 17030.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.01083, over 53.00 utterances.], tot_loss[ctc_loss=0.08167, att_loss=0.2397, loss=0.2081, over 3269227.68 frames. utt_duration=1245 frames, utt_pad_proportion=0.05435, over 10515.07 utterances.], batch size: 53, lr: 6.28e-03, grad_scale: 8.0 2023-03-08 16:27:57,047 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4950, 2.6224, 5.0222, 3.8553, 3.0000, 4.2357, 4.7214, 4.6889], device='cuda:0'), covar=tensor([0.0260, 0.1564, 0.0157, 0.0869, 0.1759, 0.0233, 0.0119, 0.0206], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0240, 0.0168, 0.0307, 0.0264, 0.0200, 0.0151, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:28:04,459 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-03-08 16:28:14,431 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5167, 2.2029, 2.1229, 2.1212, 2.7554, 2.4891, 2.2516, 2.6870], device='cuda:0'), covar=tensor([0.2056, 0.3719, 0.3140, 0.2133, 0.1986, 0.1578, 0.3027, 0.1246], device='cuda:0'), in_proj_covar=tensor([0.0101, 0.0110, 0.0113, 0.0097, 0.0106, 0.0092, 0.0113, 0.0083], device='cuda:0'), out_proj_covar=tensor([7.4804e-05, 8.3419e-05, 8.6737e-05, 7.3646e-05, 7.7359e-05, 7.2988e-05, 8.3517e-05, 6.6356e-05], device='cuda:0') 2023-03-08 16:28:20,432 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67613.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:28:53,212 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.336e+02 2.153e+02 2.695e+02 3.326e+02 5.215e+02, threshold=5.391e+02, percent-clipped=2.0 2023-03-08 16:28:55,164 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7709, 3.5038, 3.4509, 3.1037, 3.5738, 3.5556, 3.5252, 2.6101], device='cuda:0'), covar=tensor([0.1062, 0.1263, 0.3000, 0.3747, 0.1100, 0.1911, 0.0841, 0.4123], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0162, 0.0171, 0.0232, 0.0138, 0.0228, 0.0149, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:29:04,153 INFO [train2.py:809] (0/4) Epoch 17, batch 3900, loss[ctc_loss=0.08931, att_loss=0.2522, loss=0.2197, over 17235.00 frames. utt_duration=697.8 frames, utt_pad_proportion=0.1222, over 99.00 utterances.], tot_loss[ctc_loss=0.08132, att_loss=0.2392, loss=0.2076, over 3264242.60 frames. utt_duration=1250 frames, utt_pad_proportion=0.05522, over 10460.40 utterances.], batch size: 99, lr: 6.28e-03, grad_scale: 8.0 2023-03-08 16:29:25,729 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1300, 2.8174, 3.5707, 2.8247, 3.4470, 4.3376, 4.1863, 3.0601], device='cuda:0'), covar=tensor([0.0482, 0.1752, 0.1126, 0.1352, 0.0974, 0.0907, 0.0633, 0.1343], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0241, 0.0269, 0.0212, 0.0257, 0.0351, 0.0248, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:29:34,760 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67661.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:30:04,311 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67680.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:30:21,233 INFO [train2.py:809] (0/4) Epoch 17, batch 3950, loss[ctc_loss=0.08727, att_loss=0.2411, loss=0.2103, over 16276.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007417, over 43.00 utterances.], tot_loss[ctc_loss=0.08178, att_loss=0.24, loss=0.2083, over 3271021.13 frames. utt_duration=1241 frames, utt_pad_proportion=0.05599, over 10551.83 utterances.], batch size: 43, lr: 6.27e-03, grad_scale: 8.0 2023-03-08 16:30:23,118 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67692.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:31:13,297 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-17.pt 2023-03-08 16:31:39,746 INFO [train2.py:809] (0/4) Epoch 18, batch 0, loss[ctc_loss=0.08787, att_loss=0.2533, loss=0.2202, over 16673.00 frames. utt_duration=1451 frames, utt_pad_proportion=0.005185, over 46.00 utterances.], tot_loss[ctc_loss=0.08787, att_loss=0.2533, loss=0.2202, over 16673.00 frames. utt_duration=1451 frames, utt_pad_proportion=0.005185, over 46.00 utterances.], batch size: 46, lr: 6.09e-03, grad_scale: 8.0 2023-03-08 16:31:39,748 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 16:31:52,766 INFO [train2.py:843] (0/4) Epoch 18, validation: ctc_loss=0.04224, att_loss=0.2352, loss=0.1966, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 16:31:52,766 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 16:32:02,658 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 16:32:06,404 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 2.121e+02 2.560e+02 3.250e+02 7.102e+02, threshold=5.120e+02, percent-clipped=3.0 2023-03-08 16:32:09,763 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8276, 3.5349, 3.5474, 3.1129, 3.5673, 3.6268, 3.5981, 2.6677], device='cuda:0'), covar=tensor([0.1097, 0.1679, 0.2295, 0.5201, 0.2582, 0.4880, 0.1071, 0.4762], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0166, 0.0173, 0.0237, 0.0142, 0.0232, 0.0152, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:32:14,396 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=67739.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:32:17,545 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67741.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:32:35,573 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=67753.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 16:33:00,095 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2109, 3.7959, 3.2558, 3.5765, 3.9118, 3.5953, 3.1400, 4.3735], device='cuda:0'), covar=tensor([0.0961, 0.0487, 0.1150, 0.0641, 0.0742, 0.0730, 0.0872, 0.0569], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0206, 0.0217, 0.0191, 0.0261, 0.0230, 0.0194, 0.0273], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 16:33:01,568 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5574, 4.8906, 4.7198, 4.7680, 4.9758, 4.5803, 3.5160, 4.8272], device='cuda:0'), covar=tensor([0.0110, 0.0117, 0.0122, 0.0121, 0.0071, 0.0107, 0.0624, 0.0210], device='cuda:0'), in_proj_covar=tensor([0.0086, 0.0084, 0.0104, 0.0065, 0.0070, 0.0082, 0.0100, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:33:11,757 INFO [train2.py:809] (0/4) Epoch 18, batch 50, loss[ctc_loss=0.06342, att_loss=0.2388, loss=0.2037, over 16482.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006519, over 46.00 utterances.], tot_loss[ctc_loss=0.08278, att_loss=0.2429, loss=0.2109, over 746638.97 frames. utt_duration=1288 frames, utt_pad_proportion=0.03624, over 2322.20 utterances.], batch size: 46, lr: 6.09e-03, grad_scale: 8.0 2023-03-08 16:33:15,132 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9220, 5.2253, 4.8060, 5.3470, 4.7403, 4.9570, 5.4135, 5.1500], device='cuda:0'), covar=tensor([0.0552, 0.0309, 0.0767, 0.0257, 0.0364, 0.0289, 0.0204, 0.0179], device='cuda:0'), in_proj_covar=tensor([0.0372, 0.0299, 0.0350, 0.0316, 0.0301, 0.0226, 0.0284, 0.0269], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 16:33:21,415 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7753, 2.4914, 3.3124, 2.4048, 3.2778, 3.9365, 3.8553, 2.6958], device='cuda:0'), covar=tensor([0.0489, 0.1914, 0.1145, 0.1531, 0.1033, 0.1122, 0.0689, 0.1639], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0243, 0.0271, 0.0214, 0.0260, 0.0354, 0.0250, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:33:30,487 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=67787.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:34:31,359 INFO [train2.py:809] (0/4) Epoch 18, batch 100, loss[ctc_loss=0.07126, att_loss=0.2349, loss=0.2022, over 16626.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005469, over 47.00 utterances.], tot_loss[ctc_loss=0.08508, att_loss=0.2445, loss=0.2126, over 1309260.58 frames. utt_duration=1213 frames, utt_pad_proportion=0.05827, over 4323.23 utterances.], batch size: 47, lr: 6.09e-03, grad_scale: 8.0 2023-03-08 16:34:45,110 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.584e+02 2.310e+02 2.796e+02 3.498e+02 6.865e+02, threshold=5.593e+02, percent-clipped=9.0 2023-03-08 16:34:46,962 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5397, 5.0129, 4.7211, 4.8623, 5.0154, 4.6125, 3.7113, 4.9102], device='cuda:0'), covar=tensor([0.0120, 0.0119, 0.0178, 0.0105, 0.0087, 0.0141, 0.0618, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0087, 0.0085, 0.0105, 0.0066, 0.0070, 0.0083, 0.0101, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:35:14,027 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 16:35:36,090 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.95 vs. limit=2.0 2023-03-08 16:35:39,866 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=67869.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 16:35:49,981 INFO [train2.py:809] (0/4) Epoch 18, batch 150, loss[ctc_loss=0.1167, att_loss=0.2693, loss=0.2388, over 17045.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01018, over 53.00 utterances.], tot_loss[ctc_loss=0.08322, att_loss=0.2423, loss=0.2104, over 1739056.72 frames. utt_duration=1225 frames, utt_pad_proportion=0.05934, over 5683.31 utterances.], batch size: 53, lr: 6.09e-03, grad_scale: 8.0 2023-03-08 16:36:07,891 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 16:37:08,987 INFO [train2.py:809] (0/4) Epoch 18, batch 200, loss[ctc_loss=0.07875, att_loss=0.2478, loss=0.214, over 17289.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02418, over 59.00 utterances.], tot_loss[ctc_loss=0.0834, att_loss=0.2419, loss=0.2102, over 2080795.99 frames. utt_duration=1196 frames, utt_pad_proportion=0.06549, over 6965.14 utterances.], batch size: 59, lr: 6.08e-03, grad_scale: 8.0 2023-03-08 16:37:22,689 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.418e+02 2.000e+02 2.339e+02 2.884e+02 4.176e+02, threshold=4.678e+02, percent-clipped=0.0 2023-03-08 16:38:27,432 INFO [train2.py:809] (0/4) Epoch 18, batch 250, loss[ctc_loss=0.07863, att_loss=0.2465, loss=0.2129, over 16632.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004796, over 47.00 utterances.], tot_loss[ctc_loss=0.08365, att_loss=0.2421, loss=0.2104, over 2350216.23 frames. utt_duration=1190 frames, utt_pad_proportion=0.06642, over 7909.99 utterances.], batch size: 47, lr: 6.08e-03, grad_scale: 8.0 2023-03-08 16:38:36,804 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=67981.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:39:06,073 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-68000.pt 2023-03-08 16:39:16,933 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7101, 2.7030, 5.0898, 4.0841, 3.2068, 4.3476, 4.8441, 4.7383], device='cuda:0'), covar=tensor([0.0219, 0.1490, 0.0162, 0.0809, 0.1492, 0.0216, 0.0120, 0.0220], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0238, 0.0168, 0.0304, 0.0260, 0.0198, 0.0150, 0.0181], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:39:50,973 INFO [train2.py:809] (0/4) Epoch 18, batch 300, loss[ctc_loss=0.1305, att_loss=0.2728, loss=0.2443, over 13976.00 frames. utt_duration=387 frames, utt_pad_proportion=0.327, over 145.00 utterances.], tot_loss[ctc_loss=0.0828, att_loss=0.2412, loss=0.2095, over 2550765.61 frames. utt_duration=1217 frames, utt_pad_proportion=0.06098, over 8393.30 utterances.], batch size: 145, lr: 6.08e-03, grad_scale: 8.0 2023-03-08 16:40:04,757 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.007e+02 2.345e+02 2.847e+02 6.033e+02, threshold=4.689e+02, percent-clipped=3.0 2023-03-08 16:40:08,023 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68036.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:40:17,179 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68042.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:40:26,328 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68048.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 16:41:09,393 INFO [train2.py:809] (0/4) Epoch 18, batch 350, loss[ctc_loss=0.0596, att_loss=0.2066, loss=0.1772, over 15486.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.009568, over 36.00 utterances.], tot_loss[ctc_loss=0.08241, att_loss=0.2418, loss=0.2099, over 2727292.57 frames. utt_duration=1232 frames, utt_pad_proportion=0.05211, over 8867.00 utterances.], batch size: 36, lr: 6.08e-03, grad_scale: 8.0 2023-03-08 16:41:27,808 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68087.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 16:42:27,979 INFO [train2.py:809] (0/4) Epoch 18, batch 400, loss[ctc_loss=0.07308, att_loss=0.2377, loss=0.2048, over 16548.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.00527, over 45.00 utterances.], tot_loss[ctc_loss=0.08219, att_loss=0.2414, loss=0.2096, over 2849632.25 frames. utt_duration=1210 frames, utt_pad_proportion=0.05945, over 9433.48 utterances.], batch size: 45, lr: 6.07e-03, grad_scale: 8.0 2023-03-08 16:42:41,447 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.402e+02 2.125e+02 2.519e+02 3.295e+02 8.158e+02, threshold=5.039e+02, percent-clipped=7.0 2023-03-08 16:42:45,372 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.92 vs. limit=2.0 2023-03-08 16:43:03,787 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68148.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 16:43:26,003 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7717, 6.0227, 5.5302, 5.7819, 5.6689, 5.2616, 5.5446, 5.2255], device='cuda:0'), covar=tensor([0.1386, 0.0826, 0.0822, 0.0792, 0.0947, 0.1386, 0.2112, 0.2432], device='cuda:0'), in_proj_covar=tensor([0.0499, 0.0578, 0.0438, 0.0437, 0.0416, 0.0446, 0.0592, 0.0509], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 16:43:37,507 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68169.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:43:46,625 INFO [train2.py:809] (0/4) Epoch 18, batch 450, loss[ctc_loss=0.09345, att_loss=0.2598, loss=0.2265, over 17300.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01189, over 55.00 utterances.], tot_loss[ctc_loss=0.0822, att_loss=0.2411, loss=0.2093, over 2939252.95 frames. utt_duration=1224 frames, utt_pad_proportion=0.05886, over 9620.43 utterances.], batch size: 55, lr: 6.07e-03, grad_scale: 8.0 2023-03-08 16:44:53,045 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68217.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:45:05,402 INFO [train2.py:809] (0/4) Epoch 18, batch 500, loss[ctc_loss=0.09785, att_loss=0.2553, loss=0.2238, over 16882.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006759, over 49.00 utterances.], tot_loss[ctc_loss=0.08215, att_loss=0.2402, loss=0.2086, over 3011095.76 frames. utt_duration=1224 frames, utt_pad_proportion=0.05961, over 9850.80 utterances.], batch size: 49, lr: 6.07e-03, grad_scale: 8.0 2023-03-08 16:45:19,448 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.432e+02 1.924e+02 2.355e+02 3.288e+02 6.167e+02, threshold=4.710e+02, percent-clipped=3.0 2023-03-08 16:45:33,530 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6779, 5.1310, 5.1011, 5.0349, 5.1091, 5.1426, 4.9038, 4.7416], device='cuda:0'), covar=tensor([0.1403, 0.0609, 0.0335, 0.0538, 0.0518, 0.0378, 0.0387, 0.0356], device='cuda:0'), in_proj_covar=tensor([0.0500, 0.0335, 0.0316, 0.0332, 0.0395, 0.0410, 0.0335, 0.0372], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 16:46:22,159 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7020, 4.8224, 4.6471, 4.6294, 5.3782, 4.6632, 4.6892, 2.4688], device='cuda:0'), covar=tensor([0.0212, 0.0288, 0.0279, 0.0377, 0.0925, 0.0210, 0.0311, 0.1944], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0163, 0.0169, 0.0184, 0.0361, 0.0142, 0.0157, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 16:46:24,805 INFO [train2.py:809] (0/4) Epoch 18, batch 550, loss[ctc_loss=0.06867, att_loss=0.2336, loss=0.2006, over 16131.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.006064, over 42.00 utterances.], tot_loss[ctc_loss=0.08143, att_loss=0.2391, loss=0.2076, over 3064531.22 frames. utt_duration=1246 frames, utt_pad_proportion=0.0557, over 9852.94 utterances.], batch size: 42, lr: 6.07e-03, grad_scale: 8.0 2023-03-08 16:47:07,457 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2023-03-08 16:47:44,061 INFO [train2.py:809] (0/4) Epoch 18, batch 600, loss[ctc_loss=0.08762, att_loss=0.2426, loss=0.2116, over 16628.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005185, over 47.00 utterances.], tot_loss[ctc_loss=0.082, att_loss=0.2399, loss=0.2083, over 3119243.79 frames. utt_duration=1237 frames, utt_pad_proportion=0.05546, over 10101.67 utterances.], batch size: 47, lr: 6.07e-03, grad_scale: 8.0 2023-03-08 16:47:58,121 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 2.193e+02 2.599e+02 3.203e+02 7.782e+02, threshold=5.199e+02, percent-clipped=2.0 2023-03-08 16:48:01,675 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68336.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:48:03,050 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68337.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:48:19,822 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68348.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:49:03,207 INFO [train2.py:809] (0/4) Epoch 18, batch 650, loss[ctc_loss=0.07898, att_loss=0.2094, loss=0.1833, over 14094.00 frames. utt_duration=1820 frames, utt_pad_proportion=0.05057, over 31.00 utterances.], tot_loss[ctc_loss=0.08195, att_loss=0.2394, loss=0.2079, over 3152589.20 frames. utt_duration=1252 frames, utt_pad_proportion=0.05073, over 10082.42 utterances.], batch size: 31, lr: 6.06e-03, grad_scale: 8.0 2023-03-08 16:49:17,261 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68384.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:49:17,501 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0950, 2.7111, 3.5275, 2.8000, 3.3213, 4.3400, 4.1600, 2.9691], device='cuda:0'), covar=tensor([0.0474, 0.1781, 0.1266, 0.1380, 0.1213, 0.0873, 0.0651, 0.1469], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0245, 0.0275, 0.0214, 0.0263, 0.0355, 0.0252, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:49:33,601 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-03-08 16:49:35,683 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68396.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:49:40,446 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68399.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:50:22,681 INFO [train2.py:809] (0/4) Epoch 18, batch 700, loss[ctc_loss=0.08791, att_loss=0.2542, loss=0.2209, over 16786.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005563, over 48.00 utterances.], tot_loss[ctc_loss=0.08224, att_loss=0.2401, loss=0.2085, over 3184408.07 frames. utt_duration=1238 frames, utt_pad_proportion=0.05337, over 10298.05 utterances.], batch size: 48, lr: 6.06e-03, grad_scale: 8.0 2023-03-08 16:50:36,285 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.516e+02 2.020e+02 2.578e+02 3.082e+02 5.307e+02, threshold=5.155e+02, percent-clipped=1.0 2023-03-08 16:50:50,335 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68443.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 16:51:17,713 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68460.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:51:20,032 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2007, 5.1669, 4.9539, 2.4270, 1.9532, 3.1089, 2.4052, 3.9560], device='cuda:0'), covar=tensor([0.0612, 0.0293, 0.0281, 0.4880, 0.5689, 0.2260, 0.3349, 0.1622], device='cuda:0'), in_proj_covar=tensor([0.0346, 0.0257, 0.0256, 0.0235, 0.0344, 0.0334, 0.0243, 0.0360], device='cuda:0'), out_proj_covar=tensor([1.4798e-04, 9.4816e-05, 1.0964e-04, 1.0194e-04, 1.4453e-04, 1.3110e-04, 9.7106e-05, 1.4741e-04], device='cuda:0') 2023-03-08 16:51:41,304 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-03-08 16:51:41,630 INFO [train2.py:809] (0/4) Epoch 18, batch 750, loss[ctc_loss=0.07648, att_loss=0.239, loss=0.2065, over 16995.00 frames. utt_duration=688.2 frames, utt_pad_proportion=0.1376, over 99.00 utterances.], tot_loss[ctc_loss=0.08119, att_loss=0.2391, loss=0.2075, over 3202562.14 frames. utt_duration=1248 frames, utt_pad_proportion=0.05319, over 10280.50 utterances.], batch size: 99, lr: 6.06e-03, grad_scale: 8.0 2023-03-08 16:53:00,269 INFO [train2.py:809] (0/4) Epoch 18, batch 800, loss[ctc_loss=0.07897, att_loss=0.2431, loss=0.2103, over 16858.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.008228, over 49.00 utterances.], tot_loss[ctc_loss=0.0813, att_loss=0.239, loss=0.2075, over 3220447.30 frames. utt_duration=1264 frames, utt_pad_proportion=0.04838, over 10202.98 utterances.], batch size: 49, lr: 6.06e-03, grad_scale: 8.0 2023-03-08 16:53:13,826 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.459e+02 2.218e+02 2.657e+02 2.960e+02 6.407e+02, threshold=5.314e+02, percent-clipped=4.0 2023-03-08 16:54:07,450 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9496, 3.9897, 3.7611, 2.7739, 3.8551, 3.7868, 3.4920, 2.7296], device='cuda:0'), covar=tensor([0.0111, 0.0117, 0.0252, 0.0900, 0.0108, 0.0364, 0.0314, 0.1142], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0097, 0.0095, 0.0110, 0.0081, 0.0107, 0.0097, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 16:54:19,202 INFO [train2.py:809] (0/4) Epoch 18, batch 850, loss[ctc_loss=0.08278, att_loss=0.2382, loss=0.2072, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007394, over 44.00 utterances.], tot_loss[ctc_loss=0.0806, att_loss=0.2383, loss=0.2068, over 3233766.48 frames. utt_duration=1253 frames, utt_pad_proportion=0.05162, over 10339.51 utterances.], batch size: 44, lr: 6.05e-03, grad_scale: 8.0 2023-03-08 16:54:24,110 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68578.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:54:47,187 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1616, 4.5180, 4.2588, 4.6003, 2.6821, 4.4224, 2.4810, 2.0261], device='cuda:0'), covar=tensor([0.0363, 0.0227, 0.0768, 0.0192, 0.1964, 0.0207, 0.1802, 0.1925], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0146, 0.0253, 0.0141, 0.0218, 0.0127, 0.0228, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 16:55:14,284 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5400, 2.1382, 2.2534, 2.5240, 2.7967, 2.4761, 2.3815, 3.1434], device='cuda:0'), covar=tensor([0.1738, 0.3824, 0.2786, 0.1700, 0.1804, 0.1718, 0.3040, 0.0792], device='cuda:0'), in_proj_covar=tensor([0.0103, 0.0112, 0.0113, 0.0097, 0.0106, 0.0094, 0.0115, 0.0084], device='cuda:0'), out_proj_covar=tensor([7.6164e-05, 8.4677e-05, 8.7053e-05, 7.4278e-05, 7.7722e-05, 7.4570e-05, 8.5178e-05, 6.7108e-05], device='cuda:0') 2023-03-08 16:55:38,614 INFO [train2.py:809] (0/4) Epoch 18, batch 900, loss[ctc_loss=0.07259, att_loss=0.2218, loss=0.192, over 15869.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01022, over 39.00 utterances.], tot_loss[ctc_loss=0.08121, att_loss=0.2389, loss=0.2074, over 3238590.55 frames. utt_duration=1244 frames, utt_pad_proportion=0.0546, over 10422.27 utterances.], batch size: 39, lr: 6.05e-03, grad_scale: 8.0 2023-03-08 16:55:38,984 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9974, 3.6121, 3.5998, 3.1038, 3.8124, 3.7666, 3.7257, 2.7754], device='cuda:0'), covar=tensor([0.0938, 0.2029, 0.2464, 0.5385, 0.1702, 0.2747, 0.0787, 0.4648], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0167, 0.0178, 0.0242, 0.0145, 0.0240, 0.0156, 0.0207], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 16:55:52,341 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.470e+02 1.982e+02 2.432e+02 3.091e+02 4.816e+02, threshold=4.864e+02, percent-clipped=0.0 2023-03-08 16:55:57,213 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68637.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:56:00,254 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68639.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:56:33,568 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2128, 2.8454, 3.2997, 4.4872, 3.9932, 3.9796, 2.9367, 2.0728], device='cuda:0'), covar=tensor([0.0834, 0.2255, 0.0973, 0.0541, 0.0917, 0.0467, 0.1558, 0.2489], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0213, 0.0187, 0.0207, 0.0211, 0.0170, 0.0197, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 16:56:57,069 INFO [train2.py:809] (0/4) Epoch 18, batch 950, loss[ctc_loss=0.07602, att_loss=0.2442, loss=0.2105, over 17049.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009086, over 53.00 utterances.], tot_loss[ctc_loss=0.08049, att_loss=0.2387, loss=0.207, over 3242001.78 frames. utt_duration=1248 frames, utt_pad_proportion=0.05332, over 10399.24 utterances.], batch size: 53, lr: 6.05e-03, grad_scale: 8.0 2023-03-08 16:57:12,173 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:57:12,419 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=68685.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:57:39,765 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4758, 4.9241, 4.7306, 4.8628, 4.9295, 4.4897, 3.5818, 4.8236], device='cuda:0'), covar=tensor([0.0129, 0.0108, 0.0137, 0.0081, 0.0101, 0.0124, 0.0623, 0.0187], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0085, 0.0105, 0.0067, 0.0071, 0.0083, 0.0101, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 16:58:05,419 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1050, 6.3088, 5.7981, 6.1288, 6.0441, 5.5417, 5.7595, 5.5123], device='cuda:0'), covar=tensor([0.1348, 0.0822, 0.0846, 0.0748, 0.0790, 0.1485, 0.2065, 0.2178], device='cuda:0'), in_proj_covar=tensor([0.0504, 0.0577, 0.0439, 0.0439, 0.0414, 0.0446, 0.0589, 0.0510], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 16:58:15,808 INFO [train2.py:809] (0/4) Epoch 18, batch 1000, loss[ctc_loss=0.1002, att_loss=0.2541, loss=0.2234, over 17172.00 frames. utt_duration=695.4 frames, utt_pad_proportion=0.1264, over 99.00 utterances.], tot_loss[ctc_loss=0.0812, att_loss=0.239, loss=0.2074, over 3237365.01 frames. utt_duration=1255 frames, utt_pad_proportion=0.0523, over 10328.88 utterances.], batch size: 99, lr: 6.05e-03, grad_scale: 8.0 2023-03-08 16:58:29,416 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.078e+02 2.479e+02 2.964e+02 7.208e+02, threshold=4.957e+02, percent-clipped=4.0 2023-03-08 16:58:31,942 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 16:58:43,597 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=68743.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 16:58:48,366 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=68746.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:59:02,245 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68755.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 16:59:33,512 INFO [train2.py:809] (0/4) Epoch 18, batch 1050, loss[ctc_loss=0.076, att_loss=0.2421, loss=0.2089, over 17123.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01405, over 56.00 utterances.], tot_loss[ctc_loss=0.08113, att_loss=0.2391, loss=0.2075, over 3242728.25 frames. utt_duration=1246 frames, utt_pad_proportion=0.05448, over 10420.19 utterances.], batch size: 56, lr: 6.05e-03, grad_scale: 16.0 2023-03-08 16:59:58,337 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=68791.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 17:00:53,126 INFO [train2.py:809] (0/4) Epoch 18, batch 1100, loss[ctc_loss=0.07206, att_loss=0.2452, loss=0.2106, over 17050.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.0091, over 52.00 utterances.], tot_loss[ctc_loss=0.08083, att_loss=0.2391, loss=0.2074, over 3246622.94 frames. utt_duration=1247 frames, utt_pad_proportion=0.05549, over 10429.88 utterances.], batch size: 52, lr: 6.04e-03, grad_scale: 16.0 2023-03-08 17:01:06,985 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.026e+02 2.542e+02 3.113e+02 8.116e+02, threshold=5.085e+02, percent-clipped=3.0 2023-03-08 17:01:22,190 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-08 17:02:12,249 INFO [train2.py:809] (0/4) Epoch 18, batch 1150, loss[ctc_loss=0.09367, att_loss=0.2285, loss=0.2016, over 15949.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.00726, over 41.00 utterances.], tot_loss[ctc_loss=0.08067, att_loss=0.2392, loss=0.2075, over 3257060.40 frames. utt_duration=1238 frames, utt_pad_proportion=0.05639, over 10536.49 utterances.], batch size: 41, lr: 6.04e-03, grad_scale: 16.0 2023-03-08 17:02:20,964 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-03-08 17:03:28,902 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.97 vs. limit=2.0 2023-03-08 17:03:29,036 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2023-03-08 17:03:31,040 INFO [train2.py:809] (0/4) Epoch 18, batch 1200, loss[ctc_loss=0.07371, att_loss=0.2297, loss=0.1985, over 16415.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006154, over 44.00 utterances.], tot_loss[ctc_loss=0.08036, att_loss=0.2393, loss=0.2075, over 3258028.62 frames. utt_duration=1254 frames, utt_pad_proportion=0.05278, over 10406.10 utterances.], batch size: 44, lr: 6.04e-03, grad_scale: 16.0 2023-03-08 17:03:44,924 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.483e+02 2.054e+02 2.571e+02 3.261e+02 6.642e+02, threshold=5.143e+02, percent-clipped=4.0 2023-03-08 17:03:45,180 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=68934.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:04:50,127 INFO [train2.py:809] (0/4) Epoch 18, batch 1250, loss[ctc_loss=0.119, att_loss=0.2675, loss=0.2378, over 17047.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008482, over 52.00 utterances.], tot_loss[ctc_loss=0.07953, att_loss=0.2385, loss=0.2067, over 3264304.90 frames. utt_duration=1279 frames, utt_pad_proportion=0.04652, over 10224.59 utterances.], batch size: 52, lr: 6.04e-03, grad_scale: 16.0 2023-03-08 17:05:12,724 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.94 vs. limit=2.0 2023-03-08 17:06:07,425 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4679, 2.5896, 4.9513, 3.8803, 3.0997, 4.2075, 4.5762, 4.6432], device='cuda:0'), covar=tensor([0.0231, 0.1565, 0.0126, 0.0872, 0.1607, 0.0239, 0.0173, 0.0217], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0240, 0.0171, 0.0309, 0.0266, 0.0202, 0.0154, 0.0184], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 17:06:07,742 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-03-08 17:06:08,500 INFO [train2.py:809] (0/4) Epoch 18, batch 1300, loss[ctc_loss=0.0947, att_loss=0.2561, loss=0.2238, over 17353.00 frames. utt_duration=1178 frames, utt_pad_proportion=0.02095, over 59.00 utterances.], tot_loss[ctc_loss=0.07939, att_loss=0.2382, loss=0.2065, over 3275387.00 frames. utt_duration=1284 frames, utt_pad_proportion=0.04243, over 10212.20 utterances.], batch size: 59, lr: 6.04e-03, grad_scale: 16.0 2023-03-08 17:06:22,261 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.206e+02 2.084e+02 2.462e+02 2.947e+02 6.955e+02, threshold=4.923e+02, percent-clipped=4.0 2023-03-08 17:06:33,220 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69041.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:06:36,477 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69043.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:06:54,768 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69055.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:07:27,321 INFO [train2.py:809] (0/4) Epoch 18, batch 1350, loss[ctc_loss=0.09661, att_loss=0.2311, loss=0.2042, over 16168.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007389, over 41.00 utterances.], tot_loss[ctc_loss=0.07967, att_loss=0.2385, loss=0.2067, over 3284298.95 frames. utt_duration=1296 frames, utt_pad_proportion=0.03772, over 10148.45 utterances.], batch size: 41, lr: 6.03e-03, grad_scale: 16.0 2023-03-08 17:07:56,582 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9485, 3.6206, 3.6399, 3.1081, 3.7156, 3.7275, 3.6833, 2.8179], device='cuda:0'), covar=tensor([0.0865, 0.1512, 0.1972, 0.4163, 0.0952, 0.2306, 0.0853, 0.4119], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0166, 0.0174, 0.0236, 0.0141, 0.0236, 0.0155, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 17:08:10,418 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69103.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:08:10,594 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6078, 4.6484, 4.3162, 2.7425, 4.4100, 4.3449, 3.7080, 2.4783], device='cuda:0'), covar=tensor([0.0140, 0.0147, 0.0351, 0.1153, 0.0130, 0.0326, 0.0460, 0.1574], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0096, 0.0094, 0.0107, 0.0080, 0.0106, 0.0094, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 17:08:12,198 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69104.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:08:46,054 INFO [train2.py:809] (0/4) Epoch 18, batch 1400, loss[ctc_loss=0.09318, att_loss=0.2525, loss=0.2207, over 17104.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01575, over 56.00 utterances.], tot_loss[ctc_loss=0.0791, att_loss=0.2376, loss=0.2059, over 3280280.46 frames. utt_duration=1298 frames, utt_pad_proportion=0.03697, over 10117.08 utterances.], batch size: 56, lr: 6.03e-03, grad_scale: 16.0 2023-03-08 17:08:59,738 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.277e+02 2.123e+02 2.542e+02 3.068e+02 7.934e+02, threshold=5.085e+02, percent-clipped=3.0 2023-03-08 17:09:44,796 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69162.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:09:54,717 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69168.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:10:05,180 INFO [train2.py:809] (0/4) Epoch 18, batch 1450, loss[ctc_loss=0.07292, att_loss=0.2173, loss=0.1884, over 14601.00 frames. utt_duration=1827 frames, utt_pad_proportion=0.04267, over 32.00 utterances.], tot_loss[ctc_loss=0.07936, att_loss=0.2376, loss=0.2059, over 3273492.43 frames. utt_duration=1287 frames, utt_pad_proportion=0.04311, over 10184.21 utterances.], batch size: 32, lr: 6.03e-03, grad_scale: 16.0 2023-03-08 17:11:20,875 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69223.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:11:23,581 INFO [train2.py:809] (0/4) Epoch 18, batch 1500, loss[ctc_loss=0.07764, att_loss=0.2414, loss=0.2086, over 17012.00 frames. utt_duration=689 frames, utt_pad_proportion=0.1323, over 99.00 utterances.], tot_loss[ctc_loss=0.07973, att_loss=0.2383, loss=0.2066, over 3273825.53 frames. utt_duration=1257 frames, utt_pad_proportion=0.05128, over 10432.95 utterances.], batch size: 99, lr: 6.03e-03, grad_scale: 16.0 2023-03-08 17:11:23,963 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69225.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:11:25,517 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9142, 3.7745, 3.1658, 3.4254, 3.9563, 3.5788, 3.0703, 4.2391], device='cuda:0'), covar=tensor([0.1042, 0.0460, 0.1055, 0.0649, 0.0591, 0.0663, 0.0783, 0.0496], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0207, 0.0219, 0.0192, 0.0261, 0.0232, 0.0197, 0.0276], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 17:11:29,933 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69229.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:11:37,073 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.061e+02 2.532e+02 3.025e+02 5.458e+02, threshold=5.063e+02, percent-clipped=1.0 2023-03-08 17:11:37,455 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69234.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:11:54,383 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69245.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 17:12:39,642 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69273.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:12:42,420 INFO [train2.py:809] (0/4) Epoch 18, batch 1550, loss[ctc_loss=0.0923, att_loss=0.2555, loss=0.2229, over 17029.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009354, over 52.00 utterances.], tot_loss[ctc_loss=0.08018, att_loss=0.2388, loss=0.207, over 3271313.03 frames. utt_duration=1244 frames, utt_pad_proportion=0.05464, over 10534.65 utterances.], batch size: 52, lr: 6.02e-03, grad_scale: 16.0 2023-03-08 17:12:52,874 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69282.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:12:59,164 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69286.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:13:21,768 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-08 17:13:31,400 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69306.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 17:13:53,224 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69320.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:14:00,352 INFO [train2.py:809] (0/4) Epoch 18, batch 1600, loss[ctc_loss=0.08733, att_loss=0.2547, loss=0.2212, over 17328.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02304, over 59.00 utterances.], tot_loss[ctc_loss=0.07942, att_loss=0.2383, loss=0.2065, over 3271096.03 frames. utt_duration=1252 frames, utt_pad_proportion=0.05278, over 10460.60 utterances.], batch size: 59, lr: 6.02e-03, grad_scale: 16.0 2023-03-08 17:14:14,479 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.214e+02 1.944e+02 2.286e+02 2.991e+02 6.797e+02, threshold=4.572e+02, percent-clipped=3.0 2023-03-08 17:14:14,892 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69334.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 17:14:25,742 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69341.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:15:10,858 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69368.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:15:21,707 INFO [train2.py:809] (0/4) Epoch 18, batch 1650, loss[ctc_loss=0.07028, att_loss=0.2199, loss=0.19, over 16092.00 frames. utt_duration=1534 frames, utt_pad_proportion=0.005756, over 42.00 utterances.], tot_loss[ctc_loss=0.07943, att_loss=0.2391, loss=0.2072, over 3280710.71 frames. utt_duration=1262 frames, utt_pad_proportion=0.04733, over 10410.93 utterances.], batch size: 42, lr: 6.02e-03, grad_scale: 16.0 2023-03-08 17:15:31,628 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69381.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:15:44,461 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69389.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:15:52,751 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69394.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:16:02,185 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69399.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:16:42,697 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-03-08 17:16:44,764 INFO [train2.py:809] (0/4) Epoch 18, batch 1700, loss[ctc_loss=0.08745, att_loss=0.256, loss=0.2223, over 16543.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006032, over 45.00 utterances.], tot_loss[ctc_loss=0.07972, att_loss=0.2395, loss=0.2075, over 3282670.32 frames. utt_duration=1251 frames, utt_pad_proportion=0.05029, over 10508.97 utterances.], batch size: 45, lr: 6.02e-03, grad_scale: 16.0 2023-03-08 17:16:51,497 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69429.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:16:53,688 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-08 17:16:58,946 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.149e+02 2.524e+02 3.202e+02 9.381e+02, threshold=5.048e+02, percent-clipped=6.0 2023-03-08 17:17:11,831 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1327, 5.1668, 4.9878, 3.1059, 4.9775, 4.7037, 4.5925, 2.8512], device='cuda:0'), covar=tensor([0.0130, 0.0091, 0.0220, 0.0870, 0.0077, 0.0188, 0.0231, 0.1251], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0096, 0.0095, 0.0107, 0.0080, 0.0106, 0.0095, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 17:17:34,235 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69455.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:18:06,238 INFO [train2.py:809] (0/4) Epoch 18, batch 1750, loss[ctc_loss=0.07413, att_loss=0.2434, loss=0.2095, over 17360.00 frames. utt_duration=1008 frames, utt_pad_proportion=0.04933, over 69.00 utterances.], tot_loss[ctc_loss=0.07901, att_loss=0.2386, loss=0.2067, over 3281984.30 frames. utt_duration=1266 frames, utt_pad_proportion=0.04617, over 10378.51 utterances.], batch size: 69, lr: 6.02e-03, grad_scale: 16.0 2023-03-08 17:18:55,753 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4943, 5.1966, 5.3446, 5.3888, 5.1916, 5.3162, 5.1139, 4.7835], device='cuda:0'), covar=tensor([0.1653, 0.0771, 0.0338, 0.0444, 0.0738, 0.0388, 0.0371, 0.0423], device='cuda:0'), in_proj_covar=tensor([0.0499, 0.0334, 0.0315, 0.0331, 0.0393, 0.0403, 0.0332, 0.0371], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 17:19:17,080 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69518.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:19:26,186 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69524.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:19:27,554 INFO [train2.py:809] (0/4) Epoch 18, batch 1800, loss[ctc_loss=0.04605, att_loss=0.1969, loss=0.1668, over 14570.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.04361, over 32.00 utterances.], tot_loss[ctc_loss=0.07955, att_loss=0.2386, loss=0.2068, over 3273916.61 frames. utt_duration=1239 frames, utt_pad_proportion=0.05491, over 10584.82 utterances.], batch size: 32, lr: 6.01e-03, grad_scale: 16.0 2023-03-08 17:19:41,869 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.594e+02 2.055e+02 2.270e+02 2.783e+02 4.907e+02, threshold=4.541e+02, percent-clipped=0.0 2023-03-08 17:20:49,338 INFO [train2.py:809] (0/4) Epoch 18, batch 1850, loss[ctc_loss=0.07741, att_loss=0.235, loss=0.2035, over 17213.00 frames. utt_duration=872.8 frames, utt_pad_proportion=0.08314, over 79.00 utterances.], tot_loss[ctc_loss=0.07902, att_loss=0.2382, loss=0.2064, over 3275889.16 frames. utt_duration=1246 frames, utt_pad_proportion=0.05297, over 10528.44 utterances.], batch size: 79, lr: 6.01e-03, grad_scale: 16.0 2023-03-08 17:20:58,860 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69581.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:21:27,837 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5736, 4.9234, 4.7117, 4.7900, 4.9685, 4.7078, 3.4640, 4.9218], device='cuda:0'), covar=tensor([0.0141, 0.0131, 0.0153, 0.0112, 0.0120, 0.0115, 0.0793, 0.0227], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0085, 0.0106, 0.0066, 0.0071, 0.0083, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 17:21:31,762 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69601.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 17:22:09,413 INFO [train2.py:809] (0/4) Epoch 18, batch 1900, loss[ctc_loss=0.09894, att_loss=0.2527, loss=0.2219, over 17218.00 frames. utt_duration=873.4 frames, utt_pad_proportion=0.08547, over 79.00 utterances.], tot_loss[ctc_loss=0.07973, att_loss=0.2389, loss=0.207, over 3281323.14 frames. utt_duration=1261 frames, utt_pad_proportion=0.04732, over 10419.54 utterances.], batch size: 79, lr: 6.01e-03, grad_scale: 16.0 2023-03-08 17:22:15,614 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69629.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 17:22:18,016 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.37 vs. limit=5.0 2023-03-08 17:22:23,438 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 2.048e+02 2.360e+02 2.878e+02 5.569e+02, threshold=4.720e+02, percent-clipped=3.0 2023-03-08 17:23:28,935 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=69674.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 17:23:30,115 INFO [train2.py:809] (0/4) Epoch 18, batch 1950, loss[ctc_loss=0.08357, att_loss=0.25, loss=0.2167, over 16448.00 frames. utt_duration=1432 frames, utt_pad_proportion=0.007804, over 46.00 utterances.], tot_loss[ctc_loss=0.07979, att_loss=0.2385, loss=0.2068, over 3276075.07 frames. utt_duration=1254 frames, utt_pad_proportion=0.05078, over 10458.52 utterances.], batch size: 46, lr: 6.01e-03, grad_scale: 16.0 2023-03-08 17:23:31,799 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69676.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:24:09,191 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69699.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:24:48,815 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69724.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:24:50,199 INFO [train2.py:809] (0/4) Epoch 18, batch 2000, loss[ctc_loss=0.07941, att_loss=0.2457, loss=0.2124, over 16756.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006382, over 48.00 utterances.], tot_loss[ctc_loss=0.08045, att_loss=0.2392, loss=0.2074, over 3278959.24 frames. utt_duration=1257 frames, utt_pad_proportion=0.04908, over 10447.02 utterances.], batch size: 48, lr: 6.00e-03, grad_scale: 8.0 2023-03-08 17:24:59,990 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2005, 4.1103, 4.1626, 3.9945, 4.7238, 4.1809, 4.0999, 2.2851], device='cuda:0'), covar=tensor([0.0286, 0.0524, 0.0454, 0.0390, 0.0792, 0.0267, 0.0382, 0.1882], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0164, 0.0166, 0.0183, 0.0354, 0.0140, 0.0155, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 17:25:05,732 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.492e+02 2.144e+02 2.656e+02 3.349e+02 1.249e+03, threshold=5.313e+02, percent-clipped=13.0 2023-03-08 17:25:06,174 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=69735.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 17:25:07,712 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8560, 3.6363, 3.5695, 3.1502, 3.6734, 3.6540, 3.6577, 2.7516], device='cuda:0'), covar=tensor([0.0870, 0.1349, 0.2419, 0.3369, 0.1452, 0.2307, 0.0820, 0.4015], device='cuda:0'), in_proj_covar=tensor([0.0148, 0.0165, 0.0175, 0.0234, 0.0142, 0.0237, 0.0156, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 17:25:26,191 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69747.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:25:31,552 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=69750.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:26:10,679 INFO [train2.py:809] (0/4) Epoch 18, batch 2050, loss[ctc_loss=0.07352, att_loss=0.2348, loss=0.2025, over 16120.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006526, over 42.00 utterances.], tot_loss[ctc_loss=0.07972, att_loss=0.2393, loss=0.2074, over 3284810.91 frames. utt_duration=1257 frames, utt_pad_proportion=0.04823, over 10461.24 utterances.], batch size: 42, lr: 6.00e-03, grad_scale: 8.0 2023-03-08 17:26:23,798 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-08 17:27:19,472 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69818.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:27:28,515 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69824.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:27:29,777 INFO [train2.py:809] (0/4) Epoch 18, batch 2100, loss[ctc_loss=0.07629, att_loss=0.2125, loss=0.1852, over 16174.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007297, over 41.00 utterances.], tot_loss[ctc_loss=0.07954, att_loss=0.2386, loss=0.2068, over 3273538.46 frames. utt_duration=1262 frames, utt_pad_proportion=0.0501, over 10385.07 utterances.], batch size: 41, lr: 6.00e-03, grad_scale: 8.0 2023-03-08 17:27:45,295 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 1.987e+02 2.392e+02 3.047e+02 6.775e+02, threshold=4.784e+02, percent-clipped=2.0 2023-03-08 17:28:36,185 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69866.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:28:45,307 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69872.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:28:49,864 INFO [train2.py:809] (0/4) Epoch 18, batch 2150, loss[ctc_loss=0.09334, att_loss=0.2568, loss=0.2241, over 16766.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006496, over 48.00 utterances.], tot_loss[ctc_loss=0.07944, att_loss=0.2385, loss=0.2067, over 3270878.84 frames. utt_duration=1252 frames, utt_pad_proportion=0.05453, over 10460.52 utterances.], batch size: 48, lr: 6.00e-03, grad_scale: 8.0 2023-03-08 17:28:59,261 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69881.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:29:31,194 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69901.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 17:30:08,950 INFO [train2.py:809] (0/4) Epoch 18, batch 2200, loss[ctc_loss=0.07652, att_loss=0.2432, loss=0.2099, over 16500.00 frames. utt_duration=1437 frames, utt_pad_proportion=0.004489, over 46.00 utterances.], tot_loss[ctc_loss=0.08054, att_loss=0.2395, loss=0.2077, over 3280003.19 frames. utt_duration=1257 frames, utt_pad_proportion=0.05124, over 10450.41 utterances.], batch size: 46, lr: 6.00e-03, grad_scale: 8.0 2023-03-08 17:30:15,134 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69929.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:30:15,348 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69929.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 17:30:24,844 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.283e+02 2.018e+02 2.397e+02 3.065e+02 5.583e+02, threshold=4.795e+02, percent-clipped=3.0 2023-03-08 17:30:47,791 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69949.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 17:31:28,420 INFO [train2.py:809] (0/4) Epoch 18, batch 2250, loss[ctc_loss=0.0976, att_loss=0.2612, loss=0.2285, over 17398.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.02969, over 63.00 utterances.], tot_loss[ctc_loss=0.08029, att_loss=0.2389, loss=0.2072, over 3268687.25 frames. utt_duration=1227 frames, utt_pad_proportion=0.06198, over 10666.62 utterances.], batch size: 63, lr: 5.99e-03, grad_scale: 8.0 2023-03-08 17:31:30,214 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=69976.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:31:31,463 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=69977.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:31:34,742 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2805, 5.2720, 5.0796, 3.3313, 5.1404, 4.8623, 4.6871, 3.0685], device='cuda:0'), covar=tensor([0.0105, 0.0089, 0.0242, 0.0832, 0.0076, 0.0166, 0.0248, 0.1162], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0097, 0.0096, 0.0108, 0.0081, 0.0105, 0.0096, 0.0101], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 17:31:34,806 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2515, 4.0058, 3.4333, 3.6934, 4.0816, 3.8078, 3.4291, 4.4783], device='cuda:0'), covar=tensor([0.0949, 0.0532, 0.1089, 0.0613, 0.0778, 0.0768, 0.0709, 0.0475], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0211, 0.0223, 0.0193, 0.0267, 0.0234, 0.0200, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 17:32:09,797 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-70000.pt 2023-03-08 17:32:51,050 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:32:51,247 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70024.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:32:52,442 INFO [train2.py:809] (0/4) Epoch 18, batch 2300, loss[ctc_loss=0.07177, att_loss=0.239, loss=0.2055, over 16543.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005255, over 45.00 utterances.], tot_loss[ctc_loss=0.08054, att_loss=0.2393, loss=0.2076, over 3271324.37 frames. utt_duration=1203 frames, utt_pad_proportion=0.06679, over 10894.78 utterances.], batch size: 45, lr: 5.99e-03, grad_scale: 8.0 2023-03-08 17:33:01,000 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70030.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 17:33:03,985 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0529, 5.2860, 5.6084, 5.3801, 5.4908, 5.9612, 5.1073, 6.0948], device='cuda:0'), covar=tensor([0.0692, 0.0771, 0.0818, 0.1275, 0.1869, 0.0934, 0.0768, 0.0607], device='cuda:0'), in_proj_covar=tensor([0.0834, 0.0484, 0.0573, 0.0636, 0.0838, 0.0590, 0.0477, 0.0580], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 17:33:09,215 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 2.121e+02 2.733e+02 3.440e+02 9.117e+02, threshold=5.466e+02, percent-clipped=8.0 2023-03-08 17:33:34,372 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70050.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:34:02,066 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6162, 2.6056, 4.9941, 3.9033, 3.0698, 4.3088, 4.7640, 4.6189], device='cuda:0'), covar=tensor([0.0265, 0.1746, 0.0204, 0.0992, 0.1798, 0.0274, 0.0155, 0.0284], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0242, 0.0171, 0.0312, 0.0266, 0.0205, 0.0156, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 17:34:07,917 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70072.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:34:12,449 INFO [train2.py:809] (0/4) Epoch 18, batch 2350, loss[ctc_loss=0.07392, att_loss=0.2226, loss=0.1928, over 10567.00 frames. utt_duration=1839 frames, utt_pad_proportion=0.215, over 23.00 utterances.], tot_loss[ctc_loss=0.08075, att_loss=0.2392, loss=0.2075, over 3257331.14 frames. utt_duration=1208 frames, utt_pad_proportion=0.0674, over 10795.82 utterances.], batch size: 23, lr: 5.99e-03, grad_scale: 8.0 2023-03-08 17:34:50,234 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70098.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:35:08,666 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70110.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:35:31,916 INFO [train2.py:809] (0/4) Epoch 18, batch 2400, loss[ctc_loss=0.08303, att_loss=0.2547, loss=0.2203, over 17293.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01245, over 55.00 utterances.], tot_loss[ctc_loss=0.08141, att_loss=0.2398, loss=0.2081, over 3264853.44 frames. utt_duration=1212 frames, utt_pad_proportion=0.06487, over 10789.82 utterances.], batch size: 55, lr: 5.99e-03, grad_scale: 8.0 2023-03-08 17:35:48,285 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.485e+02 2.097e+02 2.428e+02 2.770e+02 6.273e+02, threshold=4.856e+02, percent-clipped=1.0 2023-03-08 17:36:44,147 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9585, 4.7539, 4.8080, 4.6749, 5.1868, 4.6963, 4.6679, 2.2593], device='cuda:0'), covar=tensor([0.0120, 0.0147, 0.0140, 0.0142, 0.0853, 0.0146, 0.0171, 0.2079], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0165, 0.0168, 0.0185, 0.0356, 0.0141, 0.0157, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 17:36:45,619 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70171.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:36:52,019 INFO [train2.py:809] (0/4) Epoch 18, batch 2450, loss[ctc_loss=0.07402, att_loss=0.2257, loss=0.1953, over 15761.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009452, over 38.00 utterances.], tot_loss[ctc_loss=0.08149, att_loss=0.2398, loss=0.2082, over 3265390.51 frames. utt_duration=1187 frames, utt_pad_proportion=0.07104, over 11015.86 utterances.], batch size: 38, lr: 5.99e-03, grad_scale: 8.0 2023-03-08 17:37:16,673 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 17:37:19,581 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 17:37:56,634 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8615, 3.5649, 3.6167, 3.1158, 3.6363, 3.7382, 3.6986, 2.6990], device='cuda:0'), covar=tensor([0.1240, 0.2084, 0.2559, 0.4076, 0.1807, 0.3066, 0.1798, 0.5287], device='cuda:0'), in_proj_covar=tensor([0.0149, 0.0168, 0.0178, 0.0238, 0.0143, 0.0239, 0.0157, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 17:38:11,065 INFO [train2.py:809] (0/4) Epoch 18, batch 2500, loss[ctc_loss=0.0717, att_loss=0.2218, loss=0.1918, over 13676.00 frames. utt_duration=1825 frames, utt_pad_proportion=0.07131, over 30.00 utterances.], tot_loss[ctc_loss=0.08136, att_loss=0.2389, loss=0.2074, over 3249150.63 frames. utt_duration=1191 frames, utt_pad_proportion=0.07467, over 10927.47 utterances.], batch size: 30, lr: 5.98e-03, grad_scale: 8.0 2023-03-08 17:38:26,724 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.323e+02 2.106e+02 2.442e+02 2.921e+02 5.704e+02, threshold=4.884e+02, percent-clipped=1.0 2023-03-08 17:38:55,355 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1876, 5.1797, 4.9048, 3.0183, 4.9222, 4.7553, 4.3426, 2.7765], device='cuda:0'), covar=tensor([0.0104, 0.0083, 0.0295, 0.0989, 0.0096, 0.0164, 0.0309, 0.1344], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0097, 0.0097, 0.0109, 0.0082, 0.0107, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 17:39:06,516 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70260.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:39:28,525 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-03-08 17:39:31,628 INFO [train2.py:809] (0/4) Epoch 18, batch 2550, loss[ctc_loss=0.056, att_loss=0.2251, loss=0.1913, over 16170.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006842, over 41.00 utterances.], tot_loss[ctc_loss=0.08082, att_loss=0.2391, loss=0.2074, over 3260331.84 frames. utt_duration=1213 frames, utt_pad_proportion=0.06702, over 10760.87 utterances.], batch size: 41, lr: 5.98e-03, grad_scale: 8.0 2023-03-08 17:39:52,739 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4612, 5.0212, 5.1669, 5.2556, 4.9524, 5.1463, 4.8524, 4.5785], device='cuda:0'), covar=tensor([0.1602, 0.0817, 0.0374, 0.0492, 0.0773, 0.0493, 0.0461, 0.0442], device='cuda:0'), in_proj_covar=tensor([0.0503, 0.0338, 0.0318, 0.0335, 0.0399, 0.0408, 0.0339, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 17:40:05,200 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7264, 3.4514, 3.4756, 3.0338, 3.6324, 3.5638, 3.5772, 2.5761], device='cuda:0'), covar=tensor([0.1028, 0.1593, 0.2117, 0.3542, 0.0826, 0.2214, 0.0777, 0.4462], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0169, 0.0179, 0.0238, 0.0143, 0.0239, 0.0157, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 17:40:08,376 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4847, 2.8909, 3.6140, 3.0754, 3.4911, 4.5246, 4.3986, 3.0419], device='cuda:0'), covar=tensor([0.0344, 0.1780, 0.1268, 0.1306, 0.1151, 0.0881, 0.0537, 0.1486], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0241, 0.0272, 0.0215, 0.0264, 0.0356, 0.0253, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 17:40:22,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8666, 4.7353, 4.7902, 4.4739, 5.4282, 4.4973, 4.7504, 2.4949], device='cuda:0'), covar=tensor([0.0162, 0.0313, 0.0276, 0.0482, 0.0675, 0.0200, 0.0293, 0.1958], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0165, 0.0167, 0.0185, 0.0355, 0.0141, 0.0157, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 17:40:36,391 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8533, 5.2433, 4.7188, 5.3038, 4.6383, 4.9152, 5.3482, 5.1231], device='cuda:0'), covar=tensor([0.0639, 0.0303, 0.0829, 0.0339, 0.0470, 0.0275, 0.0257, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0375, 0.0306, 0.0355, 0.0323, 0.0309, 0.0234, 0.0289, 0.0274], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 17:40:43,944 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=70321.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:40:50,212 INFO [train2.py:809] (0/4) Epoch 18, batch 2600, loss[ctc_loss=0.1066, att_loss=0.2612, loss=0.2303, over 17011.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.009251, over 51.00 utterances.], tot_loss[ctc_loss=0.08057, att_loss=0.239, loss=0.2074, over 3261806.42 frames. utt_duration=1237 frames, utt_pad_proportion=0.06139, over 10557.69 utterances.], batch size: 51, lr: 5.98e-03, grad_scale: 8.0 2023-03-08 17:40:58,252 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70330.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 17:41:05,530 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 1.967e+02 2.510e+02 3.094e+02 6.487e+02, threshold=5.020e+02, percent-clipped=1.0 2023-03-08 17:42:08,695 INFO [train2.py:809] (0/4) Epoch 18, batch 2650, loss[ctc_loss=0.08916, att_loss=0.2481, loss=0.2163, over 17449.00 frames. utt_duration=1013 frames, utt_pad_proportion=0.04336, over 69.00 utterances.], tot_loss[ctc_loss=0.08098, att_loss=0.2391, loss=0.2075, over 3258918.79 frames. utt_duration=1259 frames, utt_pad_proportion=0.05669, over 10364.47 utterances.], batch size: 69, lr: 5.98e-03, grad_scale: 8.0 2023-03-08 17:42:13,358 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70378.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 17:43:27,230 INFO [train2.py:809] (0/4) Epoch 18, batch 2700, loss[ctc_loss=0.07844, att_loss=0.2301, loss=0.1998, over 16008.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007342, over 40.00 utterances.], tot_loss[ctc_loss=0.0808, att_loss=0.2388, loss=0.2072, over 3259979.26 frames. utt_duration=1260 frames, utt_pad_proportion=0.05633, over 10362.51 utterances.], batch size: 40, lr: 5.98e-03, grad_scale: 8.0 2023-03-08 17:43:43,123 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.380e+02 1.957e+02 2.578e+02 3.179e+02 5.678e+02, threshold=5.155e+02, percent-clipped=6.0 2023-03-08 17:43:47,973 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5003, 2.9678, 3.3889, 4.5729, 4.0202, 3.9909, 2.9903, 2.3060], device='cuda:0'), covar=tensor([0.0723, 0.2050, 0.1022, 0.0460, 0.0829, 0.0451, 0.1605, 0.2372], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0214, 0.0186, 0.0211, 0.0213, 0.0171, 0.0200, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 17:44:17,155 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7580, 3.0892, 3.7402, 3.2890, 3.6787, 4.7528, 4.6106, 3.3294], device='cuda:0'), covar=tensor([0.0312, 0.1740, 0.1194, 0.1206, 0.1058, 0.0809, 0.0555, 0.1382], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0240, 0.0270, 0.0212, 0.0260, 0.0351, 0.0249, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 17:44:31,285 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70466.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:44:46,104 INFO [train2.py:809] (0/4) Epoch 18, batch 2750, loss[ctc_loss=0.06654, att_loss=0.2171, loss=0.187, over 15637.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009384, over 37.00 utterances.], tot_loss[ctc_loss=0.07994, att_loss=0.2385, loss=0.2068, over 3264017.70 frames. utt_duration=1237 frames, utt_pad_proportion=0.0597, over 10565.51 utterances.], batch size: 37, lr: 5.97e-03, grad_scale: 8.0 2023-03-08 17:45:33,068 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5633, 2.3048, 2.0487, 2.5677, 2.8385, 2.4290, 2.3348, 2.7497], device='cuda:0'), covar=tensor([0.1744, 0.3022, 0.2959, 0.1789, 0.1685, 0.1355, 0.2314, 0.1196], device='cuda:0'), in_proj_covar=tensor([0.0105, 0.0113, 0.0113, 0.0098, 0.0110, 0.0097, 0.0116, 0.0086], device='cuda:0'), out_proj_covar=tensor([7.8067e-05, 8.6489e-05, 8.7738e-05, 7.5574e-05, 8.0472e-05, 7.6771e-05, 8.5986e-05, 6.9235e-05], device='cuda:0') 2023-03-08 17:46:05,047 INFO [train2.py:809] (0/4) Epoch 18, batch 2800, loss[ctc_loss=0.09569, att_loss=0.2597, loss=0.2269, over 17278.00 frames. utt_duration=876.3 frames, utt_pad_proportion=0.07656, over 79.00 utterances.], tot_loss[ctc_loss=0.08028, att_loss=0.2386, loss=0.2069, over 3264113.82 frames. utt_duration=1224 frames, utt_pad_proportion=0.06309, over 10676.63 utterances.], batch size: 79, lr: 5.97e-03, grad_scale: 8.0 2023-03-08 17:46:11,359 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4478, 2.8331, 3.3004, 4.3671, 3.8867, 3.8937, 3.0164, 2.1348], device='cuda:0'), covar=tensor([0.0670, 0.2221, 0.0981, 0.0652, 0.0810, 0.0551, 0.1578, 0.2637], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0214, 0.0186, 0.0212, 0.0212, 0.0171, 0.0199, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 17:46:20,628 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 1.948e+02 2.363e+02 2.957e+02 5.460e+02, threshold=4.725e+02, percent-clipped=2.0 2023-03-08 17:47:24,018 INFO [train2.py:809] (0/4) Epoch 18, batch 2850, loss[ctc_loss=0.08111, att_loss=0.2532, loss=0.2188, over 17030.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.01009, over 53.00 utterances.], tot_loss[ctc_loss=0.08035, att_loss=0.2386, loss=0.207, over 3267485.29 frames. utt_duration=1247 frames, utt_pad_proportion=0.05752, over 10492.01 utterances.], batch size: 53, lr: 5.97e-03, grad_scale: 8.0 2023-03-08 17:47:56,392 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0582, 5.3760, 5.3259, 5.2387, 5.4045, 5.3892, 5.0330, 4.7794], device='cuda:0'), covar=tensor([0.1052, 0.0497, 0.0240, 0.0552, 0.0313, 0.0289, 0.0386, 0.0333], device='cuda:0'), in_proj_covar=tensor([0.0503, 0.0337, 0.0317, 0.0335, 0.0397, 0.0408, 0.0338, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 17:48:29,672 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=70616.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:48:43,289 INFO [train2.py:809] (0/4) Epoch 18, batch 2900, loss[ctc_loss=0.08547, att_loss=0.2274, loss=0.199, over 16012.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007063, over 40.00 utterances.], tot_loss[ctc_loss=0.08061, att_loss=0.2389, loss=0.2073, over 3266144.61 frames. utt_duration=1221 frames, utt_pad_proportion=0.06361, over 10708.94 utterances.], batch size: 40, lr: 5.97e-03, grad_scale: 8.0 2023-03-08 17:48:59,498 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 1.930e+02 2.447e+02 2.876e+02 5.952e+02, threshold=4.894e+02, percent-clipped=4.0 2023-03-08 17:50:04,765 INFO [train2.py:809] (0/4) Epoch 18, batch 2950, loss[ctc_loss=0.09406, att_loss=0.2525, loss=0.2208, over 17474.00 frames. utt_duration=1014 frames, utt_pad_proportion=0.0431, over 69.00 utterances.], tot_loss[ctc_loss=0.08105, att_loss=0.2395, loss=0.2078, over 3259182.11 frames. utt_duration=1180 frames, utt_pad_proportion=0.07436, over 11057.61 utterances.], batch size: 69, lr: 5.96e-03, grad_scale: 8.0 2023-03-08 17:50:07,802 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-08 17:51:26,429 INFO [train2.py:809] (0/4) Epoch 18, batch 3000, loss[ctc_loss=0.08722, att_loss=0.2491, loss=0.2168, over 17035.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009826, over 52.00 utterances.], tot_loss[ctc_loss=0.08067, att_loss=0.2396, loss=0.2078, over 3269408.06 frames. utt_duration=1191 frames, utt_pad_proportion=0.06877, over 10997.97 utterances.], batch size: 52, lr: 5.96e-03, grad_scale: 8.0 2023-03-08 17:51:26,431 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 17:51:43,589 INFO [train2.py:843] (0/4) Epoch 18, validation: ctc_loss=0.04147, att_loss=0.2347, loss=0.196, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 17:51:43,589 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 17:51:59,783 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.535e+02 1.987e+02 2.490e+02 3.021e+02 7.545e+02, threshold=4.980e+02, percent-clipped=3.0 2023-03-08 17:52:08,408 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.87 vs. limit=5.0 2023-03-08 17:52:40,273 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3913, 4.4779, 4.6044, 4.4228, 5.0508, 4.4068, 4.4803, 2.3670], device='cuda:0'), covar=tensor([0.0245, 0.0286, 0.0259, 0.0292, 0.0827, 0.0224, 0.0301, 0.2033], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0165, 0.0168, 0.0186, 0.0356, 0.0141, 0.0157, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 17:52:49,422 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70766.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:52:53,982 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1174, 3.9702, 3.3182, 3.5199, 4.0235, 3.6942, 3.2021, 4.3728], device='cuda:0'), covar=tensor([0.0989, 0.0523, 0.1011, 0.0745, 0.0704, 0.0744, 0.0866, 0.0472], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0214, 0.0225, 0.0194, 0.0269, 0.0235, 0.0200, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 17:53:03,421 INFO [train2.py:809] (0/4) Epoch 18, batch 3050, loss[ctc_loss=0.07411, att_loss=0.2242, loss=0.1941, over 15950.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.006798, over 41.00 utterances.], tot_loss[ctc_loss=0.08084, att_loss=0.2391, loss=0.2074, over 3259536.86 frames. utt_duration=1180 frames, utt_pad_proportion=0.07507, over 11060.48 utterances.], batch size: 41, lr: 5.96e-03, grad_scale: 8.0 2023-03-08 17:53:03,553 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9531, 6.1347, 5.5735, 5.9032, 5.7965, 5.2679, 5.5600, 5.3661], device='cuda:0'), covar=tensor([0.1148, 0.0805, 0.0819, 0.0771, 0.0788, 0.1544, 0.2033, 0.2201], device='cuda:0'), in_proj_covar=tensor([0.0501, 0.0586, 0.0439, 0.0436, 0.0414, 0.0450, 0.0594, 0.0512], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 17:53:13,000 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8093, 3.3852, 3.8530, 3.2907, 3.8145, 4.7827, 4.6739, 3.5363], device='cuda:0'), covar=tensor([0.0293, 0.1406, 0.1105, 0.1306, 0.0892, 0.0979, 0.0451, 0.1144], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0240, 0.0271, 0.0215, 0.0261, 0.0354, 0.0251, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 17:53:14,426 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7699, 5.0957, 5.3030, 5.1342, 5.2992, 5.7354, 5.0724, 5.8245], device='cuda:0'), covar=tensor([0.0763, 0.0772, 0.0914, 0.1455, 0.1660, 0.0891, 0.0782, 0.0740], device='cuda:0'), in_proj_covar=tensor([0.0839, 0.0492, 0.0574, 0.0641, 0.0842, 0.0598, 0.0473, 0.0589], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 17:53:54,018 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-03-08 17:54:07,305 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70814.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:54:26,013 INFO [train2.py:809] (0/4) Epoch 18, batch 3100, loss[ctc_loss=0.05827, att_loss=0.2194, loss=0.1872, over 12413.00 frames. utt_duration=1841 frames, utt_pad_proportion=0.126, over 27.00 utterances.], tot_loss[ctc_loss=0.0805, att_loss=0.2385, loss=0.2069, over 3253960.90 frames. utt_duration=1187 frames, utt_pad_proportion=0.07385, over 10977.54 utterances.], batch size: 27, lr: 5.96e-03, grad_scale: 8.0 2023-03-08 17:54:41,866 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 2.010e+02 2.337e+02 3.007e+02 9.236e+02, threshold=4.674e+02, percent-clipped=2.0 2023-03-08 17:55:46,922 INFO [train2.py:809] (0/4) Epoch 18, batch 3150, loss[ctc_loss=0.06844, att_loss=0.2179, loss=0.188, over 15885.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009113, over 39.00 utterances.], tot_loss[ctc_loss=0.08049, att_loss=0.2387, loss=0.2071, over 3254646.16 frames. utt_duration=1191 frames, utt_pad_proportion=0.07344, over 10945.19 utterances.], batch size: 39, lr: 5.96e-03, grad_scale: 8.0 2023-03-08 17:56:52,149 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=70916.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:57:07,451 INFO [train2.py:809] (0/4) Epoch 18, batch 3200, loss[ctc_loss=0.05994, att_loss=0.2204, loss=0.1883, over 16016.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007214, over 40.00 utterances.], tot_loss[ctc_loss=0.08056, att_loss=0.239, loss=0.2073, over 3260488.01 frames. utt_duration=1207 frames, utt_pad_proportion=0.06875, over 10815.46 utterances.], batch size: 40, lr: 5.95e-03, grad_scale: 8.0 2023-03-08 17:57:08,297 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-08 17:57:23,346 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 2.043e+02 2.547e+02 2.997e+02 7.270e+02, threshold=5.093e+02, percent-clipped=4.0 2023-03-08 17:57:27,515 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-03-08 17:57:53,882 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.66 vs. limit=5.0 2023-03-08 17:58:10,071 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=70964.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:58:28,691 INFO [train2.py:809] (0/4) Epoch 18, batch 3250, loss[ctc_loss=0.05379, att_loss=0.2076, loss=0.1769, over 15744.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.009008, over 38.00 utterances.], tot_loss[ctc_loss=0.08076, att_loss=0.2395, loss=0.2078, over 3271408.88 frames. utt_duration=1191 frames, utt_pad_proportion=0.06863, over 11002.23 utterances.], batch size: 38, lr: 5.95e-03, grad_scale: 8.0 2023-03-08 17:58:56,325 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=70992.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 17:59:47,174 INFO [train2.py:809] (0/4) Epoch 18, batch 3300, loss[ctc_loss=0.07737, att_loss=0.241, loss=0.2083, over 16619.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005827, over 47.00 utterances.], tot_loss[ctc_loss=0.08016, att_loss=0.2385, loss=0.2068, over 3269519.59 frames. utt_duration=1219 frames, utt_pad_proportion=0.06251, over 10739.35 utterances.], batch size: 47, lr: 5.95e-03, grad_scale: 8.0 2023-03-08 17:59:55,934 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71030.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:00:02,924 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 1.989e+02 2.342e+02 2.870e+02 6.957e+02, threshold=4.685e+02, percent-clipped=3.0 2023-03-08 18:00:19,839 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0538, 5.4018, 4.8118, 5.5051, 4.8746, 5.0592, 5.5111, 5.3106], device='cuda:0'), covar=tensor([0.0498, 0.0257, 0.0867, 0.0237, 0.0364, 0.0211, 0.0213, 0.0177], device='cuda:0'), in_proj_covar=tensor([0.0373, 0.0303, 0.0354, 0.0321, 0.0305, 0.0231, 0.0287, 0.0270], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 18:00:29,690 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1269, 5.1452, 4.9210, 2.9798, 4.8701, 4.6805, 4.4059, 2.9272], device='cuda:0'), covar=tensor([0.0139, 0.0089, 0.0270, 0.1027, 0.0101, 0.0192, 0.0290, 0.1335], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0098, 0.0097, 0.0109, 0.0081, 0.0107, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 18:00:31,317 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71053.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:01:05,448 INFO [train2.py:809] (0/4) Epoch 18, batch 3350, loss[ctc_loss=0.07815, att_loss=0.2258, loss=0.1962, over 16011.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006846, over 40.00 utterances.], tot_loss[ctc_loss=0.07957, att_loss=0.238, loss=0.2063, over 3273511.93 frames. utt_duration=1255 frames, utt_pad_proportion=0.05325, over 10448.42 utterances.], batch size: 40, lr: 5.95e-03, grad_scale: 8.0 2023-03-08 18:01:31,194 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71091.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:01:42,580 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3767, 4.7190, 4.6147, 4.6432, 4.7763, 4.3666, 3.0607, 4.6511], device='cuda:0'), covar=tensor([0.0116, 0.0111, 0.0129, 0.0090, 0.0094, 0.0127, 0.0822, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0085, 0.0105, 0.0066, 0.0072, 0.0082, 0.0101, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:02:06,651 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=6.21 vs. limit=5.0 2023-03-08 18:02:25,933 INFO [train2.py:809] (0/4) Epoch 18, batch 3400, loss[ctc_loss=0.07326, att_loss=0.2365, loss=0.2038, over 16555.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005387, over 45.00 utterances.], tot_loss[ctc_loss=0.07973, att_loss=0.2384, loss=0.2067, over 3278179.02 frames. utt_duration=1252 frames, utt_pad_proportion=0.0521, over 10484.49 utterances.], batch size: 45, lr: 5.95e-03, grad_scale: 8.0 2023-03-08 18:02:42,140 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.526e+02 2.089e+02 2.385e+02 3.204e+02 8.386e+02, threshold=4.770e+02, percent-clipped=4.0 2023-03-08 18:03:23,867 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1889, 4.4216, 4.9484, 4.8997, 3.0277, 4.6699, 3.1310, 1.6459], device='cuda:0'), covar=tensor([0.0353, 0.0278, 0.0429, 0.0182, 0.1459, 0.0149, 0.1176, 0.1760], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0154, 0.0258, 0.0148, 0.0225, 0.0130, 0.0233, 0.0207], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:03:46,795 INFO [train2.py:809] (0/4) Epoch 18, batch 3450, loss[ctc_loss=0.06866, att_loss=0.2335, loss=0.2005, over 15893.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008121, over 39.00 utterances.], tot_loss[ctc_loss=0.07953, att_loss=0.2381, loss=0.2064, over 3274615.71 frames. utt_duration=1269 frames, utt_pad_proportion=0.04912, over 10331.40 utterances.], batch size: 39, lr: 5.94e-03, grad_scale: 8.0 2023-03-08 18:04:01,753 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5043, 2.7288, 3.6666, 2.9671, 3.4420, 4.6361, 4.4819, 3.0065], device='cuda:0'), covar=tensor([0.0385, 0.2051, 0.1057, 0.1408, 0.1130, 0.0689, 0.0438, 0.1569], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0239, 0.0270, 0.0214, 0.0257, 0.0351, 0.0249, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:05:05,409 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2107, 2.7529, 3.5361, 2.8951, 3.2936, 4.3607, 4.1529, 2.8557], device='cuda:0'), covar=tensor([0.0390, 0.1736, 0.1133, 0.1343, 0.1084, 0.0806, 0.0638, 0.1510], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0240, 0.0271, 0.0215, 0.0259, 0.0353, 0.0250, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:05:06,664 INFO [train2.py:809] (0/4) Epoch 18, batch 3500, loss[ctc_loss=0.1034, att_loss=0.2622, loss=0.2304, over 17486.00 frames. utt_duration=1015 frames, utt_pad_proportion=0.04123, over 69.00 utterances.], tot_loss[ctc_loss=0.07881, att_loss=0.2384, loss=0.2065, over 3279581.64 frames. utt_duration=1286 frames, utt_pad_proportion=0.04364, over 10215.86 utterances.], batch size: 69, lr: 5.94e-03, grad_scale: 8.0 2023-03-08 18:05:22,701 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 1.989e+02 2.438e+02 2.819e+02 5.372e+02, threshold=4.876e+02, percent-clipped=3.0 2023-03-08 18:06:26,580 INFO [train2.py:809] (0/4) Epoch 18, batch 3550, loss[ctc_loss=0.09213, att_loss=0.2474, loss=0.2164, over 17293.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01249, over 55.00 utterances.], tot_loss[ctc_loss=0.07895, att_loss=0.2385, loss=0.2066, over 3273413.90 frames. utt_duration=1268 frames, utt_pad_proportion=0.05018, over 10341.57 utterances.], batch size: 55, lr: 5.94e-03, grad_scale: 8.0 2023-03-08 18:06:49,446 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3566, 2.4539, 4.9534, 3.8734, 2.9415, 4.2105, 4.5834, 4.5439], device='cuda:0'), covar=tensor([0.0256, 0.1693, 0.0145, 0.0886, 0.1758, 0.0263, 0.0138, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0238, 0.0168, 0.0306, 0.0263, 0.0203, 0.0154, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 18:07:46,253 INFO [train2.py:809] (0/4) Epoch 18, batch 3600, loss[ctc_loss=0.0887, att_loss=0.2458, loss=0.2144, over 16625.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.0052, over 47.00 utterances.], tot_loss[ctc_loss=0.07937, att_loss=0.2393, loss=0.2073, over 3282585.60 frames. utt_duration=1282 frames, utt_pad_proportion=0.04409, over 10257.07 utterances.], batch size: 47, lr: 5.94e-03, grad_scale: 8.0 2023-03-08 18:08:02,039 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.388e+02 1.997e+02 2.305e+02 2.777e+02 7.963e+02, threshold=4.609e+02, percent-clipped=3.0 2023-03-08 18:08:23,274 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71348.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:08:46,134 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6968, 3.2143, 3.7843, 3.3312, 3.6319, 4.7440, 4.5582, 3.4427], device='cuda:0'), covar=tensor([0.0371, 0.1439, 0.1165, 0.1204, 0.1082, 0.0936, 0.0561, 0.1278], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0240, 0.0272, 0.0214, 0.0258, 0.0353, 0.0251, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:09:09,344 INFO [train2.py:809] (0/4) Epoch 18, batch 3650, loss[ctc_loss=0.0856, att_loss=0.2523, loss=0.219, over 17068.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008803, over 53.00 utterances.], tot_loss[ctc_loss=0.07949, att_loss=0.2393, loss=0.2073, over 3278904.11 frames. utt_duration=1264 frames, utt_pad_proportion=0.04859, over 10388.31 utterances.], batch size: 53, lr: 5.94e-03, grad_scale: 8.0 2023-03-08 18:09:28,352 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71386.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:09:57,334 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.18 vs. limit=5.0 2023-03-08 18:10:07,891 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0629, 5.0074, 4.7618, 3.1243, 4.8404, 4.7570, 4.3344, 2.7781], device='cuda:0'), covar=tensor([0.0122, 0.0119, 0.0360, 0.0903, 0.0107, 0.0188, 0.0316, 0.1388], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0099, 0.0098, 0.0109, 0.0082, 0.0108, 0.0098, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 18:10:33,055 INFO [train2.py:809] (0/4) Epoch 18, batch 3700, loss[ctc_loss=0.0572, att_loss=0.2218, loss=0.1889, over 16183.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006075, over 41.00 utterances.], tot_loss[ctc_loss=0.07937, att_loss=0.239, loss=0.2071, over 3276888.97 frames. utt_duration=1268 frames, utt_pad_proportion=0.04975, over 10346.51 utterances.], batch size: 41, lr: 5.93e-03, grad_scale: 8.0 2023-03-08 18:10:49,956 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 2.052e+02 2.398e+02 2.900e+02 5.315e+02, threshold=4.797e+02, percent-clipped=3.0 2023-03-08 18:11:24,745 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-03-08 18:11:56,300 INFO [train2.py:809] (0/4) Epoch 18, batch 3750, loss[ctc_loss=0.09035, att_loss=0.2619, loss=0.2276, over 17116.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01436, over 56.00 utterances.], tot_loss[ctc_loss=0.0793, att_loss=0.2393, loss=0.2073, over 3279627.04 frames. utt_duration=1265 frames, utt_pad_proportion=0.04904, over 10385.71 utterances.], batch size: 56, lr: 5.93e-03, grad_scale: 8.0 2023-03-08 18:13:12,727 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7114, 3.2449, 3.7600, 3.3373, 3.6945, 4.7888, 4.5334, 3.4219], device='cuda:0'), covar=tensor([0.0335, 0.1517, 0.1165, 0.1208, 0.0981, 0.0746, 0.0585, 0.1278], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0239, 0.0271, 0.0213, 0.0258, 0.0352, 0.0251, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:13:18,685 INFO [train2.py:809] (0/4) Epoch 18, batch 3800, loss[ctc_loss=0.0925, att_loss=0.2394, loss=0.2101, over 15963.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.0067, over 41.00 utterances.], tot_loss[ctc_loss=0.07969, att_loss=0.2396, loss=0.2076, over 3287054.93 frames. utt_duration=1268 frames, utt_pad_proportion=0.04691, over 10384.71 utterances.], batch size: 41, lr: 5.93e-03, grad_scale: 8.0 2023-03-08 18:13:34,473 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.420e+02 2.121e+02 2.472e+02 3.058e+02 5.629e+02, threshold=4.945e+02, percent-clipped=2.0 2023-03-08 18:13:37,802 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8989, 3.7203, 3.1304, 3.3655, 3.8934, 3.5591, 3.0679, 4.1297], device='cuda:0'), covar=tensor([0.1018, 0.0577, 0.1034, 0.0717, 0.0674, 0.0733, 0.0808, 0.0711], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0214, 0.0222, 0.0194, 0.0268, 0.0233, 0.0198, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 18:14:15,411 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-08 18:14:23,914 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-03-08 18:14:37,808 INFO [train2.py:809] (0/4) Epoch 18, batch 3850, loss[ctc_loss=0.06349, att_loss=0.2204, loss=0.189, over 16005.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007419, over 40.00 utterances.], tot_loss[ctc_loss=0.07984, att_loss=0.2401, loss=0.2081, over 3292949.58 frames. utt_duration=1273 frames, utt_pad_proportion=0.04421, over 10356.80 utterances.], batch size: 40, lr: 5.93e-03, grad_scale: 8.0 2023-03-08 18:15:41,488 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71616.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:15:54,890 INFO [train2.py:809] (0/4) Epoch 18, batch 3900, loss[ctc_loss=0.06893, att_loss=0.2199, loss=0.1897, over 15741.00 frames. utt_duration=1658 frames, utt_pad_proportion=0.0105, over 38.00 utterances.], tot_loss[ctc_loss=0.07917, att_loss=0.2391, loss=0.2071, over 3291505.79 frames. utt_duration=1278 frames, utt_pad_proportion=0.04357, over 10315.83 utterances.], batch size: 38, lr: 5.93e-03, grad_scale: 8.0 2023-03-08 18:16:10,963 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 2.131e+02 2.535e+02 3.133e+02 8.645e+02, threshold=5.071e+02, percent-clipped=4.0 2023-03-08 18:16:20,889 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5957, 4.3926, 4.5942, 4.5283, 5.0718, 4.5330, 4.5002, 2.3478], device='cuda:0'), covar=tensor([0.0212, 0.0325, 0.0307, 0.0265, 0.0956, 0.0204, 0.0298, 0.1953], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0168, 0.0172, 0.0187, 0.0360, 0.0143, 0.0160, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:16:26,330 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-03-08 18:16:31,465 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71648.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:17:03,222 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.29 vs. limit=5.0 2023-03-08 18:17:13,022 INFO [train2.py:809] (0/4) Epoch 18, batch 3950, loss[ctc_loss=0.09035, att_loss=0.253, loss=0.2205, over 16885.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006406, over 49.00 utterances.], tot_loss[ctc_loss=0.07934, att_loss=0.2395, loss=0.2074, over 3284803.67 frames. utt_duration=1239 frames, utt_pad_proportion=0.05471, over 10620.94 utterances.], batch size: 49, lr: 5.92e-03, grad_scale: 8.0 2023-03-08 18:17:16,274 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71677.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:17:29,979 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=71686.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:17:31,597 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2146, 4.0343, 3.4668, 3.7616, 4.1679, 3.8472, 3.5286, 4.5546], device='cuda:0'), covar=tensor([0.0910, 0.0443, 0.0999, 0.0592, 0.0604, 0.0662, 0.0674, 0.0415], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0214, 0.0223, 0.0194, 0.0269, 0.0234, 0.0198, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 18:17:33,337 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6231, 3.8593, 3.9569, 2.3724, 2.3303, 2.8592, 2.4371, 3.4754], device='cuda:0'), covar=tensor([0.0701, 0.0372, 0.0339, 0.3816, 0.4090, 0.2046, 0.2709, 0.1251], device='cuda:0'), in_proj_covar=tensor([0.0345, 0.0260, 0.0255, 0.0235, 0.0338, 0.0327, 0.0246, 0.0354], device='cuda:0'), out_proj_covar=tensor([1.4693e-04, 9.6565e-05, 1.0897e-04, 1.0166e-04, 1.4208e-04, 1.2833e-04, 9.8449e-05, 1.4525e-04], device='cuda:0') 2023-03-08 18:17:45,662 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71696.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:18:01,143 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7340, 2.3515, 2.5696, 3.2895, 3.0461, 3.1680, 2.5329, 2.2352], device='cuda:0'), covar=tensor([0.0738, 0.1862, 0.1098, 0.0792, 0.0929, 0.0548, 0.1379, 0.1783], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0214, 0.0185, 0.0209, 0.0215, 0.0171, 0.0199, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:18:03,461 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-18.pt 2023-03-08 18:18:26,971 INFO [train2.py:809] (0/4) Epoch 19, batch 0, loss[ctc_loss=0.08306, att_loss=0.2476, loss=0.2147, over 17300.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02374, over 59.00 utterances.], tot_loss[ctc_loss=0.08306, att_loss=0.2476, loss=0.2147, over 17300.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02374, over 59.00 utterances.], batch size: 59, lr: 5.76e-03, grad_scale: 16.0 2023-03-08 18:18:26,973 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 18:18:38,963 INFO [train2.py:843] (0/4) Epoch 19, validation: ctc_loss=0.04291, att_loss=0.2348, loss=0.1964, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 18:18:38,964 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 18:19:20,132 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=71734.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:19:21,552 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.522e+02 2.028e+02 2.615e+02 3.173e+02 7.413e+02, threshold=5.229e+02, percent-clipped=5.0 2023-03-08 18:19:57,600 INFO [train2.py:809] (0/4) Epoch 19, batch 50, loss[ctc_loss=0.07835, att_loss=0.2134, loss=0.1864, over 15376.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01077, over 35.00 utterances.], tot_loss[ctc_loss=0.0813, att_loss=0.2411, loss=0.2091, over 745426.55 frames. utt_duration=1271 frames, utt_pad_proportion=0.03906, over 2347.85 utterances.], batch size: 35, lr: 5.76e-03, grad_scale: 16.0 2023-03-08 18:20:00,371 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2367, 5.2305, 5.1424, 2.6674, 2.0334, 2.8185, 2.8670, 3.9468], device='cuda:0'), covar=tensor([0.0658, 0.0339, 0.0223, 0.4290, 0.5901, 0.2767, 0.2875, 0.1742], device='cuda:0'), in_proj_covar=tensor([0.0347, 0.0261, 0.0257, 0.0236, 0.0340, 0.0330, 0.0248, 0.0358], device='cuda:0'), out_proj_covar=tensor([1.4805e-04, 9.6979e-05, 1.0978e-04, 1.0217e-04, 1.4298e-04, 1.2979e-04, 9.9099e-05, 1.4648e-04], device='cuda:0') 2023-03-08 18:20:46,396 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1042, 6.3257, 5.7530, 6.0126, 5.9305, 5.4062, 5.7355, 5.5163], device='cuda:0'), covar=tensor([0.0962, 0.0652, 0.0732, 0.0729, 0.0924, 0.1399, 0.1904, 0.2200], device='cuda:0'), in_proj_covar=tensor([0.0502, 0.0584, 0.0440, 0.0438, 0.0417, 0.0455, 0.0594, 0.0513], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 18:20:56,932 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.09 vs. limit=5.0 2023-03-08 18:21:17,851 INFO [train2.py:809] (0/4) Epoch 19, batch 100, loss[ctc_loss=0.1121, att_loss=0.2333, loss=0.2091, over 15492.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.009169, over 36.00 utterances.], tot_loss[ctc_loss=0.08001, att_loss=0.2387, loss=0.207, over 1303411.86 frames. utt_duration=1291 frames, utt_pad_proportion=0.04129, over 4044.36 utterances.], batch size: 36, lr: 5.76e-03, grad_scale: 16.0 2023-03-08 18:21:58,052 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2305, 4.5551, 4.4906, 4.7647, 3.0189, 4.6373, 2.8453, 2.0542], device='cuda:0'), covar=tensor([0.0384, 0.0234, 0.0759, 0.0205, 0.1831, 0.0170, 0.1574, 0.1870], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0155, 0.0258, 0.0150, 0.0227, 0.0131, 0.0234, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:22:00,750 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.482e+02 2.106e+02 2.518e+02 3.036e+02 5.583e+02, threshold=5.037e+02, percent-clipped=2.0 2023-03-08 18:22:36,458 INFO [train2.py:809] (0/4) Epoch 19, batch 150, loss[ctc_loss=0.0538, att_loss=0.2388, loss=0.2018, over 16762.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006718, over 48.00 utterances.], tot_loss[ctc_loss=0.08011, att_loss=0.239, loss=0.2072, over 1738753.09 frames. utt_duration=1276 frames, utt_pad_proportion=0.04476, over 5455.65 utterances.], batch size: 48, lr: 5.76e-03, grad_scale: 16.0 2023-03-08 18:23:10,828 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.99 vs. limit=5.0 2023-03-08 18:23:37,778 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4083, 2.4406, 4.9327, 3.9031, 2.9223, 4.0859, 4.7481, 4.5904], device='cuda:0'), covar=tensor([0.0265, 0.1663, 0.0152, 0.0893, 0.1842, 0.0264, 0.0128, 0.0226], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0240, 0.0172, 0.0310, 0.0267, 0.0205, 0.0157, 0.0187], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 18:23:56,060 INFO [train2.py:809] (0/4) Epoch 19, batch 200, loss[ctc_loss=0.08126, att_loss=0.2432, loss=0.2108, over 17251.00 frames. utt_duration=698.6 frames, utt_pad_proportion=0.1246, over 99.00 utterances.], tot_loss[ctc_loss=0.07915, att_loss=0.2387, loss=0.2068, over 2086743.94 frames. utt_duration=1255 frames, utt_pad_proportion=0.04684, over 6658.27 utterances.], batch size: 99, lr: 5.75e-03, grad_scale: 16.0 2023-03-08 18:24:02,184 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4737, 4.4962, 4.2692, 3.0515, 4.3183, 4.1390, 3.9640, 2.7202], device='cuda:0'), covar=tensor([0.0113, 0.0108, 0.0267, 0.0798, 0.0106, 0.0280, 0.0270, 0.1258], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0098, 0.0098, 0.0108, 0.0081, 0.0107, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 18:24:05,815 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.48 vs. limit=5.0 2023-03-08 18:24:14,372 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=71919.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:24:39,024 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.141e+02 1.975e+02 2.308e+02 2.986e+02 5.349e+02, threshold=4.617e+02, percent-clipped=2.0 2023-03-08 18:25:14,792 INFO [train2.py:809] (0/4) Epoch 19, batch 250, loss[ctc_loss=0.06321, att_loss=0.2324, loss=0.1986, over 15948.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.007291, over 41.00 utterances.], tot_loss[ctc_loss=0.07897, att_loss=0.2384, loss=0.2065, over 2355355.89 frames. utt_duration=1261 frames, utt_pad_proportion=0.04377, over 7482.24 utterances.], batch size: 41, lr: 5.75e-03, grad_scale: 16.0 2023-03-08 18:25:37,764 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=71972.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:25:51,051 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=71980.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:26:22,759 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-72000.pt 2023-03-08 18:26:31,186 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-08 18:26:40,273 INFO [train2.py:809] (0/4) Epoch 19, batch 300, loss[ctc_loss=0.06378, att_loss=0.229, loss=0.196, over 16270.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007315, over 43.00 utterances.], tot_loss[ctc_loss=0.07759, att_loss=0.2376, loss=0.2056, over 2558028.74 frames. utt_duration=1275 frames, utt_pad_proportion=0.043, over 8032.89 utterances.], batch size: 43, lr: 5.75e-03, grad_scale: 16.0 2023-03-08 18:27:22,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 2.009e+02 2.404e+02 3.052e+02 6.487e+02, threshold=4.807e+02, percent-clipped=6.0 2023-03-08 18:27:59,502 INFO [train2.py:809] (0/4) Epoch 19, batch 350, loss[ctc_loss=0.09457, att_loss=0.2513, loss=0.22, over 16970.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.006228, over 50.00 utterances.], tot_loss[ctc_loss=0.07723, att_loss=0.2371, loss=0.2051, over 2717111.67 frames. utt_duration=1276 frames, utt_pad_proportion=0.04337, over 8530.69 utterances.], batch size: 50, lr: 5.75e-03, grad_scale: 16.0 2023-03-08 18:29:05,082 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5409, 2.1496, 2.1416, 2.2884, 2.6312, 2.6685, 2.5996, 2.8652], device='cuda:0'), covar=tensor([0.3245, 0.5094, 0.3423, 0.3141, 0.3167, 0.2193, 0.3246, 0.1881], device='cuda:0'), in_proj_covar=tensor([0.0110, 0.0117, 0.0114, 0.0102, 0.0114, 0.0100, 0.0120, 0.0089], device='cuda:0'), out_proj_covar=tensor([8.1433e-05, 8.9542e-05, 8.9041e-05, 7.8818e-05, 8.3845e-05, 7.9193e-05, 8.9544e-05, 7.1851e-05], device='cuda:0') 2023-03-08 18:29:07,987 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1363, 5.4769, 5.3615, 5.3171, 5.4882, 5.4275, 5.2044, 4.8768], device='cuda:0'), covar=tensor([0.0948, 0.0399, 0.0255, 0.0410, 0.0247, 0.0262, 0.0286, 0.0307], device='cuda:0'), in_proj_covar=tensor([0.0507, 0.0339, 0.0322, 0.0336, 0.0400, 0.0413, 0.0343, 0.0378], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 18:29:17,681 INFO [train2.py:809] (0/4) Epoch 19, batch 400, loss[ctc_loss=0.06374, att_loss=0.2358, loss=0.2014, over 16613.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006201, over 47.00 utterances.], tot_loss[ctc_loss=0.07707, att_loss=0.2374, loss=0.2053, over 2844807.70 frames. utt_duration=1284 frames, utt_pad_proportion=0.0417, over 8871.38 utterances.], batch size: 47, lr: 5.75e-03, grad_scale: 8.0 2023-03-08 18:29:18,655 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9615, 3.6407, 3.6604, 3.1826, 3.6845, 3.8206, 3.7402, 2.6915], device='cuda:0'), covar=tensor([0.0867, 0.1228, 0.1678, 0.3958, 0.0985, 0.4359, 0.0759, 0.3981], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0171, 0.0184, 0.0244, 0.0146, 0.0239, 0.0163, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:29:51,757 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72129.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:30:02,016 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.245e+02 2.032e+02 2.343e+02 2.887e+02 1.413e+03, threshold=4.687e+02, percent-clipped=3.0 2023-03-08 18:30:03,847 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6795, 5.0692, 4.9938, 4.9755, 5.0483, 4.8639, 3.5674, 4.8960], device='cuda:0'), covar=tensor([0.0117, 0.0153, 0.0127, 0.0091, 0.0100, 0.0115, 0.0766, 0.0293], device='cuda:0'), in_proj_covar=tensor([0.0088, 0.0085, 0.0105, 0.0065, 0.0071, 0.0082, 0.0101, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:30:37,548 INFO [train2.py:809] (0/4) Epoch 19, batch 450, loss[ctc_loss=0.06492, att_loss=0.2301, loss=0.1971, over 15939.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007881, over 41.00 utterances.], tot_loss[ctc_loss=0.07794, att_loss=0.2382, loss=0.2061, over 2940030.91 frames. utt_duration=1236 frames, utt_pad_proportion=0.05442, over 9525.92 utterances.], batch size: 41, lr: 5.74e-03, grad_scale: 8.0 2023-03-08 18:30:43,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.87 vs. limit=5.0 2023-03-08 18:31:28,115 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72190.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:31:28,693 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-08 18:31:54,723 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9668, 4.9311, 4.7294, 2.9745, 4.7379, 4.6468, 4.2166, 2.7320], device='cuda:0'), covar=tensor([0.0102, 0.0110, 0.0264, 0.0911, 0.0104, 0.0186, 0.0308, 0.1307], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0098, 0.0098, 0.0108, 0.0081, 0.0107, 0.0098, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 18:31:56,578 INFO [train2.py:809] (0/4) Epoch 19, batch 500, loss[ctc_loss=0.07321, att_loss=0.2415, loss=0.2079, over 17160.00 frames. utt_duration=1227 frames, utt_pad_proportion=0.01274, over 56.00 utterances.], tot_loss[ctc_loss=0.07733, att_loss=0.2374, loss=0.2054, over 3015457.95 frames. utt_duration=1269 frames, utt_pad_proportion=0.04679, over 9512.73 utterances.], batch size: 56, lr: 5.74e-03, grad_scale: 8.0 2023-03-08 18:32:08,260 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.20 vs. limit=5.0 2023-03-08 18:32:10,467 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9090, 5.2219, 5.4899, 5.3549, 5.4368, 5.8533, 5.2340, 6.0157], device='cuda:0'), covar=tensor([0.0678, 0.0688, 0.0752, 0.1199, 0.1639, 0.0919, 0.0612, 0.0581], device='cuda:0'), in_proj_covar=tensor([0.0830, 0.0487, 0.0570, 0.0630, 0.0841, 0.0592, 0.0465, 0.0578], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:32:39,381 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 2.024e+02 2.496e+02 3.106e+02 1.224e+03, threshold=4.991e+02, percent-clipped=5.0 2023-03-08 18:33:07,073 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-08 18:33:14,205 INFO [train2.py:809] (0/4) Epoch 19, batch 550, loss[ctc_loss=0.08565, att_loss=0.2209, loss=0.1939, over 14573.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.0376, over 32.00 utterances.], tot_loss[ctc_loss=0.07859, att_loss=0.2375, loss=0.2057, over 3064531.07 frames. utt_duration=1242 frames, utt_pad_proportion=0.05547, over 9885.20 utterances.], batch size: 32, lr: 5.74e-03, grad_scale: 8.0 2023-03-08 18:33:33,906 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0410, 5.1262, 4.8807, 2.1815, 1.8047, 2.7196, 2.5370, 3.7036], device='cuda:0'), covar=tensor([0.0717, 0.0233, 0.0251, 0.5297, 0.6106, 0.2657, 0.3384, 0.1878], device='cuda:0'), in_proj_covar=tensor([0.0348, 0.0260, 0.0260, 0.0236, 0.0341, 0.0332, 0.0249, 0.0357], device='cuda:0'), out_proj_covar=tensor([1.4858e-04, 9.7271e-05, 1.1108e-04, 1.0239e-04, 1.4364e-04, 1.3061e-04, 9.9666e-05, 1.4664e-04], device='cuda:0') 2023-03-08 18:33:35,272 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1079, 2.4794, 3.0691, 4.0801, 3.6816, 3.7084, 2.7088, 2.0771], device='cuda:0'), covar=tensor([0.0783, 0.2155, 0.0980, 0.0696, 0.0840, 0.0502, 0.1552, 0.2349], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0212, 0.0185, 0.0210, 0.0214, 0.0170, 0.0199, 0.0186], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:33:37,393 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72272.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:33:41,755 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72275.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:34:32,964 INFO [train2.py:809] (0/4) Epoch 19, batch 600, loss[ctc_loss=0.07151, att_loss=0.2516, loss=0.2156, over 16634.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004871, over 47.00 utterances.], tot_loss[ctc_loss=0.07855, att_loss=0.2375, loss=0.2057, over 3114582.41 frames. utt_duration=1269 frames, utt_pad_proportion=0.04762, over 9828.18 utterances.], batch size: 47, lr: 5.74e-03, grad_scale: 8.0 2023-03-08 18:34:45,876 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1725, 5.4674, 5.7399, 5.4953, 5.6798, 6.1135, 5.3527, 6.2097], device='cuda:0'), covar=tensor([0.0614, 0.0647, 0.0689, 0.1097, 0.1587, 0.0855, 0.0561, 0.0612], device='cuda:0'), in_proj_covar=tensor([0.0833, 0.0489, 0.0573, 0.0631, 0.0844, 0.0593, 0.0466, 0.0580], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:34:52,092 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72320.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:35:02,570 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72326.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:35:18,135 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.518e+02 1.966e+02 2.354e+02 3.021e+02 6.403e+02, threshold=4.709e+02, percent-clipped=3.0 2023-03-08 18:35:53,224 INFO [train2.py:809] (0/4) Epoch 19, batch 650, loss[ctc_loss=0.07686, att_loss=0.211, loss=0.1842, over 15393.00 frames. utt_duration=1761 frames, utt_pad_proportion=0.009658, over 35.00 utterances.], tot_loss[ctc_loss=0.07824, att_loss=0.2375, loss=0.2057, over 3155943.17 frames. utt_duration=1267 frames, utt_pad_proportion=0.04603, over 9971.49 utterances.], batch size: 35, lr: 5.74e-03, grad_scale: 8.0 2023-03-08 18:36:40,899 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72387.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:37:13,708 INFO [train2.py:809] (0/4) Epoch 19, batch 700, loss[ctc_loss=0.0672, att_loss=0.2307, loss=0.198, over 16426.00 frames. utt_duration=1495 frames, utt_pad_proportion=0.005489, over 44.00 utterances.], tot_loss[ctc_loss=0.07783, att_loss=0.2377, loss=0.2057, over 3188304.85 frames. utt_duration=1284 frames, utt_pad_proportion=0.04124, over 9945.63 utterances.], batch size: 44, lr: 5.73e-03, grad_scale: 8.0 2023-03-08 18:37:57,757 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 2.010e+02 2.298e+02 2.722e+02 6.112e+02, threshold=4.595e+02, percent-clipped=4.0 2023-03-08 18:38:32,590 INFO [train2.py:809] (0/4) Epoch 19, batch 750, loss[ctc_loss=0.08807, att_loss=0.2489, loss=0.2167, over 17277.00 frames. utt_duration=876.2 frames, utt_pad_proportion=0.08156, over 79.00 utterances.], tot_loss[ctc_loss=0.07737, att_loss=0.2369, loss=0.205, over 3206358.14 frames. utt_duration=1283 frames, utt_pad_proportion=0.04215, over 10007.62 utterances.], batch size: 79, lr: 5.73e-03, grad_scale: 8.0 2023-03-08 18:39:06,420 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6618, 2.4587, 2.5510, 2.4214, 2.7324, 2.3021, 2.4156, 2.8652], device='cuda:0'), covar=tensor([0.1591, 0.2878, 0.2304, 0.1719, 0.1285, 0.1540, 0.2391, 0.0871], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0117, 0.0115, 0.0104, 0.0115, 0.0101, 0.0123, 0.0090], device='cuda:0'), out_proj_covar=tensor([8.1942e-05, 8.9460e-05, 8.9808e-05, 8.0153e-05, 8.4330e-05, 7.9976e-05, 9.1039e-05, 7.2159e-05], device='cuda:0') 2023-03-08 18:39:16,270 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72485.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:39:39,280 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5779, 5.0264, 4.7848, 4.8638, 5.0557, 4.7209, 3.5568, 4.9854], device='cuda:0'), covar=tensor([0.0143, 0.0134, 0.0167, 0.0119, 0.0118, 0.0129, 0.0711, 0.0210], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0086, 0.0107, 0.0066, 0.0072, 0.0083, 0.0102, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:39:51,880 INFO [train2.py:809] (0/4) Epoch 19, batch 800, loss[ctc_loss=0.05714, att_loss=0.2133, loss=0.1821, over 15650.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008545, over 37.00 utterances.], tot_loss[ctc_loss=0.07772, att_loss=0.2377, loss=0.2057, over 3227525.00 frames. utt_duration=1275 frames, utt_pad_proportion=0.04412, over 10135.33 utterances.], batch size: 37, lr: 5.73e-03, grad_scale: 8.0 2023-03-08 18:40:37,047 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.504e+02 2.045e+02 2.501e+02 3.023e+02 6.959e+02, threshold=5.001e+02, percent-clipped=5.0 2023-03-08 18:41:12,444 INFO [train2.py:809] (0/4) Epoch 19, batch 850, loss[ctc_loss=0.07862, att_loss=0.2412, loss=0.2087, over 17300.00 frames. utt_duration=877.5 frames, utt_pad_proportion=0.08016, over 79.00 utterances.], tot_loss[ctc_loss=0.07751, att_loss=0.2381, loss=0.206, over 3242886.82 frames. utt_duration=1243 frames, utt_pad_proportion=0.05018, over 10447.43 utterances.], batch size: 79, lr: 5.73e-03, grad_scale: 8.0 2023-03-08 18:41:18,777 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1848, 5.4764, 5.4103, 5.4152, 5.5006, 5.4439, 5.1750, 4.9117], device='cuda:0'), covar=tensor([0.0865, 0.0408, 0.0238, 0.0487, 0.0256, 0.0268, 0.0344, 0.0300], device='cuda:0'), in_proj_covar=tensor([0.0506, 0.0338, 0.0321, 0.0337, 0.0398, 0.0411, 0.0340, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 18:41:39,214 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72575.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:42:31,642 INFO [train2.py:809] (0/4) Epoch 19, batch 900, loss[ctc_loss=0.07605, att_loss=0.252, loss=0.2168, over 17598.00 frames. utt_duration=1007 frames, utt_pad_proportion=0.04711, over 70.00 utterances.], tot_loss[ctc_loss=0.07744, att_loss=0.2378, loss=0.2057, over 3251615.01 frames. utt_duration=1246 frames, utt_pad_proportion=0.04962, over 10454.77 utterances.], batch size: 70, lr: 5.73e-03, grad_scale: 8.0 2023-03-08 18:42:55,509 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72623.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:43:16,190 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 2.045e+02 2.502e+02 3.069e+02 5.173e+02, threshold=5.005e+02, percent-clipped=1.0 2023-03-08 18:43:25,156 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6320, 4.6265, 4.5854, 4.5887, 5.1816, 4.4834, 4.5541, 2.7011], device='cuda:0'), covar=tensor([0.0208, 0.0295, 0.0270, 0.0268, 0.0798, 0.0216, 0.0282, 0.1683], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0169, 0.0171, 0.0187, 0.0357, 0.0143, 0.0160, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:43:51,338 INFO [train2.py:809] (0/4) Epoch 19, batch 950, loss[ctc_loss=0.06386, att_loss=0.248, loss=0.2112, over 16954.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008298, over 50.00 utterances.], tot_loss[ctc_loss=0.07748, att_loss=0.2382, loss=0.2061, over 3259333.56 frames. utt_duration=1252 frames, utt_pad_proportion=0.04977, over 10428.36 utterances.], batch size: 50, lr: 5.72e-03, grad_scale: 8.0 2023-03-08 18:43:51,715 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8235, 3.5081, 3.6249, 3.1008, 3.6381, 3.6792, 3.6054, 2.5712], device='cuda:0'), covar=tensor([0.0913, 0.1912, 0.1816, 0.3908, 0.1544, 0.2014, 0.1108, 0.4446], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0170, 0.0184, 0.0245, 0.0145, 0.0240, 0.0162, 0.0208], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 18:44:30,062 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=72682.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:45:11,226 INFO [train2.py:809] (0/4) Epoch 19, batch 1000, loss[ctc_loss=0.06807, att_loss=0.2295, loss=0.1972, over 16162.00 frames. utt_duration=1578 frames, utt_pad_proportion=0.007332, over 41.00 utterances.], tot_loss[ctc_loss=0.07839, att_loss=0.2388, loss=0.2067, over 3264329.45 frames. utt_duration=1250 frames, utt_pad_proportion=0.04936, over 10456.44 utterances.], batch size: 41, lr: 5.72e-03, grad_scale: 8.0 2023-03-08 18:45:44,332 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1109, 5.0966, 4.7987, 2.8413, 4.9725, 4.6820, 4.1091, 2.8729], device='cuda:0'), covar=tensor([0.0098, 0.0095, 0.0324, 0.1046, 0.0085, 0.0204, 0.0387, 0.1321], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0099, 0.0099, 0.0109, 0.0082, 0.0109, 0.0099, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 18:45:53,517 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6903, 3.6101, 3.0017, 3.2691, 3.7692, 3.4658, 2.7845, 3.9243], device='cuda:0'), covar=tensor([0.1112, 0.0521, 0.1056, 0.0663, 0.0647, 0.0659, 0.0896, 0.0457], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0215, 0.0223, 0.0193, 0.0270, 0.0234, 0.0198, 0.0281], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-08 18:45:56,177 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 2.016e+02 2.281e+02 2.809e+02 6.026e+02, threshold=4.562e+02, percent-clipped=5.0 2023-03-08 18:46:31,107 INFO [train2.py:809] (0/4) Epoch 19, batch 1050, loss[ctc_loss=0.08284, att_loss=0.2572, loss=0.2223, over 17365.00 frames. utt_duration=1179 frames, utt_pad_proportion=0.02003, over 59.00 utterances.], tot_loss[ctc_loss=0.07796, att_loss=0.2385, loss=0.2064, over 3250353.04 frames. utt_duration=1225 frames, utt_pad_proportion=0.05912, over 10627.11 utterances.], batch size: 59, lr: 5.72e-03, grad_scale: 8.0 2023-03-08 18:47:09,280 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6988, 2.1196, 2.4728, 2.1252, 2.8704, 2.4140, 2.4771, 2.9068], device='cuda:0'), covar=tensor([0.2012, 0.3677, 0.2784, 0.3066, 0.1525, 0.1450, 0.2787, 0.1380], device='cuda:0'), in_proj_covar=tensor([0.0109, 0.0117, 0.0114, 0.0103, 0.0113, 0.0100, 0.0122, 0.0090], device='cuda:0'), out_proj_covar=tensor([8.1246e-05, 8.9482e-05, 8.9402e-05, 7.9676e-05, 8.3547e-05, 7.9220e-05, 9.0687e-05, 7.2184e-05], device='cuda:0') 2023-03-08 18:47:10,706 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8700, 5.2099, 5.1308, 5.1196, 5.3394, 4.9239, 4.0457, 5.1938], device='cuda:0'), covar=tensor([0.0106, 0.0119, 0.0111, 0.0084, 0.0080, 0.0119, 0.0542, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0087, 0.0109, 0.0067, 0.0073, 0.0085, 0.0104, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:47:13,807 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72785.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:47:33,008 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8397, 4.0579, 3.8432, 4.2804, 2.8076, 4.0973, 2.4892, 1.9636], device='cuda:0'), covar=tensor([0.0427, 0.0218, 0.0760, 0.0213, 0.1490, 0.0204, 0.1568, 0.1677], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0151, 0.0253, 0.0149, 0.0219, 0.0130, 0.0229, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:47:42,129 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-03-08 18:47:50,612 INFO [train2.py:809] (0/4) Epoch 19, batch 1100, loss[ctc_loss=0.0519, att_loss=0.2038, loss=0.1734, over 15505.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008562, over 36.00 utterances.], tot_loss[ctc_loss=0.07698, att_loss=0.238, loss=0.2058, over 3258243.96 frames. utt_duration=1237 frames, utt_pad_proportion=0.05581, over 10546.26 utterances.], batch size: 36, lr: 5.72e-03, grad_scale: 8.0 2023-03-08 18:48:30,778 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=72833.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:48:35,373 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-08 18:48:35,937 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 1.909e+02 2.229e+02 2.585e+02 5.838e+02, threshold=4.459e+02, percent-clipped=1.0 2023-03-08 18:49:10,571 INFO [train2.py:809] (0/4) Epoch 19, batch 1150, loss[ctc_loss=0.05641, att_loss=0.2359, loss=0.2, over 17023.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.008262, over 51.00 utterances.], tot_loss[ctc_loss=0.07753, att_loss=0.2383, loss=0.2061, over 3259150.85 frames. utt_duration=1229 frames, utt_pad_proportion=0.05937, over 10616.40 utterances.], batch size: 51, lr: 5.72e-03, grad_scale: 8.0 2023-03-08 18:49:59,022 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3740, 4.4330, 4.4849, 4.3787, 4.9837, 4.3584, 4.3599, 2.5597], device='cuda:0'), covar=tensor([0.0260, 0.0310, 0.0277, 0.0312, 0.0678, 0.0242, 0.0311, 0.1778], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0167, 0.0170, 0.0186, 0.0355, 0.0143, 0.0159, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:50:08,370 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72894.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:50:24,637 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8294, 2.6735, 3.3582, 2.5788, 3.2793, 4.0089, 3.8903, 2.9359], device='cuda:0'), covar=tensor([0.0501, 0.1737, 0.1182, 0.1478, 0.1088, 0.1053, 0.0613, 0.1266], device='cuda:0'), in_proj_covar=tensor([0.0241, 0.0241, 0.0273, 0.0215, 0.0259, 0.0355, 0.0252, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:50:30,466 INFO [train2.py:809] (0/4) Epoch 19, batch 1200, loss[ctc_loss=0.05518, att_loss=0.2347, loss=0.1988, over 16964.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007553, over 50.00 utterances.], tot_loss[ctc_loss=0.0772, att_loss=0.238, loss=0.2058, over 3266607.83 frames. utt_duration=1239 frames, utt_pad_proportion=0.0541, over 10559.57 utterances.], batch size: 50, lr: 5.71e-03, grad_scale: 8.0 2023-03-08 18:50:58,314 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5362, 2.3719, 2.2398, 2.2050, 2.9528, 2.2849, 2.2126, 2.7448], device='cuda:0'), covar=tensor([0.1641, 0.2778, 0.2528, 0.2180, 0.1461, 0.1415, 0.2603, 0.1309], device='cuda:0'), in_proj_covar=tensor([0.0108, 0.0114, 0.0113, 0.0102, 0.0112, 0.0098, 0.0122, 0.0089], device='cuda:0'), out_proj_covar=tensor([8.0128e-05, 8.7895e-05, 8.8586e-05, 7.8625e-05, 8.2272e-05, 7.7843e-05, 8.9894e-05, 7.1414e-05], device='cuda:0') 2023-03-08 18:51:02,745 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3766, 5.3145, 5.0959, 3.0824, 5.1835, 4.9278, 4.5972, 3.3151], device='cuda:0'), covar=tensor([0.0094, 0.0086, 0.0266, 0.0877, 0.0073, 0.0167, 0.0265, 0.1064], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0099, 0.0099, 0.0108, 0.0082, 0.0108, 0.0098, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 18:51:15,278 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.411e+02 1.921e+02 2.301e+02 2.976e+02 5.020e+02, threshold=4.601e+02, percent-clipped=3.0 2023-03-08 18:51:18,104 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.74 vs. limit=5.0 2023-03-08 18:51:45,218 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=72955.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:51:50,077 INFO [train2.py:809] (0/4) Epoch 19, batch 1250, loss[ctc_loss=0.07622, att_loss=0.2407, loss=0.2078, over 17058.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.008781, over 53.00 utterances.], tot_loss[ctc_loss=0.07724, att_loss=0.2375, loss=0.2054, over 3266762.25 frames. utt_duration=1269 frames, utt_pad_proportion=0.04741, over 10311.00 utterances.], batch size: 53, lr: 5.71e-03, grad_scale: 8.0 2023-03-08 18:51:59,960 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=72964.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 18:52:15,970 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.74 vs. limit=2.0 2023-03-08 18:52:23,177 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0319, 4.4256, 4.2338, 4.6006, 2.5218, 4.4602, 2.6001, 1.7818], device='cuda:0'), covar=tensor([0.0393, 0.0202, 0.0690, 0.0225, 0.1733, 0.0180, 0.1491, 0.1763], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0153, 0.0255, 0.0150, 0.0222, 0.0131, 0.0229, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:52:27,719 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=72982.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:53:08,947 INFO [train2.py:809] (0/4) Epoch 19, batch 1300, loss[ctc_loss=0.07103, att_loss=0.2292, loss=0.1976, over 16121.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006068, over 42.00 utterances.], tot_loss[ctc_loss=0.07733, att_loss=0.2373, loss=0.2053, over 3265737.39 frames. utt_duration=1263 frames, utt_pad_proportion=0.05108, over 10357.07 utterances.], batch size: 42, lr: 5.71e-03, grad_scale: 8.0 2023-03-08 18:53:37,528 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73025.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 18:53:45,234 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73030.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:53:55,194 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.296e+02 2.017e+02 2.532e+02 3.108e+02 5.795e+02, threshold=5.064e+02, percent-clipped=4.0 2023-03-08 18:54:30,468 INFO [train2.py:809] (0/4) Epoch 19, batch 1350, loss[ctc_loss=0.08129, att_loss=0.2517, loss=0.2176, over 17299.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01229, over 55.00 utterances.], tot_loss[ctc_loss=0.07726, att_loss=0.2373, loss=0.2053, over 3273652.99 frames. utt_duration=1266 frames, utt_pad_proportion=0.04855, over 10355.39 utterances.], batch size: 55, lr: 5.71e-03, grad_scale: 8.0 2023-03-08 18:55:31,170 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9691, 5.2719, 4.8415, 5.2833, 4.6626, 5.0343, 5.4119, 5.1908], device='cuda:0'), covar=tensor([0.0509, 0.0264, 0.0797, 0.0319, 0.0417, 0.0239, 0.0202, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0376, 0.0306, 0.0356, 0.0328, 0.0309, 0.0231, 0.0291, 0.0275], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 18:55:50,786 INFO [train2.py:809] (0/4) Epoch 19, batch 1400, loss[ctc_loss=0.08649, att_loss=0.227, loss=0.1989, over 11974.00 frames. utt_duration=1843 frames, utt_pad_proportion=0.155, over 26.00 utterances.], tot_loss[ctc_loss=0.07726, att_loss=0.2372, loss=0.2052, over 3268345.56 frames. utt_duration=1258 frames, utt_pad_proportion=0.05221, over 10406.59 utterances.], batch size: 26, lr: 5.71e-03, grad_scale: 8.0 2023-03-08 18:55:57,098 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4424, 2.8210, 4.8925, 3.8207, 2.8655, 4.1788, 4.5950, 4.4970], device='cuda:0'), covar=tensor([0.0226, 0.1500, 0.0145, 0.0992, 0.1931, 0.0258, 0.0157, 0.0262], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0241, 0.0175, 0.0309, 0.0266, 0.0205, 0.0160, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 18:56:25,721 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8496, 3.7364, 3.1262, 3.3220, 3.8817, 3.4840, 2.9491, 4.0174], device='cuda:0'), covar=tensor([0.1070, 0.0493, 0.1073, 0.0730, 0.0626, 0.0747, 0.0876, 0.0474], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0217, 0.0225, 0.0195, 0.0273, 0.0237, 0.0200, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 18:56:35,399 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.220e+02 2.120e+02 2.397e+02 2.926e+02 6.281e+02, threshold=4.794e+02, percent-clipped=3.0 2023-03-08 18:57:10,509 INFO [train2.py:809] (0/4) Epoch 19, batch 1450, loss[ctc_loss=0.08055, att_loss=0.244, loss=0.2113, over 17329.00 frames. utt_duration=878.9 frames, utt_pad_proportion=0.08069, over 79.00 utterances.], tot_loss[ctc_loss=0.07735, att_loss=0.2375, loss=0.2055, over 3266462.56 frames. utt_duration=1237 frames, utt_pad_proportion=0.05801, over 10575.11 utterances.], batch size: 79, lr: 5.70e-03, grad_scale: 8.0 2023-03-08 18:57:40,677 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2271, 2.9673, 3.3059, 4.3555, 3.8632, 3.8414, 2.8446, 2.1508], device='cuda:0'), covar=tensor([0.0738, 0.1850, 0.0926, 0.0606, 0.0930, 0.0529, 0.1583, 0.2309], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0209, 0.0182, 0.0208, 0.0212, 0.0169, 0.0196, 0.0183], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:58:12,532 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3819, 5.6512, 5.1167, 5.4297, 5.2876, 4.8253, 5.1437, 4.8909], device='cuda:0'), covar=tensor([0.1208, 0.0884, 0.0907, 0.0890, 0.0933, 0.1412, 0.2019, 0.2180], device='cuda:0'), in_proj_covar=tensor([0.0506, 0.0585, 0.0436, 0.0443, 0.0413, 0.0449, 0.0593, 0.0512], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 18:58:30,640 INFO [train2.py:809] (0/4) Epoch 19, batch 1500, loss[ctc_loss=0.09595, att_loss=0.2392, loss=0.2105, over 16123.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.005852, over 42.00 utterances.], tot_loss[ctc_loss=0.07774, att_loss=0.2383, loss=0.2062, over 3275875.62 frames. utt_duration=1234 frames, utt_pad_proportion=0.0555, over 10628.45 utterances.], batch size: 42, lr: 5.70e-03, grad_scale: 8.0 2023-03-08 18:58:46,514 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7074, 5.1230, 4.9226, 5.0067, 5.2220, 4.7596, 3.5634, 5.0977], device='cuda:0'), covar=tensor([0.0115, 0.0123, 0.0127, 0.0081, 0.0067, 0.0110, 0.0682, 0.0173], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0087, 0.0109, 0.0066, 0.0072, 0.0085, 0.0104, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 18:59:15,228 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 2.048e+02 2.505e+02 3.116e+02 1.439e+03, threshold=5.009e+02, percent-clipped=5.0 2023-03-08 18:59:19,046 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.85 vs. limit=2.0 2023-03-08 18:59:31,021 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2609, 4.2777, 4.3162, 4.2940, 4.7585, 4.2226, 4.2660, 2.3597], device='cuda:0'), covar=tensor([0.0285, 0.0334, 0.0300, 0.0278, 0.0947, 0.0275, 0.0302, 0.2083], device='cuda:0'), in_proj_covar=tensor([0.0144, 0.0168, 0.0170, 0.0184, 0.0354, 0.0143, 0.0157, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 18:59:36,939 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73250.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 18:59:49,904 INFO [train2.py:809] (0/4) Epoch 19, batch 1550, loss[ctc_loss=0.06705, att_loss=0.2294, loss=0.1969, over 16532.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006246, over 45.00 utterances.], tot_loss[ctc_loss=0.07836, att_loss=0.2389, loss=0.2068, over 3277772.54 frames. utt_duration=1253 frames, utt_pad_proportion=0.05122, over 10476.74 utterances.], batch size: 45, lr: 5.70e-03, grad_scale: 8.0 2023-03-08 19:00:30,584 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4470, 4.9368, 4.7653, 4.8836, 5.0062, 4.5735, 3.4089, 4.8203], device='cuda:0'), covar=tensor([0.0131, 0.0126, 0.0129, 0.0088, 0.0090, 0.0116, 0.0725, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0087, 0.0108, 0.0066, 0.0072, 0.0084, 0.0103, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 19:00:45,863 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73293.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:01:09,921 INFO [train2.py:809] (0/4) Epoch 19, batch 1600, loss[ctc_loss=0.08743, att_loss=0.2176, loss=0.1916, over 15644.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.00836, over 37.00 utterances.], tot_loss[ctc_loss=0.0784, att_loss=0.2383, loss=0.2063, over 3269604.74 frames. utt_duration=1247 frames, utt_pad_proportion=0.05439, over 10501.16 utterances.], batch size: 37, lr: 5.70e-03, grad_scale: 8.0 2023-03-08 19:01:28,651 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73320.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 19:01:54,272 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 2.054e+02 2.356e+02 2.874e+02 5.257e+02, threshold=4.713e+02, percent-clipped=1.0 2023-03-08 19:02:23,058 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73354.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:02:29,411 INFO [train2.py:809] (0/4) Epoch 19, batch 1650, loss[ctc_loss=0.06859, att_loss=0.213, loss=0.1841, over 15893.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008864, over 39.00 utterances.], tot_loss[ctc_loss=0.07822, att_loss=0.238, loss=0.206, over 3258821.52 frames. utt_duration=1221 frames, utt_pad_proportion=0.06506, over 10685.29 utterances.], batch size: 39, lr: 5.70e-03, grad_scale: 8.0 2023-03-08 19:03:02,446 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4996, 2.9564, 3.6601, 2.9043, 3.5704, 4.6921, 4.5029, 3.2796], device='cuda:0'), covar=tensor([0.0429, 0.1791, 0.1230, 0.1488, 0.1023, 0.0733, 0.0504, 0.1372], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0242, 0.0273, 0.0215, 0.0259, 0.0354, 0.0252, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 19:03:48,204 INFO [train2.py:809] (0/4) Epoch 19, batch 1700, loss[ctc_loss=0.07515, att_loss=0.2368, loss=0.2045, over 16329.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006171, over 45.00 utterances.], tot_loss[ctc_loss=0.07818, att_loss=0.2379, loss=0.206, over 3261745.69 frames. utt_duration=1208 frames, utt_pad_proportion=0.06761, over 10814.85 utterances.], batch size: 45, lr: 5.69e-03, grad_scale: 8.0 2023-03-08 19:04:02,493 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 19:04:32,148 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 2.058e+02 2.463e+02 2.943e+02 5.957e+02, threshold=4.927e+02, percent-clipped=3.0 2023-03-08 19:05:06,957 INFO [train2.py:809] (0/4) Epoch 19, batch 1750, loss[ctc_loss=0.1408, att_loss=0.2748, loss=0.248, over 14291.00 frames. utt_duration=390.6 frames, utt_pad_proportion=0.3135, over 147.00 utterances.], tot_loss[ctc_loss=0.0786, att_loss=0.2383, loss=0.2064, over 3261938.11 frames. utt_duration=1183 frames, utt_pad_proportion=0.07331, over 11044.61 utterances.], batch size: 147, lr: 5.69e-03, grad_scale: 8.0 2023-03-08 19:06:25,915 INFO [train2.py:809] (0/4) Epoch 19, batch 1800, loss[ctc_loss=0.08781, att_loss=0.2493, loss=0.217, over 17017.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008786, over 51.00 utterances.], tot_loss[ctc_loss=0.07839, att_loss=0.2385, loss=0.2065, over 3269296.70 frames. utt_duration=1205 frames, utt_pad_proportion=0.06653, over 10862.98 utterances.], batch size: 51, lr: 5.69e-03, grad_scale: 8.0 2023-03-08 19:07:10,632 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.143e+02 1.911e+02 2.242e+02 2.883e+02 9.834e+02, threshold=4.485e+02, percent-clipped=3.0 2023-03-08 19:07:23,149 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7480, 2.2941, 2.5768, 2.4265, 2.8506, 2.5973, 2.7186, 3.0093], device='cuda:0'), covar=tensor([0.2017, 0.3274, 0.2596, 0.2324, 0.1720, 0.1312, 0.2259, 0.1338], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0113, 0.0112, 0.0101, 0.0110, 0.0097, 0.0120, 0.0089], device='cuda:0'), out_proj_covar=tensor([7.9600e-05, 8.6683e-05, 8.7562e-05, 7.8029e-05, 8.1313e-05, 7.7361e-05, 8.8642e-05, 7.1143e-05], device='cuda:0') 2023-03-08 19:07:33,044 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73550.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:07:45,874 INFO [train2.py:809] (0/4) Epoch 19, batch 1850, loss[ctc_loss=0.04376, att_loss=0.194, loss=0.164, over 15373.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.0108, over 35.00 utterances.], tot_loss[ctc_loss=0.07791, att_loss=0.2377, loss=0.2058, over 3272786.11 frames. utt_duration=1240 frames, utt_pad_proportion=0.05702, over 10570.41 utterances.], batch size: 35, lr: 5.69e-03, grad_scale: 8.0 2023-03-08 19:07:50,932 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5067, 2.8039, 4.9514, 3.9272, 3.0093, 4.2726, 4.8221, 4.5586], device='cuda:0'), covar=tensor([0.0250, 0.1449, 0.0221, 0.1029, 0.1743, 0.0268, 0.0145, 0.0284], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0240, 0.0174, 0.0307, 0.0264, 0.0204, 0.0160, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 19:08:19,900 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7543, 3.1754, 3.8672, 3.0876, 3.8211, 4.8692, 4.6362, 3.5301], device='cuda:0'), covar=tensor([0.0347, 0.1541, 0.1052, 0.1350, 0.0879, 0.0672, 0.0542, 0.1190], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0242, 0.0275, 0.0216, 0.0260, 0.0357, 0.0253, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 19:08:36,593 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73590.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:08:49,385 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73598.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:09:04,976 INFO [train2.py:809] (0/4) Epoch 19, batch 1900, loss[ctc_loss=0.08089, att_loss=0.2478, loss=0.2144, over 17394.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.04714, over 69.00 utterances.], tot_loss[ctc_loss=0.07795, att_loss=0.238, loss=0.206, over 3275663.22 frames. utt_duration=1244 frames, utt_pad_proportion=0.05554, over 10547.48 utterances.], batch size: 69, lr: 5.69e-03, grad_scale: 8.0 2023-03-08 19:09:23,752 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73620.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 19:09:38,945 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.19 vs. limit=5.0 2023-03-08 19:09:48,771 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 1.956e+02 2.391e+02 2.954e+02 5.478e+02, threshold=4.781e+02, percent-clipped=4.0 2023-03-08 19:10:10,620 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73649.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:10:11,667 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-03-08 19:10:13,962 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73651.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:10:20,776 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 19:10:24,368 INFO [train2.py:809] (0/4) Epoch 19, batch 1950, loss[ctc_loss=0.06337, att_loss=0.2281, loss=0.1951, over 15949.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007384, over 41.00 utterances.], tot_loss[ctc_loss=0.07703, att_loss=0.2369, loss=0.2049, over 3264841.97 frames. utt_duration=1270 frames, utt_pad_proportion=0.05057, over 10298.45 utterances.], batch size: 41, lr: 5.68e-03, grad_scale: 8.0 2023-03-08 19:10:40,448 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73668.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 19:11:43,429 INFO [train2.py:809] (0/4) Epoch 19, batch 2000, loss[ctc_loss=0.09475, att_loss=0.2592, loss=0.2263, over 16758.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.00569, over 48.00 utterances.], tot_loss[ctc_loss=0.07794, att_loss=0.2373, loss=0.2054, over 3269740.94 frames. utt_duration=1259 frames, utt_pad_proportion=0.05201, over 10403.37 utterances.], batch size: 48, lr: 5.68e-03, grad_scale: 8.0 2023-03-08 19:12:22,458 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=73732.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:12:28,299 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 2.150e+02 2.565e+02 3.312e+02 1.867e+03, threshold=5.130e+02, percent-clipped=8.0 2023-03-08 19:13:03,765 INFO [train2.py:809] (0/4) Epoch 19, batch 2050, loss[ctc_loss=0.07396, att_loss=0.2215, loss=0.192, over 15874.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009939, over 39.00 utterances.], tot_loss[ctc_loss=0.07756, att_loss=0.2376, loss=0.2056, over 3275031.42 frames. utt_duration=1272 frames, utt_pad_proportion=0.04667, over 10314.18 utterances.], batch size: 39, lr: 5.68e-03, grad_scale: 8.0 2023-03-08 19:13:07,918 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5342, 2.1136, 2.2234, 2.1860, 2.8165, 2.3906, 2.4661, 2.7283], device='cuda:0'), covar=tensor([0.1536, 0.3184, 0.2513, 0.2120, 0.1764, 0.1284, 0.2242, 0.1161], device='cuda:0'), in_proj_covar=tensor([0.0107, 0.0112, 0.0111, 0.0101, 0.0110, 0.0097, 0.0120, 0.0088], device='cuda:0'), out_proj_covar=tensor([7.9542e-05, 8.6675e-05, 8.7506e-05, 7.8114e-05, 8.1345e-05, 7.6899e-05, 8.8601e-05, 7.0925e-05], device='cuda:0') 2023-03-08 19:13:58,643 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.77 vs. limit=2.0 2023-03-08 19:13:59,687 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=73793.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:14:24,522 INFO [train2.py:809] (0/4) Epoch 19, batch 2100, loss[ctc_loss=0.05766, att_loss=0.2255, loss=0.192, over 16285.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006899, over 43.00 utterances.], tot_loss[ctc_loss=0.07756, att_loss=0.237, loss=0.2051, over 3273373.52 frames. utt_duration=1278 frames, utt_pad_proportion=0.04613, over 10253.37 utterances.], batch size: 43, lr: 5.68e-03, grad_scale: 8.0 2023-03-08 19:15:08,599 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 2.020e+02 2.356e+02 2.940e+02 5.527e+02, threshold=4.712e+02, percent-clipped=4.0 2023-03-08 19:15:43,844 INFO [train2.py:809] (0/4) Epoch 19, batch 2150, loss[ctc_loss=0.09898, att_loss=0.2521, loss=0.2214, over 17310.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01147, over 55.00 utterances.], tot_loss[ctc_loss=0.07847, att_loss=0.2386, loss=0.2065, over 3277712.57 frames. utt_duration=1229 frames, utt_pad_proportion=0.05802, over 10684.97 utterances.], batch size: 55, lr: 5.68e-03, grad_scale: 8.0 2023-03-08 19:17:03,551 INFO [train2.py:809] (0/4) Epoch 19, batch 2200, loss[ctc_loss=0.09484, att_loss=0.2493, loss=0.2184, over 17063.00 frames. utt_duration=865.6 frames, utt_pad_proportion=0.09169, over 79.00 utterances.], tot_loss[ctc_loss=0.07817, att_loss=0.2386, loss=0.2066, over 3287450.99 frames. utt_duration=1220 frames, utt_pad_proportion=0.05691, over 10791.19 utterances.], batch size: 79, lr: 5.68e-03, grad_scale: 8.0 2023-03-08 19:17:47,492 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.450e+02 2.057e+02 2.535e+02 3.122e+02 7.194e+02, threshold=5.070e+02, percent-clipped=8.0 2023-03-08 19:18:04,608 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=73946.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:18:09,408 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=73949.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:18:23,579 INFO [train2.py:809] (0/4) Epoch 19, batch 2250, loss[ctc_loss=0.09464, att_loss=0.2397, loss=0.2107, over 16290.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006488, over 43.00 utterances.], tot_loss[ctc_loss=0.07831, att_loss=0.2391, loss=0.207, over 3291728.70 frames. utt_duration=1217 frames, utt_pad_proportion=0.05674, over 10835.31 utterances.], batch size: 43, lr: 5.67e-03, grad_scale: 8.0 2023-03-08 19:19:25,457 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=73997.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:19:30,241 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-74000.pt 2023-03-08 19:19:47,856 INFO [train2.py:809] (0/4) Epoch 19, batch 2300, loss[ctc_loss=0.06469, att_loss=0.2359, loss=0.2017, over 16678.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.007152, over 46.00 utterances.], tot_loss[ctc_loss=0.07892, att_loss=0.2402, loss=0.208, over 3296864.44 frames. utt_duration=1209 frames, utt_pad_proportion=0.05748, over 10921.72 utterances.], batch size: 46, lr: 5.67e-03, grad_scale: 8.0 2023-03-08 19:20:09,104 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.35 vs. limit=5.0 2023-03-08 19:20:30,626 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 2.198e+02 2.569e+02 3.135e+02 6.426e+02, threshold=5.138e+02, percent-clipped=5.0 2023-03-08 19:21:05,898 INFO [train2.py:809] (0/4) Epoch 19, batch 2350, loss[ctc_loss=0.08598, att_loss=0.2473, loss=0.215, over 17321.00 frames. utt_duration=878.3 frames, utt_pad_proportion=0.08027, over 79.00 utterances.], tot_loss[ctc_loss=0.0781, att_loss=0.2387, loss=0.2066, over 3288706.02 frames. utt_duration=1223 frames, utt_pad_proportion=0.05692, over 10770.38 utterances.], batch size: 79, lr: 5.67e-03, grad_scale: 8.0 2023-03-08 19:21:15,269 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74064.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:21:52,215 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74088.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:22:07,273 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0383, 4.3304, 4.2074, 4.5519, 2.7983, 4.4449, 2.5030, 1.6493], device='cuda:0'), covar=tensor([0.0356, 0.0242, 0.0724, 0.0222, 0.1506, 0.0177, 0.1496, 0.1649], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0153, 0.0257, 0.0149, 0.0220, 0.0132, 0.0230, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:22:24,337 INFO [train2.py:809] (0/4) Epoch 19, batch 2400, loss[ctc_loss=0.09309, att_loss=0.258, loss=0.2251, over 17178.00 frames. utt_duration=688.6 frames, utt_pad_proportion=0.1294, over 100.00 utterances.], tot_loss[ctc_loss=0.07809, att_loss=0.2385, loss=0.2064, over 3281112.91 frames. utt_duration=1224 frames, utt_pad_proportion=0.05843, over 10739.55 utterances.], batch size: 100, lr: 5.67e-03, grad_scale: 16.0 2023-03-08 19:22:51,364 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74125.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:23:07,413 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 2.133e+02 2.520e+02 3.216e+02 6.079e+02, threshold=5.039e+02, percent-clipped=1.0 2023-03-08 19:23:43,615 INFO [train2.py:809] (0/4) Epoch 19, batch 2450, loss[ctc_loss=0.06063, att_loss=0.2101, loss=0.1802, over 15344.00 frames. utt_duration=1755 frames, utt_pad_proportion=0.01187, over 35.00 utterances.], tot_loss[ctc_loss=0.07877, att_loss=0.2391, loss=0.207, over 3281775.52 frames. utt_duration=1199 frames, utt_pad_proportion=0.06455, over 10965.48 utterances.], batch size: 35, lr: 5.67e-03, grad_scale: 16.0 2023-03-08 19:23:47,248 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 19:23:48,353 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1282, 5.0975, 4.8280, 2.7050, 4.8942, 4.6724, 4.4648, 2.7050], device='cuda:0'), covar=tensor([0.0098, 0.0097, 0.0313, 0.1145, 0.0100, 0.0203, 0.0286, 0.1449], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0101, 0.0101, 0.0111, 0.0084, 0.0112, 0.0099, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 19:24:38,799 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1802, 5.4838, 5.3884, 5.4444, 5.4646, 5.4432, 5.1653, 4.9446], device='cuda:0'), covar=tensor([0.0985, 0.0509, 0.0249, 0.0401, 0.0279, 0.0296, 0.0377, 0.0324], device='cuda:0'), in_proj_covar=tensor([0.0515, 0.0348, 0.0331, 0.0344, 0.0406, 0.0418, 0.0348, 0.0384], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 19:24:45,715 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.79 vs. limit=2.0 2023-03-08 19:25:02,842 INFO [train2.py:809] (0/4) Epoch 19, batch 2500, loss[ctc_loss=0.0582, att_loss=0.2247, loss=0.1914, over 16002.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007667, over 40.00 utterances.], tot_loss[ctc_loss=0.07805, att_loss=0.2387, loss=0.2066, over 3276752.12 frames. utt_duration=1225 frames, utt_pad_proportion=0.05982, over 10716.62 utterances.], batch size: 40, lr: 5.66e-03, grad_scale: 16.0 2023-03-08 19:25:47,410 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.096e+02 2.486e+02 2.922e+02 5.921e+02, threshold=4.971e+02, percent-clipped=2.0 2023-03-08 19:25:54,809 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74240.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:26:03,959 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74246.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:26:23,586 INFO [train2.py:809] (0/4) Epoch 19, batch 2550, loss[ctc_loss=0.06384, att_loss=0.2337, loss=0.1997, over 16272.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007676, over 43.00 utterances.], tot_loss[ctc_loss=0.07754, att_loss=0.2384, loss=0.2063, over 3269342.71 frames. utt_duration=1221 frames, utt_pad_proportion=0.06138, over 10719.64 utterances.], batch size: 43, lr: 5.66e-03, grad_scale: 16.0 2023-03-08 19:27:20,421 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74294.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:27:30,890 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6193, 4.4023, 4.5320, 4.5184, 5.0880, 4.4961, 4.4536, 2.4508], device='cuda:0'), covar=tensor([0.0196, 0.0362, 0.0303, 0.0256, 0.0730, 0.0218, 0.0328, 0.1944], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0174, 0.0177, 0.0192, 0.0363, 0.0149, 0.0165, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:27:32,364 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74301.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:27:43,386 INFO [train2.py:809] (0/4) Epoch 19, batch 2600, loss[ctc_loss=0.06995, att_loss=0.2321, loss=0.1997, over 16550.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005732, over 45.00 utterances.], tot_loss[ctc_loss=0.07689, att_loss=0.2373, loss=0.2052, over 3265403.35 frames. utt_duration=1232 frames, utt_pad_proportion=0.06166, over 10614.69 utterances.], batch size: 45, lr: 5.66e-03, grad_scale: 8.0 2023-03-08 19:28:22,722 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3767, 2.7847, 3.2657, 4.4423, 3.9056, 3.8824, 2.9499, 2.2542], device='cuda:0'), covar=tensor([0.0741, 0.2169, 0.1003, 0.0620, 0.0836, 0.0545, 0.1502, 0.2407], device='cuda:0'), in_proj_covar=tensor([0.0177, 0.0212, 0.0186, 0.0212, 0.0217, 0.0175, 0.0198, 0.0185], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:28:29,339 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.314e+02 1.890e+02 2.336e+02 2.791e+02 4.929e+02, threshold=4.673e+02, percent-clipped=0.0 2023-03-08 19:29:04,102 INFO [train2.py:809] (0/4) Epoch 19, batch 2650, loss[ctc_loss=0.08073, att_loss=0.2514, loss=0.2172, over 17015.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.007992, over 51.00 utterances.], tot_loss[ctc_loss=0.07669, att_loss=0.2375, loss=0.2053, over 3269840.85 frames. utt_duration=1253 frames, utt_pad_proportion=0.05406, over 10447.14 utterances.], batch size: 51, lr: 5.66e-03, grad_scale: 8.0 2023-03-08 19:29:07,309 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7571, 6.0008, 5.4688, 5.7351, 5.6618, 5.2136, 5.4248, 5.1731], device='cuda:0'), covar=tensor([0.1244, 0.0882, 0.0905, 0.0833, 0.0941, 0.1413, 0.2328, 0.2393], device='cuda:0'), in_proj_covar=tensor([0.0508, 0.0591, 0.0447, 0.0441, 0.0418, 0.0455, 0.0604, 0.0514], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 19:29:11,060 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.45 vs. limit=5.0 2023-03-08 19:29:52,057 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74388.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:30:23,754 INFO [train2.py:809] (0/4) Epoch 19, batch 2700, loss[ctc_loss=0.09279, att_loss=0.2385, loss=0.2094, over 15931.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007222, over 41.00 utterances.], tot_loss[ctc_loss=0.07719, att_loss=0.2376, loss=0.2055, over 3270649.76 frames. utt_duration=1256 frames, utt_pad_proportion=0.05391, over 10432.53 utterances.], batch size: 41, lr: 5.66e-03, grad_scale: 8.0 2023-03-08 19:30:42,774 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74420.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:31:00,786 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4453, 2.5540, 4.8983, 3.7845, 2.9618, 4.1644, 4.7179, 4.6152], device='cuda:0'), covar=tensor([0.0284, 0.1642, 0.0209, 0.1146, 0.1961, 0.0273, 0.0157, 0.0250], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0240, 0.0176, 0.0308, 0.0263, 0.0204, 0.0161, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 19:31:08,858 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74436.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:31:10,817 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 2.123e+02 2.465e+02 2.927e+02 6.214e+02, threshold=4.930e+02, percent-clipped=3.0 2023-03-08 19:31:24,612 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.17 vs. limit=2.0 2023-03-08 19:31:46,095 INFO [train2.py:809] (0/4) Epoch 19, batch 2750, loss[ctc_loss=0.06761, att_loss=0.2239, loss=0.1926, over 15886.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009066, over 39.00 utterances.], tot_loss[ctc_loss=0.07821, att_loss=0.239, loss=0.2069, over 3278423.91 frames. utt_duration=1225 frames, utt_pad_proportion=0.0585, over 10717.30 utterances.], batch size: 39, lr: 5.65e-03, grad_scale: 8.0 2023-03-08 19:33:01,709 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2023-03-08 19:33:07,058 INFO [train2.py:809] (0/4) Epoch 19, batch 2800, loss[ctc_loss=0.06474, att_loss=0.2436, loss=0.2079, over 16878.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007737, over 49.00 utterances.], tot_loss[ctc_loss=0.07792, att_loss=0.2382, loss=0.2062, over 3266774.05 frames. utt_duration=1235 frames, utt_pad_proportion=0.059, over 10595.44 utterances.], batch size: 49, lr: 5.65e-03, grad_scale: 8.0 2023-03-08 19:33:21,990 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-08 19:33:53,760 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 2.053e+02 2.431e+02 2.775e+02 7.062e+02, threshold=4.862e+02, percent-clipped=1.0 2023-03-08 19:34:27,616 INFO [train2.py:809] (0/4) Epoch 19, batch 2850, loss[ctc_loss=0.06673, att_loss=0.2168, loss=0.1868, over 15482.00 frames. utt_duration=1722 frames, utt_pad_proportion=0.009478, over 36.00 utterances.], tot_loss[ctc_loss=0.07839, att_loss=0.2378, loss=0.2059, over 3254341.74 frames. utt_duration=1212 frames, utt_pad_proportion=0.06715, over 10751.64 utterances.], batch size: 36, lr: 5.65e-03, grad_scale: 8.0 2023-03-08 19:34:37,211 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0631, 5.3545, 5.2352, 5.2787, 5.3652, 5.3268, 5.0084, 4.8267], device='cuda:0'), covar=tensor([0.1118, 0.0501, 0.0308, 0.0528, 0.0278, 0.0354, 0.0420, 0.0311], device='cuda:0'), in_proj_covar=tensor([0.0514, 0.0349, 0.0331, 0.0344, 0.0403, 0.0417, 0.0346, 0.0384], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 19:35:17,763 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9135, 3.5866, 3.5834, 3.1780, 3.5588, 3.7291, 3.6421, 2.7549], device='cuda:0'), covar=tensor([0.1067, 0.1924, 0.2566, 0.3805, 0.2379, 0.3146, 0.1211, 0.4221], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0173, 0.0187, 0.0246, 0.0149, 0.0247, 0.0166, 0.0209], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:35:26,126 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1911, 5.5061, 5.3552, 5.4184, 5.4885, 5.4646, 5.1600, 4.9466], device='cuda:0'), covar=tensor([0.0982, 0.0506, 0.0273, 0.0457, 0.0286, 0.0268, 0.0362, 0.0278], device='cuda:0'), in_proj_covar=tensor([0.0510, 0.0347, 0.0329, 0.0341, 0.0400, 0.0414, 0.0344, 0.0382], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 19:35:27,782 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74595.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:35:29,185 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74596.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:35:48,357 INFO [train2.py:809] (0/4) Epoch 19, batch 2900, loss[ctc_loss=0.0785, att_loss=0.2457, loss=0.2122, over 17144.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01355, over 56.00 utterances.], tot_loss[ctc_loss=0.07767, att_loss=0.2375, loss=0.2055, over 3261129.18 frames. utt_duration=1231 frames, utt_pad_proportion=0.06143, over 10608.78 utterances.], batch size: 56, lr: 5.65e-03, grad_scale: 8.0 2023-03-08 19:36:15,693 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-03-08 19:36:34,196 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 1.992e+02 2.354e+02 2.922e+02 1.423e+03, threshold=4.708e+02, percent-clipped=1.0 2023-03-08 19:37:06,215 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74656.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:37:08,928 INFO [train2.py:809] (0/4) Epoch 19, batch 2950, loss[ctc_loss=0.07821, att_loss=0.2476, loss=0.2137, over 16182.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.00673, over 41.00 utterances.], tot_loss[ctc_loss=0.07795, att_loss=0.2381, loss=0.2061, over 3265385.49 frames. utt_duration=1214 frames, utt_pad_proportion=0.06505, over 10776.41 utterances.], batch size: 41, lr: 5.65e-03, grad_scale: 8.0 2023-03-08 19:37:29,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8896, 6.1151, 5.6324, 5.8451, 5.8099, 5.3024, 5.5905, 5.3910], device='cuda:0'), covar=tensor([0.1198, 0.0886, 0.0888, 0.0792, 0.1020, 0.1431, 0.2190, 0.2265], device='cuda:0'), in_proj_covar=tensor([0.0516, 0.0600, 0.0452, 0.0447, 0.0421, 0.0460, 0.0608, 0.0521], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 19:38:28,658 INFO [train2.py:809] (0/4) Epoch 19, batch 3000, loss[ctc_loss=0.06149, att_loss=0.2168, loss=0.1857, over 15993.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.0087, over 40.00 utterances.], tot_loss[ctc_loss=0.07802, att_loss=0.2375, loss=0.2056, over 3260490.14 frames. utt_duration=1228 frames, utt_pad_proportion=0.06103, over 10632.91 utterances.], batch size: 40, lr: 5.64e-03, grad_scale: 8.0 2023-03-08 19:38:28,661 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 19:38:42,983 INFO [train2.py:843] (0/4) Epoch 19, validation: ctc_loss=0.04253, att_loss=0.235, loss=0.1965, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 19:38:42,984 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 19:39:02,192 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74720.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:39:29,765 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 2.054e+02 2.397e+02 2.986e+02 9.234e+02, threshold=4.795e+02, percent-clipped=4.0 2023-03-08 19:40:05,292 INFO [train2.py:809] (0/4) Epoch 19, batch 3050, loss[ctc_loss=0.06765, att_loss=0.2221, loss=0.1912, over 15958.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006284, over 41.00 utterances.], tot_loss[ctc_loss=0.07849, att_loss=0.2378, loss=0.2059, over 3268379.27 frames. utt_duration=1224 frames, utt_pad_proportion=0.05951, over 10695.57 utterances.], batch size: 41, lr: 5.64e-03, grad_scale: 8.0 2023-03-08 19:40:20,695 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74768.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:41:26,170 INFO [train2.py:809] (0/4) Epoch 19, batch 3100, loss[ctc_loss=0.09187, att_loss=0.2574, loss=0.2243, over 17262.00 frames. utt_duration=1098 frames, utt_pad_proportion=0.0406, over 63.00 utterances.], tot_loss[ctc_loss=0.07855, att_loss=0.2376, loss=0.2058, over 3265991.80 frames. utt_duration=1234 frames, utt_pad_proportion=0.05877, over 10597.23 utterances.], batch size: 63, lr: 5.64e-03, grad_scale: 8.0 2023-03-08 19:42:11,533 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 1.971e+02 2.423e+02 3.077e+02 6.495e+02, threshold=4.846e+02, percent-clipped=7.0 2023-03-08 19:42:22,666 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0969, 5.1570, 4.9885, 2.4117, 2.0467, 2.8959, 2.3727, 3.9837], device='cuda:0'), covar=tensor([0.0692, 0.0272, 0.0267, 0.4708, 0.5697, 0.2597, 0.3897, 0.1522], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0266, 0.0264, 0.0239, 0.0343, 0.0335, 0.0250, 0.0362], device='cuda:0'), out_proj_covar=tensor([1.5067e-04, 9.8764e-05, 1.1246e-04, 1.0305e-04, 1.4408e-04, 1.3114e-04, 1.0007e-04, 1.4784e-04], device='cuda:0') 2023-03-08 19:42:46,964 INFO [train2.py:809] (0/4) Epoch 19, batch 3150, loss[ctc_loss=0.05356, att_loss=0.2206, loss=0.1872, over 16182.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006228, over 41.00 utterances.], tot_loss[ctc_loss=0.07802, att_loss=0.2375, loss=0.2056, over 3271581.82 frames. utt_duration=1266 frames, utt_pad_proportion=0.0503, over 10345.13 utterances.], batch size: 41, lr: 5.64e-03, grad_scale: 8.0 2023-03-08 19:43:01,197 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0896, 2.7195, 3.5023, 2.8097, 3.3014, 4.2765, 4.0304, 2.8465], device='cuda:0'), covar=tensor([0.0399, 0.1766, 0.1045, 0.1332, 0.1045, 0.0738, 0.0625, 0.1430], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0243, 0.0275, 0.0215, 0.0261, 0.0358, 0.0256, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 19:43:46,994 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=74896.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:44:06,750 INFO [train2.py:809] (0/4) Epoch 19, batch 3200, loss[ctc_loss=0.06954, att_loss=0.2335, loss=0.2007, over 16276.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006782, over 43.00 utterances.], tot_loss[ctc_loss=0.07828, att_loss=0.2378, loss=0.2059, over 3273284.65 frames. utt_duration=1258 frames, utt_pad_proportion=0.05173, over 10420.14 utterances.], batch size: 43, lr: 5.64e-03, grad_scale: 8.0 2023-03-08 19:44:08,696 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74909.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:44:33,295 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5355, 4.6475, 4.6859, 4.6463, 5.1667, 4.5377, 4.5893, 2.3075], device='cuda:0'), covar=tensor([0.0214, 0.0285, 0.0311, 0.0293, 0.0906, 0.0210, 0.0307, 0.2099], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0176, 0.0181, 0.0196, 0.0365, 0.0150, 0.0167, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:44:36,408 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5765, 4.6225, 4.6977, 4.6818, 5.2117, 4.5619, 4.6043, 2.3188], device='cuda:0'), covar=tensor([0.0191, 0.0266, 0.0266, 0.0285, 0.0798, 0.0187, 0.0301, 0.2090], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0175, 0.0180, 0.0196, 0.0365, 0.0150, 0.0167, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:44:51,312 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.517e+02 1.907e+02 2.285e+02 3.012e+02 5.502e+02, threshold=4.570e+02, percent-clipped=3.0 2023-03-08 19:45:03,893 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=74944.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:45:15,355 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=74951.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:45:20,248 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3837, 5.3420, 5.1089, 3.3128, 5.2222, 4.9020, 4.7237, 3.0896], device='cuda:0'), covar=tensor([0.0101, 0.0078, 0.0240, 0.0803, 0.0072, 0.0162, 0.0230, 0.1195], device='cuda:0'), in_proj_covar=tensor([0.0071, 0.0099, 0.0099, 0.0108, 0.0082, 0.0108, 0.0097, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 19:45:26,715 INFO [train2.py:809] (0/4) Epoch 19, batch 3250, loss[ctc_loss=0.06846, att_loss=0.2462, loss=0.2107, over 17447.00 frames. utt_duration=1109 frames, utt_pad_proportion=0.03016, over 63.00 utterances.], tot_loss[ctc_loss=0.07852, att_loss=0.2381, loss=0.2062, over 3273900.72 frames. utt_duration=1252 frames, utt_pad_proportion=0.05314, over 10471.50 utterances.], batch size: 63, lr: 5.64e-03, grad_scale: 8.0 2023-03-08 19:45:36,425 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=74964.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:45:45,817 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=74970.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:46:46,968 INFO [train2.py:809] (0/4) Epoch 19, batch 3300, loss[ctc_loss=0.08622, att_loss=0.2257, loss=0.1978, over 16179.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006259, over 41.00 utterances.], tot_loss[ctc_loss=0.07802, att_loss=0.2371, loss=0.2053, over 3264675.14 frames. utt_duration=1259 frames, utt_pad_proportion=0.05506, over 10381.06 utterances.], batch size: 41, lr: 5.63e-03, grad_scale: 8.0 2023-03-08 19:47:02,794 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3842, 4.5291, 4.5807, 4.5969, 5.1564, 4.4771, 4.5532, 2.4100], device='cuda:0'), covar=tensor([0.0234, 0.0349, 0.0311, 0.0275, 0.0904, 0.0218, 0.0322, 0.1895], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0175, 0.0180, 0.0195, 0.0363, 0.0149, 0.0166, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:47:13,506 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75025.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:47:23,418 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-08 19:47:31,947 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 2.051e+02 2.410e+02 2.967e+02 4.688e+02, threshold=4.820e+02, percent-clipped=1.0 2023-03-08 19:48:01,069 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 19:48:06,385 INFO [train2.py:809] (0/4) Epoch 19, batch 3350, loss[ctc_loss=0.06625, att_loss=0.2221, loss=0.1909, over 15967.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.005538, over 41.00 utterances.], tot_loss[ctc_loss=0.07794, att_loss=0.2373, loss=0.2054, over 3264407.10 frames. utt_duration=1257 frames, utt_pad_proportion=0.05609, over 10401.75 utterances.], batch size: 41, lr: 5.63e-03, grad_scale: 8.0 2023-03-08 19:49:03,333 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7257, 6.0329, 5.4515, 5.7725, 5.7059, 5.2773, 5.4527, 5.1529], device='cuda:0'), covar=tensor([0.1288, 0.0858, 0.0887, 0.0775, 0.0820, 0.1464, 0.2263, 0.2343], device='cuda:0'), in_proj_covar=tensor([0.0506, 0.0589, 0.0444, 0.0439, 0.0415, 0.0457, 0.0600, 0.0515], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 19:49:10,312 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3067, 4.2044, 4.3118, 4.3774, 4.9393, 4.2796, 4.2750, 2.3091], device='cuda:0'), covar=tensor([0.0243, 0.0358, 0.0349, 0.0273, 0.0852, 0.0263, 0.0371, 0.2126], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0175, 0.0180, 0.0195, 0.0364, 0.0149, 0.0166, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:49:26,185 INFO [train2.py:809] (0/4) Epoch 19, batch 3400, loss[ctc_loss=0.06959, att_loss=0.2254, loss=0.1942, over 15957.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006315, over 41.00 utterances.], tot_loss[ctc_loss=0.07765, att_loss=0.238, loss=0.2059, over 3270159.49 frames. utt_duration=1269 frames, utt_pad_proportion=0.05011, over 10322.33 utterances.], batch size: 41, lr: 5.63e-03, grad_scale: 8.0 2023-03-08 19:49:36,123 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6145, 4.5608, 4.5964, 4.6083, 5.2679, 4.5053, 4.5608, 2.4886], device='cuda:0'), covar=tensor([0.0198, 0.0347, 0.0306, 0.0317, 0.0638, 0.0208, 0.0366, 0.1847], device='cuda:0'), in_proj_covar=tensor([0.0152, 0.0174, 0.0180, 0.0194, 0.0362, 0.0149, 0.0166, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 19:50:04,399 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5174, 2.5875, 5.0345, 4.0156, 2.9230, 4.2980, 4.8394, 4.6474], device='cuda:0'), covar=tensor([0.0276, 0.1562, 0.0199, 0.0885, 0.1807, 0.0250, 0.0139, 0.0265], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0241, 0.0178, 0.0308, 0.0264, 0.0206, 0.0163, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 19:50:08,941 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9459, 5.2452, 4.8490, 5.3539, 4.7106, 4.8828, 5.3820, 5.1525], device='cuda:0'), covar=tensor([0.0592, 0.0290, 0.0743, 0.0277, 0.0398, 0.0261, 0.0237, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0382, 0.0313, 0.0360, 0.0335, 0.0315, 0.0235, 0.0296, 0.0278], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 19:50:11,773 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.156e+02 2.118e+02 2.457e+02 2.938e+02 6.241e+02, threshold=4.915e+02, percent-clipped=4.0 2023-03-08 19:50:46,058 INFO [train2.py:809] (0/4) Epoch 19, batch 3450, loss[ctc_loss=0.05796, att_loss=0.2053, loss=0.1759, over 14066.00 frames. utt_duration=1816 frames, utt_pad_proportion=0.05052, over 31.00 utterances.], tot_loss[ctc_loss=0.0777, att_loss=0.2377, loss=0.2057, over 3259963.72 frames. utt_duration=1241 frames, utt_pad_proportion=0.06061, over 10519.00 utterances.], batch size: 31, lr: 5.63e-03, grad_scale: 8.0 2023-03-08 19:52:06,153 INFO [train2.py:809] (0/4) Epoch 19, batch 3500, loss[ctc_loss=0.09148, att_loss=0.2583, loss=0.2249, over 16631.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004916, over 47.00 utterances.], tot_loss[ctc_loss=0.07746, att_loss=0.2374, loss=0.2054, over 3253668.05 frames. utt_duration=1249 frames, utt_pad_proportion=0.05816, over 10428.72 utterances.], batch size: 47, lr: 5.63e-03, grad_scale: 8.0 2023-03-08 19:52:10,253 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-08 19:52:15,061 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-08 19:52:38,228 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75228.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:52:52,420 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.557e+02 2.026e+02 2.447e+02 3.089e+02 8.845e+02, threshold=4.893e+02, percent-clipped=4.0 2023-03-08 19:53:15,776 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75251.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:53:26,627 INFO [train2.py:809] (0/4) Epoch 19, batch 3550, loss[ctc_loss=0.06523, att_loss=0.2179, loss=0.1873, over 16179.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006929, over 41.00 utterances.], tot_loss[ctc_loss=0.0768, att_loss=0.2368, loss=0.2048, over 3257334.81 frames. utt_duration=1254 frames, utt_pad_proportion=0.05676, over 10401.80 utterances.], batch size: 41, lr: 5.62e-03, grad_scale: 8.0 2023-03-08 19:53:37,467 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75265.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:53:45,173 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=75270.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:54:15,755 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75289.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:54:32,147 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75299.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:54:46,280 INFO [train2.py:809] (0/4) Epoch 19, batch 3600, loss[ctc_loss=0.08326, att_loss=0.2226, loss=0.1947, over 16006.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007957, over 40.00 utterances.], tot_loss[ctc_loss=0.077, att_loss=0.2364, loss=0.2046, over 3249209.00 frames. utt_duration=1255 frames, utt_pad_proportion=0.05844, over 10364.87 utterances.], batch size: 40, lr: 5.62e-03, grad_scale: 8.0 2023-03-08 19:55:04,942 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75320.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 19:55:23,534 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=75331.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 19:55:32,225 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 1.965e+02 2.324e+02 2.872e+02 5.380e+02, threshold=4.648e+02, percent-clipped=3.0 2023-03-08 19:56:05,131 INFO [train2.py:809] (0/4) Epoch 19, batch 3650, loss[ctc_loss=0.09738, att_loss=0.2457, loss=0.216, over 16134.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005833, over 42.00 utterances.], tot_loss[ctc_loss=0.07764, att_loss=0.2375, loss=0.2055, over 3260279.55 frames. utt_duration=1238 frames, utt_pad_proportion=0.06024, over 10547.17 utterances.], batch size: 42, lr: 5.62e-03, grad_scale: 8.0 2023-03-08 19:56:21,536 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.82 vs. limit=2.0 2023-03-08 19:57:25,710 INFO [train2.py:809] (0/4) Epoch 19, batch 3700, loss[ctc_loss=0.07656, att_loss=0.2454, loss=0.2116, over 17028.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.008218, over 51.00 utterances.], tot_loss[ctc_loss=0.07799, att_loss=0.2377, loss=0.2057, over 3259174.36 frames. utt_duration=1220 frames, utt_pad_proportion=0.06596, over 10702.87 utterances.], batch size: 51, lr: 5.62e-03, grad_scale: 8.0 2023-03-08 19:58:09,149 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-03-08 19:58:11,211 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.172e+02 2.064e+02 2.535e+02 3.260e+02 9.270e+02, threshold=5.071e+02, percent-clipped=6.0 2023-03-08 19:58:44,484 INFO [train2.py:809] (0/4) Epoch 19, batch 3750, loss[ctc_loss=0.05998, att_loss=0.2233, loss=0.1907, over 16012.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007017, over 40.00 utterances.], tot_loss[ctc_loss=0.07814, att_loss=0.2374, loss=0.2056, over 3260066.27 frames. utt_duration=1251 frames, utt_pad_proportion=0.05737, over 10432.49 utterances.], batch size: 40, lr: 5.62e-03, grad_scale: 8.0 2023-03-08 19:59:19,414 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4366, 2.3892, 4.8803, 3.8681, 2.9589, 4.1989, 4.7523, 4.6541], device='cuda:0'), covar=tensor([0.0245, 0.1765, 0.0192, 0.0917, 0.1738, 0.0251, 0.0144, 0.0219], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0241, 0.0178, 0.0308, 0.0263, 0.0205, 0.0162, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002], device='cuda:0') 2023-03-08 20:00:03,970 INFO [train2.py:809] (0/4) Epoch 19, batch 3800, loss[ctc_loss=0.1223, att_loss=0.2754, loss=0.2448, over 17421.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03284, over 63.00 utterances.], tot_loss[ctc_loss=0.07805, att_loss=0.2376, loss=0.2057, over 3260212.87 frames. utt_duration=1263 frames, utt_pad_proportion=0.05401, over 10335.45 utterances.], batch size: 63, lr: 5.62e-03, grad_scale: 8.0 2023-03-08 20:00:19,673 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1356, 5.1803, 4.8929, 2.9479, 4.9542, 4.7623, 4.5604, 3.0696], device='cuda:0'), covar=tensor([0.0112, 0.0078, 0.0273, 0.1047, 0.0087, 0.0180, 0.0249, 0.1215], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0100, 0.0101, 0.0110, 0.0083, 0.0110, 0.0098, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 20:00:50,186 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.424e+02 2.066e+02 2.537e+02 3.345e+02 7.460e+02, threshold=5.074e+02, percent-clipped=4.0 2023-03-08 20:01:20,856 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5843, 2.1357, 2.3031, 2.5785, 3.0439, 2.5717, 2.2052, 2.8913], device='cuda:0'), covar=tensor([0.2257, 0.4383, 0.3341, 0.1941, 0.2013, 0.1760, 0.3659, 0.1206], device='cuda:0'), in_proj_covar=tensor([0.0111, 0.0118, 0.0117, 0.0104, 0.0117, 0.0103, 0.0125, 0.0092], device='cuda:0'), out_proj_covar=tensor([8.3026e-05, 9.1353e-05, 9.1577e-05, 8.0609e-05, 8.5959e-05, 8.1892e-05, 9.2431e-05, 7.4220e-05], device='cuda:0') 2023-03-08 20:01:23,605 INFO [train2.py:809] (0/4) Epoch 19, batch 3850, loss[ctc_loss=0.06283, att_loss=0.2195, loss=0.1882, over 14552.00 frames. utt_duration=1820 frames, utt_pad_proportion=0.04043, over 32.00 utterances.], tot_loss[ctc_loss=0.07738, att_loss=0.2369, loss=0.205, over 3249210.84 frames. utt_duration=1257 frames, utt_pad_proportion=0.05872, over 10349.25 utterances.], batch size: 32, lr: 5.61e-03, grad_scale: 8.0 2023-03-08 20:01:34,612 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75565.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:01:46,579 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3394, 3.0017, 3.3373, 4.2929, 3.8992, 3.9848, 2.9245, 2.3632], device='cuda:0'), covar=tensor([0.0746, 0.1994, 0.0926, 0.0623, 0.0964, 0.0459, 0.1565, 0.2251], device='cuda:0'), in_proj_covar=tensor([0.0178, 0.0214, 0.0188, 0.0215, 0.0220, 0.0176, 0.0203, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:02:03,497 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75584.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:02:40,366 INFO [train2.py:809] (0/4) Epoch 19, batch 3900, loss[ctc_loss=0.08658, att_loss=0.2446, loss=0.213, over 16403.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.00685, over 44.00 utterances.], tot_loss[ctc_loss=0.07759, att_loss=0.2374, loss=0.2055, over 3260184.00 frames. utt_duration=1231 frames, utt_pad_proportion=0.06153, over 10606.96 utterances.], batch size: 44, lr: 5.61e-03, grad_scale: 8.0 2023-03-08 20:02:47,783 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75613.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:02:58,537 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75620.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:03:07,684 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=75626.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:03:21,326 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0711, 5.2410, 5.5488, 5.4816, 5.5438, 5.9741, 5.2130, 6.1206], device='cuda:0'), covar=tensor([0.0693, 0.0805, 0.0798, 0.1232, 0.1776, 0.0887, 0.0710, 0.0624], device='cuda:0'), in_proj_covar=tensor([0.0845, 0.0498, 0.0585, 0.0643, 0.0855, 0.0604, 0.0480, 0.0585], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 20:03:24,257 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 1.998e+02 2.333e+02 2.845e+02 5.732e+02, threshold=4.667e+02, percent-clipped=1.0 2023-03-08 20:03:57,008 INFO [train2.py:809] (0/4) Epoch 19, batch 3950, loss[ctc_loss=0.0721, att_loss=0.2377, loss=0.2046, over 16883.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006568, over 49.00 utterances.], tot_loss[ctc_loss=0.07685, att_loss=0.2364, loss=0.2045, over 3252231.25 frames. utt_duration=1252 frames, utt_pad_proportion=0.0577, over 10405.66 utterances.], batch size: 49, lr: 5.61e-03, grad_scale: 8.0 2023-03-08 20:04:12,252 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75668.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:04:47,745 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-19.pt 2023-03-08 20:05:12,140 INFO [train2.py:809] (0/4) Epoch 20, batch 0, loss[ctc_loss=0.06043, att_loss=0.2112, loss=0.181, over 15961.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006513, over 41.00 utterances.], tot_loss[ctc_loss=0.06043, att_loss=0.2112, loss=0.181, over 15961.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006513, over 41.00 utterances.], batch size: 41, lr: 5.46e-03, grad_scale: 8.0 2023-03-08 20:05:12,142 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 20:05:24,208 INFO [train2.py:843] (0/4) Epoch 20, validation: ctc_loss=0.04136, att_loss=0.235, loss=0.1963, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 20:05:24,209 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 20:06:35,312 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 2.109e+02 2.667e+02 3.273e+02 6.239e+02, threshold=5.334e+02, percent-clipped=6.0 2023-03-08 20:06:37,681 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-08 20:06:43,082 INFO [train2.py:809] (0/4) Epoch 20, batch 50, loss[ctc_loss=0.06332, att_loss=0.2396, loss=0.2044, over 16327.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006323, over 45.00 utterances.], tot_loss[ctc_loss=0.07618, att_loss=0.2397, loss=0.207, over 737747.21 frames. utt_duration=1233 frames, utt_pad_proportion=0.06546, over 2396.38 utterances.], batch size: 45, lr: 5.46e-03, grad_scale: 8.0 2023-03-08 20:07:27,229 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-03-08 20:08:03,712 INFO [train2.py:809] (0/4) Epoch 20, batch 100, loss[ctc_loss=0.07212, att_loss=0.2477, loss=0.2126, over 16774.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006052, over 48.00 utterances.], tot_loss[ctc_loss=0.07649, att_loss=0.239, loss=0.2065, over 1302695.62 frames. utt_duration=1276 frames, utt_pad_proportion=0.04618, over 4087.79 utterances.], batch size: 48, lr: 5.46e-03, grad_scale: 8.0 2023-03-08 20:09:16,371 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 1.985e+02 2.361e+02 2.843e+02 6.853e+02, threshold=4.722e+02, percent-clipped=2.0 2023-03-08 20:09:24,381 INFO [train2.py:809] (0/4) Epoch 20, batch 150, loss[ctc_loss=0.07589, att_loss=0.2383, loss=0.2058, over 16480.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006428, over 46.00 utterances.], tot_loss[ctc_loss=0.07673, att_loss=0.2379, loss=0.2056, over 1736517.95 frames. utt_duration=1248 frames, utt_pad_proportion=0.05424, over 5570.28 utterances.], batch size: 46, lr: 5.46e-03, grad_scale: 8.0 2023-03-08 20:10:33,713 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75884.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:10:46,186 INFO [train2.py:809] (0/4) Epoch 20, batch 200, loss[ctc_loss=0.06705, att_loss=0.2093, loss=0.1809, over 14462.00 frames. utt_duration=1809 frames, utt_pad_proportion=0.04466, over 32.00 utterances.], tot_loss[ctc_loss=0.0759, att_loss=0.2359, loss=0.2039, over 2069217.64 frames. utt_duration=1277 frames, utt_pad_proportion=0.05082, over 6488.39 utterances.], batch size: 32, lr: 5.46e-03, grad_scale: 8.0 2023-03-08 20:11:42,056 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=75926.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 20:11:51,158 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75932.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:11:59,000 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.248e+02 2.084e+02 2.421e+02 2.926e+02 5.998e+02, threshold=4.842e+02, percent-clipped=2.0 2023-03-08 20:12:06,829 INFO [train2.py:809] (0/4) Epoch 20, batch 250, loss[ctc_loss=0.06707, att_loss=0.2148, loss=0.1853, over 15637.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009478, over 37.00 utterances.], tot_loss[ctc_loss=0.07541, att_loss=0.2347, loss=0.2028, over 2328574.76 frames. utt_duration=1286 frames, utt_pad_proportion=0.05137, over 7248.68 utterances.], batch size: 37, lr: 5.45e-03, grad_scale: 8.0 2023-03-08 20:12:59,203 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=75974.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:13:08,604 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6203, 4.7032, 4.6832, 4.8248, 5.2617, 4.5677, 4.5621, 2.4939], device='cuda:0'), covar=tensor([0.0212, 0.0303, 0.0279, 0.0229, 0.0738, 0.0208, 0.0312, 0.1874], device='cuda:0'), in_proj_covar=tensor([0.0151, 0.0174, 0.0176, 0.0192, 0.0356, 0.0147, 0.0165, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:13:09,181 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-03-08 20:13:10,179 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1037, 4.3858, 4.4438, 4.7131, 2.7448, 4.5183, 2.6666, 1.3825], device='cuda:0'), covar=tensor([0.0470, 0.0213, 0.0627, 0.0180, 0.1634, 0.0153, 0.1521, 0.1912], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0153, 0.0253, 0.0150, 0.0215, 0.0133, 0.0227, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:13:25,300 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0165, 5.3212, 4.6560, 5.4406, 4.7872, 4.9737, 5.4288, 5.2282], device='cuda:0'), covar=tensor([0.0490, 0.0297, 0.0896, 0.0279, 0.0375, 0.0237, 0.0247, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0378, 0.0311, 0.0360, 0.0336, 0.0313, 0.0233, 0.0295, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 20:13:26,591 INFO [train2.py:809] (0/4) Epoch 20, batch 300, loss[ctc_loss=0.06633, att_loss=0.2262, loss=0.1942, over 16131.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005772, over 42.00 utterances.], tot_loss[ctc_loss=0.07535, att_loss=0.2349, loss=0.203, over 2538845.53 frames. utt_duration=1309 frames, utt_pad_proportion=0.04535, over 7769.22 utterances.], batch size: 42, lr: 5.45e-03, grad_scale: 8.0 2023-03-08 20:13:39,503 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-76000.pt 2023-03-08 20:14:42,638 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 2.144e+02 2.516e+02 3.059e+02 1.142e+03, threshold=5.032e+02, percent-clipped=7.0 2023-03-08 20:14:50,244 INFO [train2.py:809] (0/4) Epoch 20, batch 350, loss[ctc_loss=0.09015, att_loss=0.2513, loss=0.219, over 17321.00 frames. utt_duration=1006 frames, utt_pad_proportion=0.05118, over 69.00 utterances.], tot_loss[ctc_loss=0.07554, att_loss=0.2358, loss=0.2037, over 2704762.58 frames. utt_duration=1288 frames, utt_pad_proportion=0.04852, over 8408.97 utterances.], batch size: 69, lr: 5.45e-03, grad_scale: 8.0 2023-03-08 20:16:10,635 INFO [train2.py:809] (0/4) Epoch 20, batch 400, loss[ctc_loss=0.05965, att_loss=0.2347, loss=0.1997, over 16865.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.007581, over 49.00 utterances.], tot_loss[ctc_loss=0.07582, att_loss=0.2355, loss=0.2036, over 2824250.45 frames. utt_duration=1256 frames, utt_pad_proportion=0.05878, over 9008.57 utterances.], batch size: 49, lr: 5.45e-03, grad_scale: 8.0 2023-03-08 20:17:22,263 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.475e+02 1.983e+02 2.448e+02 2.910e+02 4.561e+02, threshold=4.897e+02, percent-clipped=0.0 2023-03-08 20:17:29,974 INFO [train2.py:809] (0/4) Epoch 20, batch 450, loss[ctc_loss=0.1122, att_loss=0.254, loss=0.2257, over 14133.00 frames. utt_duration=388.8 frames, utt_pad_proportion=0.3238, over 146.00 utterances.], tot_loss[ctc_loss=0.0765, att_loss=0.2364, loss=0.2044, over 2919882.07 frames. utt_duration=1235 frames, utt_pad_proportion=0.06309, over 9470.82 utterances.], batch size: 146, lr: 5.45e-03, grad_scale: 8.0 2023-03-08 20:17:59,786 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8103, 5.2552, 5.2243, 5.1637, 5.2788, 5.2712, 5.0008, 4.8165], device='cuda:0'), covar=tensor([0.1383, 0.0599, 0.0322, 0.0579, 0.0436, 0.0348, 0.0367, 0.0357], device='cuda:0'), in_proj_covar=tensor([0.0524, 0.0355, 0.0335, 0.0351, 0.0414, 0.0427, 0.0348, 0.0390], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 20:18:18,986 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6996, 3.4306, 3.3838, 2.8861, 3.3571, 3.4742, 3.4609, 2.4236], device='cuda:0'), covar=tensor([0.1194, 0.1383, 0.4085, 0.4622, 0.2206, 0.2633, 0.1016, 0.4823], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0175, 0.0188, 0.0251, 0.0151, 0.0249, 0.0168, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:18:49,883 INFO [train2.py:809] (0/4) Epoch 20, batch 500, loss[ctc_loss=0.09306, att_loss=0.2488, loss=0.2176, over 16638.00 frames. utt_duration=673.7 frames, utt_pad_proportion=0.1526, over 99.00 utterances.], tot_loss[ctc_loss=0.07629, att_loss=0.2363, loss=0.2043, over 2998970.57 frames. utt_duration=1237 frames, utt_pad_proportion=0.0605, over 9710.35 utterances.], batch size: 99, lr: 5.45e-03, grad_scale: 8.0 2023-03-08 20:19:11,440 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-03-08 20:20:01,813 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 1.846e+02 2.334e+02 2.812e+02 5.634e+02, threshold=4.669e+02, percent-clipped=1.0 2023-03-08 20:20:09,795 INFO [train2.py:809] (0/4) Epoch 20, batch 550, loss[ctc_loss=0.07941, att_loss=0.244, loss=0.211, over 17054.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.007655, over 52.00 utterances.], tot_loss[ctc_loss=0.07587, att_loss=0.2356, loss=0.2037, over 3052860.46 frames. utt_duration=1231 frames, utt_pad_proportion=0.06192, over 9931.22 utterances.], batch size: 52, lr: 5.44e-03, grad_scale: 8.0 2023-03-08 20:20:10,104 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1745, 3.8198, 3.7916, 3.2124, 3.7688, 3.9032, 3.7897, 2.7920], device='cuda:0'), covar=tensor([0.0933, 0.0987, 0.2153, 0.3758, 0.1086, 0.4002, 0.1043, 0.4003], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0175, 0.0187, 0.0249, 0.0150, 0.0248, 0.0167, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:20:54,228 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76269.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:21:29,922 INFO [train2.py:809] (0/4) Epoch 20, batch 600, loss[ctc_loss=0.07528, att_loss=0.2203, loss=0.1913, over 15628.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.008116, over 37.00 utterances.], tot_loss[ctc_loss=0.07568, att_loss=0.2356, loss=0.2037, over 3103518.69 frames. utt_duration=1240 frames, utt_pad_proportion=0.05648, over 10023.38 utterances.], batch size: 37, lr: 5.44e-03, grad_scale: 16.0 2023-03-08 20:21:33,154 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76294.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:21:44,444 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1027, 3.8801, 3.3748, 3.5934, 4.0921, 3.8253, 3.1770, 4.3461], device='cuda:0'), covar=tensor([0.0991, 0.0495, 0.1056, 0.0712, 0.0720, 0.0637, 0.0864, 0.0555], device='cuda:0'), in_proj_covar=tensor([0.0200, 0.0213, 0.0225, 0.0198, 0.0274, 0.0237, 0.0200, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 20:22:20,674 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76323.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:22:31,287 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76330.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 20:22:41,630 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.413e+02 1.945e+02 2.306e+02 2.690e+02 7.016e+02, threshold=4.612e+02, percent-clipped=2.0 2023-03-08 20:22:49,615 INFO [train2.py:809] (0/4) Epoch 20, batch 650, loss[ctc_loss=0.07501, att_loss=0.2405, loss=0.2074, over 16475.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006805, over 46.00 utterances.], tot_loss[ctc_loss=0.07522, att_loss=0.2356, loss=0.2035, over 3146221.07 frames. utt_duration=1267 frames, utt_pad_proportion=0.04875, over 9944.83 utterances.], batch size: 46, lr: 5.44e-03, grad_scale: 16.0 2023-03-08 20:23:11,312 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76355.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:23:18,771 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1898, 5.2108, 4.9981, 3.0356, 4.9777, 4.8047, 4.5215, 2.9254], device='cuda:0'), covar=tensor([0.0102, 0.0081, 0.0236, 0.0989, 0.0085, 0.0184, 0.0295, 0.1270], device='cuda:0'), in_proj_covar=tensor([0.0072, 0.0100, 0.0101, 0.0110, 0.0083, 0.0110, 0.0097, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 20:23:56,945 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76384.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 20:24:09,245 INFO [train2.py:809] (0/4) Epoch 20, batch 700, loss[ctc_loss=0.07463, att_loss=0.2416, loss=0.2082, over 16412.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007042, over 44.00 utterances.], tot_loss[ctc_loss=0.0756, att_loss=0.2359, loss=0.2038, over 3182009.80 frames. utt_duration=1294 frames, utt_pad_proportion=0.04125, over 9847.95 utterances.], batch size: 44, lr: 5.44e-03, grad_scale: 16.0 2023-03-08 20:25:20,467 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 1.882e+02 2.322e+02 2.702e+02 6.742e+02, threshold=4.645e+02, percent-clipped=2.0 2023-03-08 20:25:28,802 INFO [train2.py:809] (0/4) Epoch 20, batch 750, loss[ctc_loss=0.05725, att_loss=0.2093, loss=0.1789, over 15952.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006424, over 41.00 utterances.], tot_loss[ctc_loss=0.07538, att_loss=0.2357, loss=0.2037, over 3201112.96 frames. utt_duration=1289 frames, utt_pad_proportion=0.04325, over 9945.15 utterances.], batch size: 41, lr: 5.44e-03, grad_scale: 16.0 2023-03-08 20:26:20,847 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 20:26:48,171 INFO [train2.py:809] (0/4) Epoch 20, batch 800, loss[ctc_loss=0.06924, att_loss=0.2364, loss=0.203, over 16530.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.007022, over 45.00 utterances.], tot_loss[ctc_loss=0.07668, att_loss=0.2366, loss=0.2046, over 3218683.87 frames. utt_duration=1282 frames, utt_pad_proportion=0.04529, over 10051.14 utterances.], batch size: 45, lr: 5.44e-03, grad_scale: 16.0 2023-03-08 20:28:00,598 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.342e+02 1.827e+02 2.351e+02 2.962e+02 7.560e+02, threshold=4.702e+02, percent-clipped=5.0 2023-03-08 20:28:07,270 INFO [train2.py:809] (0/4) Epoch 20, batch 850, loss[ctc_loss=0.06965, att_loss=0.2423, loss=0.2078, over 17009.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.00853, over 51.00 utterances.], tot_loss[ctc_loss=0.07635, att_loss=0.2367, loss=0.2047, over 3240238.32 frames. utt_duration=1279 frames, utt_pad_proportion=0.04383, over 10145.42 utterances.], batch size: 51, lr: 5.43e-03, grad_scale: 8.0 2023-03-08 20:28:45,906 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7880, 4.7298, 4.7098, 4.7666, 5.2967, 4.7187, 4.7224, 2.7505], device='cuda:0'), covar=tensor([0.0172, 0.0347, 0.0291, 0.0289, 0.0951, 0.0175, 0.0275, 0.1885], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0175, 0.0178, 0.0193, 0.0360, 0.0149, 0.0166, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:28:55,430 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-08 20:29:26,681 INFO [train2.py:809] (0/4) Epoch 20, batch 900, loss[ctc_loss=0.1041, att_loss=0.2627, loss=0.231, over 17340.00 frames. utt_duration=1177 frames, utt_pad_proportion=0.02175, over 59.00 utterances.], tot_loss[ctc_loss=0.07721, att_loss=0.237, loss=0.205, over 3237192.51 frames. utt_duration=1232 frames, utt_pad_proportion=0.05957, over 10520.88 utterances.], batch size: 59, lr: 5.43e-03, grad_scale: 8.0 2023-03-08 20:30:21,578 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76625.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 20:30:28,638 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.54 vs. limit=2.0 2023-03-08 20:30:41,671 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.428e+02 2.021e+02 2.452e+02 3.045e+02 6.796e+02, threshold=4.904e+02, percent-clipped=5.0 2023-03-08 20:30:47,767 INFO [train2.py:809] (0/4) Epoch 20, batch 950, loss[ctc_loss=0.103, att_loss=0.2622, loss=0.2303, over 17047.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008366, over 52.00 utterances.], tot_loss[ctc_loss=0.07674, att_loss=0.2368, loss=0.2048, over 3240577.49 frames. utt_duration=1251 frames, utt_pad_proportion=0.05484, over 10374.77 utterances.], batch size: 52, lr: 5.43e-03, grad_scale: 8.0 2023-03-08 20:31:02,270 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76650.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:31:44,700 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5404, 4.4495, 4.6005, 4.6299, 5.1625, 4.4810, 4.5164, 2.4120], device='cuda:0'), covar=tensor([0.0196, 0.0316, 0.0255, 0.0246, 0.0760, 0.0206, 0.0290, 0.1904], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0176, 0.0178, 0.0193, 0.0359, 0.0149, 0.0166, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:31:46,157 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1716, 5.0934, 4.8623, 2.7142, 4.8812, 4.8749, 4.5734, 2.7915], device='cuda:0'), covar=tensor([0.0115, 0.0129, 0.0301, 0.1240, 0.0111, 0.0188, 0.0305, 0.1468], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0101, 0.0102, 0.0111, 0.0083, 0.0110, 0.0098, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 20:31:47,535 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=76679.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:32:08,845 INFO [train2.py:809] (0/4) Epoch 20, batch 1000, loss[ctc_loss=0.07264, att_loss=0.2403, loss=0.2068, over 17067.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008818, over 53.00 utterances.], tot_loss[ctc_loss=0.07651, att_loss=0.2367, loss=0.2047, over 3241955.95 frames. utt_duration=1217 frames, utt_pad_proportion=0.0643, over 10665.56 utterances.], batch size: 53, lr: 5.43e-03, grad_scale: 8.0 2023-03-08 20:33:21,295 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 1.956e+02 2.404e+02 2.887e+02 6.438e+02, threshold=4.808e+02, percent-clipped=2.0 2023-03-08 20:33:28,407 INFO [train2.py:809] (0/4) Epoch 20, batch 1050, loss[ctc_loss=0.09167, att_loss=0.2624, loss=0.2282, over 17471.00 frames. utt_duration=1111 frames, utt_pad_proportion=0.02989, over 63.00 utterances.], tot_loss[ctc_loss=0.07562, att_loss=0.2362, loss=0.2041, over 3255273.62 frames. utt_duration=1247 frames, utt_pad_proportion=0.05523, over 10454.84 utterances.], batch size: 63, lr: 5.43e-03, grad_scale: 8.0 2023-03-08 20:34:07,669 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.48 vs. limit=5.0 2023-03-08 20:34:48,261 INFO [train2.py:809] (0/4) Epoch 20, batch 1100, loss[ctc_loss=0.07598, att_loss=0.247, loss=0.2128, over 17108.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01553, over 56.00 utterances.], tot_loss[ctc_loss=0.07678, att_loss=0.2373, loss=0.2052, over 3262066.60 frames. utt_duration=1215 frames, utt_pad_proportion=0.06175, over 10749.04 utterances.], batch size: 56, lr: 5.42e-03, grad_scale: 8.0 2023-03-08 20:34:53,846 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3110, 3.9824, 3.4716, 3.7207, 4.1849, 3.8251, 3.0279, 4.5046], device='cuda:0'), covar=tensor([0.0862, 0.0561, 0.0924, 0.0557, 0.0750, 0.0653, 0.0894, 0.0505], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0212, 0.0223, 0.0196, 0.0272, 0.0236, 0.0199, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 20:35:01,855 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1276, 5.3299, 5.2976, 5.2692, 5.3758, 5.3134, 5.0433, 4.8810], device='cuda:0'), covar=tensor([0.0991, 0.0555, 0.0266, 0.0539, 0.0296, 0.0319, 0.0362, 0.0296], device='cuda:0'), in_proj_covar=tensor([0.0524, 0.0358, 0.0339, 0.0354, 0.0414, 0.0429, 0.0350, 0.0390], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 20:36:02,081 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.131e+02 2.008e+02 2.417e+02 3.031e+02 6.018e+02, threshold=4.833e+02, percent-clipped=3.0 2023-03-08 20:36:03,940 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=76839.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:36:08,258 INFO [train2.py:809] (0/4) Epoch 20, batch 1150, loss[ctc_loss=0.06347, att_loss=0.2157, loss=0.1853, over 15899.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.00849, over 39.00 utterances.], tot_loss[ctc_loss=0.07644, att_loss=0.2377, loss=0.2055, over 3267397.01 frames. utt_duration=1225 frames, utt_pad_proportion=0.05933, over 10682.40 utterances.], batch size: 39, lr: 5.42e-03, grad_scale: 8.0 2023-03-08 20:36:31,193 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7217, 5.0284, 5.2729, 5.1226, 5.2119, 5.6693, 5.1576, 5.7628], device='cuda:0'), covar=tensor([0.0816, 0.0858, 0.0911, 0.1366, 0.2008, 0.0963, 0.0727, 0.0765], device='cuda:0'), in_proj_covar=tensor([0.0856, 0.0501, 0.0583, 0.0646, 0.0857, 0.0611, 0.0477, 0.0592], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 20:37:28,704 INFO [train2.py:809] (0/4) Epoch 20, batch 1200, loss[ctc_loss=0.06081, att_loss=0.2318, loss=0.1976, over 16699.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006082, over 46.00 utterances.], tot_loss[ctc_loss=0.07665, att_loss=0.2386, loss=0.2062, over 3278111.20 frames. utt_duration=1229 frames, utt_pad_proportion=0.0556, over 10682.43 utterances.], batch size: 46, lr: 5.42e-03, grad_scale: 8.0 2023-03-08 20:37:41,987 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=76900.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:38:21,806 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76925.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:38:42,102 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 1.866e+02 2.347e+02 2.914e+02 5.639e+02, threshold=4.693e+02, percent-clipped=1.0 2023-03-08 20:38:49,690 INFO [train2.py:809] (0/4) Epoch 20, batch 1250, loss[ctc_loss=0.0675, att_loss=0.2364, loss=0.2027, over 16628.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005095, over 47.00 utterances.], tot_loss[ctc_loss=0.07706, att_loss=0.2387, loss=0.2064, over 3275984.50 frames. utt_duration=1223 frames, utt_pad_proportion=0.06001, over 10730.47 utterances.], batch size: 47, lr: 5.42e-03, grad_scale: 8.0 2023-03-08 20:39:02,583 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76950.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:39:19,861 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5157, 2.4368, 2.4473, 2.2712, 2.4548, 2.3607, 2.5281, 1.8799], device='cuda:0'), covar=tensor([0.1120, 0.2119, 0.2565, 0.3409, 0.1663, 0.2585, 0.1386, 0.4101], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0175, 0.0187, 0.0246, 0.0150, 0.0246, 0.0166, 0.0208], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:39:38,068 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76973.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:39:48,060 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=76979.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:40:10,272 INFO [train2.py:809] (0/4) Epoch 20, batch 1300, loss[ctc_loss=0.06107, att_loss=0.2208, loss=0.1889, over 16001.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007636, over 40.00 utterances.], tot_loss[ctc_loss=0.077, att_loss=0.2385, loss=0.2062, over 3269915.69 frames. utt_duration=1219 frames, utt_pad_proportion=0.06104, over 10745.04 utterances.], batch size: 40, lr: 5.42e-03, grad_scale: 8.0 2023-03-08 20:40:19,483 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=76998.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:41:04,685 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77027.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 20:41:22,762 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 2.046e+02 2.509e+02 3.001e+02 6.074e+02, threshold=5.017e+02, percent-clipped=4.0 2023-03-08 20:41:29,839 INFO [train2.py:809] (0/4) Epoch 20, batch 1350, loss[ctc_loss=0.09392, att_loss=0.253, loss=0.2212, over 17076.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008383, over 53.00 utterances.], tot_loss[ctc_loss=0.07662, att_loss=0.238, loss=0.2058, over 3271991.71 frames. utt_duration=1241 frames, utt_pad_proportion=0.05577, over 10562.30 utterances.], batch size: 53, lr: 5.42e-03, grad_scale: 8.0 2023-03-08 20:42:48,966 INFO [train2.py:809] (0/4) Epoch 20, batch 1400, loss[ctc_loss=0.05309, att_loss=0.226, loss=0.1914, over 16123.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006418, over 42.00 utterances.], tot_loss[ctc_loss=0.07587, att_loss=0.2377, loss=0.2053, over 3275424.02 frames. utt_duration=1266 frames, utt_pad_proportion=0.04844, over 10358.26 utterances.], batch size: 42, lr: 5.41e-03, grad_scale: 8.0 2023-03-08 20:42:54,044 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77095.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:43:15,359 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9579, 5.1819, 5.5234, 5.2418, 5.3746, 5.8872, 5.2144, 6.0272], device='cuda:0'), covar=tensor([0.0708, 0.0774, 0.0793, 0.1362, 0.1800, 0.0955, 0.0647, 0.0591], device='cuda:0'), in_proj_covar=tensor([0.0863, 0.0499, 0.0585, 0.0645, 0.0858, 0.0611, 0.0476, 0.0590], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 20:44:01,976 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.377e+02 1.974e+02 2.358e+02 2.926e+02 6.381e+02, threshold=4.717e+02, percent-clipped=2.0 2023-03-08 20:44:05,919 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-08 20:44:08,004 INFO [train2.py:809] (0/4) Epoch 20, batch 1450, loss[ctc_loss=0.06351, att_loss=0.2073, loss=0.1786, over 15371.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.009513, over 35.00 utterances.], tot_loss[ctc_loss=0.07634, att_loss=0.2374, loss=0.2052, over 3273240.66 frames. utt_duration=1267 frames, utt_pad_proportion=0.04827, over 10348.86 utterances.], batch size: 35, lr: 5.41e-03, grad_scale: 8.0 2023-03-08 20:44:29,595 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77156.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 20:45:07,769 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6540, 2.2203, 2.5933, 2.5262, 2.7913, 2.6603, 2.7271, 3.2368], device='cuda:0'), covar=tensor([0.1514, 0.3030, 0.1956, 0.1345, 0.1308, 0.1094, 0.1757, 0.1014], device='cuda:0'), in_proj_covar=tensor([0.0113, 0.0122, 0.0116, 0.0105, 0.0117, 0.0101, 0.0125, 0.0093], device='cuda:0'), out_proj_covar=tensor([8.4490e-05, 9.3599e-05, 9.1947e-05, 8.1595e-05, 8.6381e-05, 8.1347e-05, 9.3374e-05, 7.4969e-05], device='cuda:0') 2023-03-08 20:45:27,436 INFO [train2.py:809] (0/4) Epoch 20, batch 1500, loss[ctc_loss=0.08743, att_loss=0.2468, loss=0.2149, over 17344.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03614, over 63.00 utterances.], tot_loss[ctc_loss=0.07618, att_loss=0.2373, loss=0.2051, over 3280136.34 frames. utt_duration=1279 frames, utt_pad_proportion=0.04387, over 10271.70 utterances.], batch size: 63, lr: 5.41e-03, grad_scale: 8.0 2023-03-08 20:45:31,944 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77195.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:45:57,154 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77211.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:46:34,111 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-08 20:46:40,578 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 1.945e+02 2.357e+02 2.826e+02 4.354e+02, threshold=4.714e+02, percent-clipped=0.0 2023-03-08 20:46:46,767 INFO [train2.py:809] (0/4) Epoch 20, batch 1550, loss[ctc_loss=0.0722, att_loss=0.2467, loss=0.2118, over 17458.00 frames. utt_duration=1014 frames, utt_pad_proportion=0.04458, over 69.00 utterances.], tot_loss[ctc_loss=0.07662, att_loss=0.2378, loss=0.2056, over 3274093.80 frames. utt_duration=1244 frames, utt_pad_proportion=0.05452, over 10540.91 utterances.], batch size: 69, lr: 5.41e-03, grad_scale: 8.0 2023-03-08 20:47:29,069 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7800, 2.3447, 2.7075, 2.5196, 2.8883, 2.6922, 2.7279, 3.3260], device='cuda:0'), covar=tensor([0.1885, 0.3176, 0.2402, 0.1765, 0.1515, 0.1395, 0.2285, 0.1369], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0124, 0.0119, 0.0107, 0.0119, 0.0104, 0.0127, 0.0095], device='cuda:0'), out_proj_covar=tensor([8.6777e-05, 9.4969e-05, 9.3537e-05, 8.3091e-05, 8.7929e-05, 8.3297e-05, 9.4958e-05, 7.6389e-05], device='cuda:0') 2023-03-08 20:47:34,325 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77272.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:47:35,784 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77273.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:47:42,393 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77277.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:48:05,772 INFO [train2.py:809] (0/4) Epoch 20, batch 1600, loss[ctc_loss=0.06758, att_loss=0.225, loss=0.1935, over 15968.00 frames. utt_duration=1560 frames, utt_pad_proportion=0.006047, over 41.00 utterances.], tot_loss[ctc_loss=0.07571, att_loss=0.2365, loss=0.2043, over 3268554.85 frames. utt_duration=1252 frames, utt_pad_proportion=0.05388, over 10455.36 utterances.], batch size: 41, lr: 5.41e-03, grad_scale: 8.0 2023-03-08 20:48:58,745 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-03-08 20:49:11,739 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77334.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:49:18,252 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.314e+02 1.917e+02 2.281e+02 2.951e+02 6.682e+02, threshold=4.563e+02, percent-clipped=3.0 2023-03-08 20:49:18,694 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77338.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:49:25,149 INFO [train2.py:809] (0/4) Epoch 20, batch 1650, loss[ctc_loss=0.08828, att_loss=0.2473, loss=0.2155, over 17107.00 frames. utt_duration=867.8 frames, utt_pad_proportion=0.08942, over 79.00 utterances.], tot_loss[ctc_loss=0.07524, att_loss=0.2363, loss=0.2041, over 3270914.63 frames. utt_duration=1267 frames, utt_pad_proportion=0.05027, over 10334.86 utterances.], batch size: 79, lr: 5.41e-03, grad_scale: 8.0 2023-03-08 20:50:26,644 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7443, 3.3140, 3.4519, 2.7595, 3.4291, 3.3771, 3.5250, 2.1271], device='cuda:0'), covar=tensor([0.1232, 0.2150, 0.2119, 0.6030, 0.2736, 0.3030, 0.1079, 0.7822], device='cuda:0'), in_proj_covar=tensor([0.0164, 0.0178, 0.0190, 0.0249, 0.0152, 0.0248, 0.0170, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:50:44,523 INFO [train2.py:809] (0/4) Epoch 20, batch 1700, loss[ctc_loss=0.09305, att_loss=0.2627, loss=0.2288, over 17056.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.007742, over 52.00 utterances.], tot_loss[ctc_loss=0.07579, att_loss=0.2367, loss=0.2045, over 3260572.95 frames. utt_duration=1273 frames, utt_pad_proportion=0.0513, over 10257.40 utterances.], batch size: 52, lr: 5.40e-03, grad_scale: 8.0 2023-03-08 20:51:30,372 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8854, 6.1859, 5.6418, 5.9050, 5.8448, 5.3279, 5.5569, 5.4164], device='cuda:0'), covar=tensor([0.1197, 0.0872, 0.0891, 0.0683, 0.0837, 0.1488, 0.2197, 0.2020], device='cuda:0'), in_proj_covar=tensor([0.0516, 0.0604, 0.0457, 0.0446, 0.0430, 0.0467, 0.0606, 0.0525], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 20:51:57,823 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.333e+02 1.985e+02 2.370e+02 2.830e+02 5.351e+02, threshold=4.740e+02, percent-clipped=1.0 2023-03-08 20:52:04,402 INFO [train2.py:809] (0/4) Epoch 20, batch 1750, loss[ctc_loss=0.09935, att_loss=0.2445, loss=0.2155, over 16541.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005616, over 45.00 utterances.], tot_loss[ctc_loss=0.07546, att_loss=0.2359, loss=0.2038, over 3264194.91 frames. utt_duration=1285 frames, utt_pad_proportion=0.04683, over 10176.58 utterances.], batch size: 45, lr: 5.40e-03, grad_scale: 8.0 2023-03-08 20:52:18,524 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77451.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 20:52:26,323 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0572, 5.2852, 5.6198, 5.4452, 5.5817, 6.0370, 5.1739, 6.1285], device='cuda:0'), covar=tensor([0.0688, 0.0733, 0.0746, 0.1196, 0.1575, 0.0831, 0.0648, 0.0609], device='cuda:0'), in_proj_covar=tensor([0.0867, 0.0504, 0.0588, 0.0649, 0.0861, 0.0611, 0.0482, 0.0597], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 20:53:24,303 INFO [train2.py:809] (0/4) Epoch 20, batch 1800, loss[ctc_loss=0.06222, att_loss=0.2225, loss=0.1905, over 16280.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007402, over 43.00 utterances.], tot_loss[ctc_loss=0.07598, att_loss=0.2359, loss=0.2039, over 3265296.31 frames. utt_duration=1277 frames, utt_pad_proportion=0.0476, over 10240.28 utterances.], batch size: 43, lr: 5.40e-03, grad_scale: 8.0 2023-03-08 20:53:29,103 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77495.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:54:37,624 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 1.944e+02 2.387e+02 2.834e+02 5.542e+02, threshold=4.773e+02, percent-clipped=3.0 2023-03-08 20:54:44,321 INFO [train2.py:809] (0/4) Epoch 20, batch 1850, loss[ctc_loss=0.07522, att_loss=0.2359, loss=0.2038, over 16260.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.008346, over 43.00 utterances.], tot_loss[ctc_loss=0.07557, att_loss=0.2362, loss=0.2041, over 3269812.06 frames. utt_duration=1301 frames, utt_pad_proportion=0.04106, over 10064.33 utterances.], batch size: 43, lr: 5.40e-03, grad_scale: 8.0 2023-03-08 20:54:45,963 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77543.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:55:23,530 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77567.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:55:43,005 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2053, 3.8963, 3.9037, 3.3887, 3.8557, 3.9488, 3.8907, 2.7562], device='cuda:0'), covar=tensor([0.0925, 0.1025, 0.1466, 0.3170, 0.2105, 0.1430, 0.0692, 0.3964], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0177, 0.0188, 0.0246, 0.0151, 0.0246, 0.0168, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:55:45,244 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2051, 3.8797, 3.8558, 3.3994, 3.8906, 3.9084, 3.8875, 2.9344], device='cuda:0'), covar=tensor([0.0914, 0.1080, 0.1782, 0.3624, 0.0976, 0.1945, 0.0790, 0.3857], device='cuda:0'), in_proj_covar=tensor([0.0162, 0.0177, 0.0188, 0.0246, 0.0151, 0.0246, 0.0168, 0.0212], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:55:48,467 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9049, 4.6552, 4.6771, 2.3238, 2.0460, 2.7525, 2.2785, 3.7563], device='cuda:0'), covar=tensor([0.0706, 0.0239, 0.0218, 0.4566, 0.5186, 0.2502, 0.3097, 0.1430], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0267, 0.0263, 0.0240, 0.0342, 0.0332, 0.0248, 0.0364], device='cuda:0'), out_proj_covar=tensor([1.4933e-04, 9.8921e-05, 1.1176e-04, 1.0348e-04, 1.4345e-04, 1.3001e-04, 9.9507e-05, 1.4776e-04], device='cuda:0') 2023-03-08 20:56:03,930 INFO [train2.py:809] (0/4) Epoch 20, batch 1900, loss[ctc_loss=0.07073, att_loss=0.2415, loss=0.2073, over 16902.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.005657, over 49.00 utterances.], tot_loss[ctc_loss=0.075, att_loss=0.2355, loss=0.2034, over 3267933.76 frames. utt_duration=1289 frames, utt_pad_proportion=0.04387, over 10152.43 utterances.], batch size: 49, lr: 5.40e-03, grad_scale: 8.0 2023-03-08 20:56:20,237 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5196, 4.5930, 4.6318, 4.6986, 5.2655, 4.5426, 4.6468, 2.6331], device='cuda:0'), covar=tensor([0.0230, 0.0294, 0.0292, 0.0257, 0.0767, 0.0223, 0.0274, 0.1740], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0179, 0.0180, 0.0196, 0.0365, 0.0150, 0.0168, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 20:56:42,714 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77616.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:57:03,832 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77629.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:57:10,132 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77633.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:57:17,662 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.149e+02 1.953e+02 2.465e+02 3.107e+02 2.003e+03, threshold=4.931e+02, percent-clipped=5.0 2023-03-08 20:57:24,504 INFO [train2.py:809] (0/4) Epoch 20, batch 1950, loss[ctc_loss=0.07907, att_loss=0.2538, loss=0.2188, over 17303.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01093, over 55.00 utterances.], tot_loss[ctc_loss=0.07526, att_loss=0.2356, loss=0.2036, over 3263504.51 frames. utt_duration=1270 frames, utt_pad_proportion=0.05067, over 10287.64 utterances.], batch size: 55, lr: 5.39e-03, grad_scale: 8.0 2023-03-08 20:57:55,437 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0229, 5.2906, 5.5420, 5.3922, 5.5021, 5.9688, 5.2890, 6.0521], device='cuda:0'), covar=tensor([0.0770, 0.0814, 0.0809, 0.1404, 0.1831, 0.0955, 0.0681, 0.0784], device='cuda:0'), in_proj_covar=tensor([0.0874, 0.0511, 0.0591, 0.0661, 0.0871, 0.0614, 0.0485, 0.0604], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 20:58:19,848 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:58:19,877 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77677.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 20:58:44,125 INFO [train2.py:809] (0/4) Epoch 20, batch 2000, loss[ctc_loss=0.07375, att_loss=0.2427, loss=0.2089, over 17041.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.008831, over 52.00 utterances.], tot_loss[ctc_loss=0.07465, att_loss=0.2357, loss=0.2035, over 3271072.30 frames. utt_duration=1285 frames, utt_pad_proportion=0.04664, over 10197.26 utterances.], batch size: 52, lr: 5.39e-03, grad_scale: 8.0 2023-03-08 20:59:22,163 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9436, 5.3084, 4.8569, 5.3284, 4.7373, 4.8760, 5.4254, 5.2086], device='cuda:0'), covar=tensor([0.0583, 0.0245, 0.0774, 0.0309, 0.0403, 0.0276, 0.0218, 0.0189], device='cuda:0'), in_proj_covar=tensor([0.0382, 0.0313, 0.0361, 0.0338, 0.0313, 0.0234, 0.0295, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 20:59:43,362 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6782, 4.6790, 4.4690, 2.7482, 4.4234, 4.3576, 4.0410, 2.9048], device='cuda:0'), covar=tensor([0.0111, 0.0118, 0.0248, 0.1028, 0.0125, 0.0247, 0.0300, 0.1228], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0101, 0.0101, 0.0110, 0.0083, 0.0109, 0.0098, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 20:59:57,578 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 2.073e+02 2.649e+02 3.244e+02 1.218e+03, threshold=5.298e+02, percent-clipped=5.0 2023-03-08 20:59:57,919 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=77738.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:00:03,837 INFO [train2.py:809] (0/4) Epoch 20, batch 2050, loss[ctc_loss=0.07206, att_loss=0.2439, loss=0.2095, over 17344.00 frames. utt_duration=879.8 frames, utt_pad_proportion=0.07876, over 79.00 utterances.], tot_loss[ctc_loss=0.07508, att_loss=0.2359, loss=0.2037, over 3260569.37 frames. utt_duration=1255 frames, utt_pad_proportion=0.05652, over 10401.08 utterances.], batch size: 79, lr: 5.39e-03, grad_scale: 8.0 2023-03-08 21:00:17,856 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77751.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 21:01:23,886 INFO [train2.py:809] (0/4) Epoch 20, batch 2100, loss[ctc_loss=0.08454, att_loss=0.2489, loss=0.216, over 17306.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01099, over 55.00 utterances.], tot_loss[ctc_loss=0.07563, att_loss=0.2365, loss=0.2043, over 3262082.24 frames. utt_duration=1244 frames, utt_pad_proportion=0.059, over 10501.59 utterances.], batch size: 55, lr: 5.39e-03, grad_scale: 8.0 2023-03-08 21:01:34,647 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77799.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:02:36,993 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.913e+02 2.227e+02 2.675e+02 5.227e+02, threshold=4.453e+02, percent-clipped=0.0 2023-03-08 21:02:43,921 INFO [train2.py:809] (0/4) Epoch 20, batch 2150, loss[ctc_loss=0.07878, att_loss=0.2226, loss=0.1938, over 14598.00 frames. utt_duration=1826 frames, utt_pad_proportion=0.04293, over 32.00 utterances.], tot_loss[ctc_loss=0.07537, att_loss=0.2358, loss=0.2038, over 3260603.63 frames. utt_duration=1268 frames, utt_pad_proportion=0.05432, over 10294.52 utterances.], batch size: 32, lr: 5.39e-03, grad_scale: 8.0 2023-03-08 21:03:24,191 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77867.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:03:53,609 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5883, 2.6942, 5.1571, 4.0285, 3.0304, 4.4181, 4.8828, 4.6740], device='cuda:0'), covar=tensor([0.0269, 0.1591, 0.0169, 0.0891, 0.1786, 0.0252, 0.0146, 0.0260], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0243, 0.0180, 0.0311, 0.0267, 0.0211, 0.0169, 0.0198], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:04:04,226 INFO [train2.py:809] (0/4) Epoch 20, batch 2200, loss[ctc_loss=0.07879, att_loss=0.2488, loss=0.2148, over 17322.00 frames. utt_duration=878.6 frames, utt_pad_proportion=0.07997, over 79.00 utterances.], tot_loss[ctc_loss=0.07589, att_loss=0.2367, loss=0.2046, over 3273986.58 frames. utt_duration=1254 frames, utt_pad_proportion=0.05332, over 10458.77 utterances.], batch size: 79, lr: 5.39e-03, grad_scale: 8.0 2023-03-08 21:04:21,042 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2023-03-08 21:04:32,552 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.78 vs. limit=5.0 2023-03-08 21:04:41,045 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77915.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:05:03,623 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77929.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:05:10,186 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=77933.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:05:17,372 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.298e+02 1.939e+02 2.291e+02 2.587e+02 5.643e+02, threshold=4.582e+02, percent-clipped=3.0 2023-03-08 21:05:23,446 INFO [train2.py:809] (0/4) Epoch 20, batch 2250, loss[ctc_loss=0.08065, att_loss=0.2445, loss=0.2117, over 16941.00 frames. utt_duration=686 frames, utt_pad_proportion=0.1339, over 99.00 utterances.], tot_loss[ctc_loss=0.07718, att_loss=0.2381, loss=0.2059, over 3272763.81 frames. utt_duration=1188 frames, utt_pad_proportion=0.07026, over 11034.54 utterances.], batch size: 99, lr: 5.38e-03, grad_scale: 8.0 2023-03-08 21:05:42,161 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-08 21:06:11,698 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=77972.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:06:19,554 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77977.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:06:26,421 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=77981.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:06:43,280 INFO [train2.py:809] (0/4) Epoch 20, batch 2300, loss[ctc_loss=0.07463, att_loss=0.2439, loss=0.21, over 17020.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.007831, over 51.00 utterances.], tot_loss[ctc_loss=0.07621, att_loss=0.2376, loss=0.2053, over 3277443.57 frames. utt_duration=1216 frames, utt_pad_proportion=0.06162, over 10793.58 utterances.], batch size: 51, lr: 5.38e-03, grad_scale: 8.0 2023-03-08 21:06:56,013 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-78000.pt 2023-03-08 21:07:53,468 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78033.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:07:58,291 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6531, 5.0853, 4.9193, 4.9792, 5.1020, 4.7831, 3.4544, 5.0759], device='cuda:0'), covar=tensor([0.0114, 0.0111, 0.0138, 0.0099, 0.0087, 0.0114, 0.0744, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0084, 0.0106, 0.0066, 0.0072, 0.0082, 0.0101, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:08:01,052 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.567e+02 2.189e+02 2.492e+02 2.915e+02 6.479e+02, threshold=4.984e+02, percent-clipped=5.0 2023-03-08 21:08:07,066 INFO [train2.py:809] (0/4) Epoch 20, batch 2350, loss[ctc_loss=0.07421, att_loss=0.2309, loss=0.1996, over 16539.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.004958, over 45.00 utterances.], tot_loss[ctc_loss=0.07691, att_loss=0.2383, loss=0.206, over 3281308.80 frames. utt_duration=1210 frames, utt_pad_proportion=0.06279, over 10860.52 utterances.], batch size: 45, lr: 5.38e-03, grad_scale: 8.0 2023-03-08 21:08:21,600 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6228, 4.9478, 4.8234, 4.8575, 5.0128, 4.6765, 3.5620, 4.9301], device='cuda:0'), covar=tensor([0.0108, 0.0115, 0.0114, 0.0088, 0.0087, 0.0107, 0.0679, 0.0201], device='cuda:0'), in_proj_covar=tensor([0.0089, 0.0084, 0.0106, 0.0066, 0.0072, 0.0082, 0.0101, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:09:26,270 INFO [train2.py:809] (0/4) Epoch 20, batch 2400, loss[ctc_loss=0.06391, att_loss=0.2246, loss=0.1925, over 15901.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.00835, over 39.00 utterances.], tot_loss[ctc_loss=0.07665, att_loss=0.2383, loss=0.2059, over 3283903.12 frames. utt_duration=1226 frames, utt_pad_proportion=0.05778, over 10728.84 utterances.], batch size: 39, lr: 5.38e-03, grad_scale: 8.0 2023-03-08 21:10:10,105 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2512, 5.1742, 4.9907, 3.2229, 4.9030, 4.8135, 4.6050, 3.1335], device='cuda:0'), covar=tensor([0.0123, 0.0094, 0.0272, 0.0911, 0.0106, 0.0188, 0.0261, 0.1224], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0101, 0.0102, 0.0110, 0.0084, 0.0110, 0.0098, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 21:10:39,631 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 2.004e+02 2.481e+02 3.189e+02 5.932e+02, threshold=4.963e+02, percent-clipped=2.0 2023-03-08 21:10:46,372 INFO [train2.py:809] (0/4) Epoch 20, batch 2450, loss[ctc_loss=0.05593, att_loss=0.2048, loss=0.175, over 15891.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008879, over 39.00 utterances.], tot_loss[ctc_loss=0.07636, att_loss=0.2373, loss=0.2051, over 3272289.47 frames. utt_duration=1214 frames, utt_pad_proportion=0.06378, over 10794.14 utterances.], batch size: 39, lr: 5.38e-03, grad_scale: 8.0 2023-03-08 21:12:06,741 INFO [train2.py:809] (0/4) Epoch 20, batch 2500, loss[ctc_loss=0.08208, att_loss=0.2503, loss=0.2167, over 17102.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01595, over 56.00 utterances.], tot_loss[ctc_loss=0.07658, att_loss=0.2381, loss=0.2058, over 3267914.66 frames. utt_duration=1202 frames, utt_pad_proportion=0.06822, over 10889.38 utterances.], batch size: 56, lr: 5.38e-03, grad_scale: 8.0 2023-03-08 21:12:37,558 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.58 vs. limit=5.0 2023-03-08 21:12:38,440 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9798, 4.9769, 4.8499, 2.1256, 1.8517, 2.8423, 2.4717, 3.8455], device='cuda:0'), covar=tensor([0.0745, 0.0269, 0.0231, 0.5331, 0.5942, 0.2617, 0.3408, 0.1624], device='cuda:0'), in_proj_covar=tensor([0.0352, 0.0271, 0.0263, 0.0242, 0.0341, 0.0335, 0.0251, 0.0365], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 21:13:12,112 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78232.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:13:21,002 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.184e+02 1.956e+02 2.237e+02 2.783e+02 5.149e+02, threshold=4.474e+02, percent-clipped=1.0 2023-03-08 21:13:26,986 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-03-08 21:13:27,749 INFO [train2.py:809] (0/4) Epoch 20, batch 2550, loss[ctc_loss=0.07974, att_loss=0.2363, loss=0.205, over 17057.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008003, over 52.00 utterances.], tot_loss[ctc_loss=0.07618, att_loss=0.2373, loss=0.2051, over 3266752.44 frames. utt_duration=1217 frames, utt_pad_proportion=0.06531, over 10751.73 utterances.], batch size: 52, lr: 5.37e-03, grad_scale: 8.0 2023-03-08 21:13:32,829 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4372, 2.6130, 5.0372, 3.9006, 2.9417, 4.2700, 4.8450, 4.5835], device='cuda:0'), covar=tensor([0.0293, 0.1712, 0.0161, 0.1012, 0.1806, 0.0275, 0.0206, 0.0285], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0243, 0.0179, 0.0311, 0.0266, 0.0211, 0.0170, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:13:32,863 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0482, 5.0569, 4.8910, 2.2702, 1.9336, 2.8822, 2.5137, 3.7996], device='cuda:0'), covar=tensor([0.0791, 0.0316, 0.0257, 0.5440, 0.5819, 0.2578, 0.3510, 0.1816], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0271, 0.0264, 0.0244, 0.0343, 0.0336, 0.0253, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 21:13:55,576 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.87 vs. limit=2.0 2023-03-08 21:14:15,800 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78272.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:14:23,991 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7551, 5.1241, 5.3559, 5.1201, 5.2952, 5.6967, 5.1405, 5.8105], device='cuda:0'), covar=tensor([0.0649, 0.0687, 0.0706, 0.1303, 0.1532, 0.0865, 0.0730, 0.0607], device='cuda:0'), in_proj_covar=tensor([0.0854, 0.0501, 0.0580, 0.0646, 0.0848, 0.0603, 0.0475, 0.0590], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:14:28,903 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2689, 3.7878, 3.3096, 3.4050, 4.0108, 3.6760, 3.1658, 4.3284], device='cuda:0'), covar=tensor([0.0915, 0.0572, 0.1055, 0.0728, 0.0708, 0.0737, 0.0840, 0.0515], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0214, 0.0226, 0.0199, 0.0275, 0.0239, 0.0200, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 21:14:47,849 INFO [train2.py:809] (0/4) Epoch 20, batch 2600, loss[ctc_loss=0.0724, att_loss=0.2133, loss=0.1851, over 15770.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008793, over 38.00 utterances.], tot_loss[ctc_loss=0.0753, att_loss=0.2365, loss=0.2043, over 3271898.88 frames. utt_duration=1232 frames, utt_pad_proportion=0.05909, over 10632.85 utterances.], batch size: 38, lr: 5.37e-03, grad_scale: 8.0 2023-03-08 21:14:49,840 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78293.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 21:15:32,440 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78320.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:15:53,574 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78333.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:16:00,906 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.091e+02 1.986e+02 2.327e+02 2.880e+02 6.602e+02, threshold=4.653e+02, percent-clipped=4.0 2023-03-08 21:16:04,257 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 21:16:07,831 INFO [train2.py:809] (0/4) Epoch 20, batch 2650, loss[ctc_loss=0.07228, att_loss=0.2385, loss=0.2052, over 17402.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.04589, over 69.00 utterances.], tot_loss[ctc_loss=0.07556, att_loss=0.2372, loss=0.2049, over 3281150.11 frames. utt_duration=1222 frames, utt_pad_proportion=0.05893, over 10752.58 utterances.], batch size: 69, lr: 5.37e-03, grad_scale: 8.0 2023-03-08 21:16:54,874 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1049, 4.5473, 4.6391, 4.7806, 2.9552, 4.7791, 2.9438, 1.9880], device='cuda:0'), covar=tensor([0.0469, 0.0230, 0.0599, 0.0193, 0.1426, 0.0158, 0.1248, 0.1558], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0159, 0.0261, 0.0154, 0.0220, 0.0139, 0.0230, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:17:10,076 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78381.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:17:23,910 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-03-08 21:17:27,553 INFO [train2.py:809] (0/4) Epoch 20, batch 2700, loss[ctc_loss=0.0696, att_loss=0.2133, loss=0.1846, over 14591.00 frames. utt_duration=1825 frames, utt_pad_proportion=0.04131, over 32.00 utterances.], tot_loss[ctc_loss=0.07563, att_loss=0.2364, loss=0.2043, over 3271838.82 frames. utt_duration=1216 frames, utt_pad_proportion=0.06127, over 10775.32 utterances.], batch size: 32, lr: 5.37e-03, grad_scale: 8.0 2023-03-08 21:17:30,033 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.93 vs. limit=2.0 2023-03-08 21:17:55,232 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3818, 4.3417, 4.3653, 4.4433, 4.8915, 4.2879, 4.2956, 2.3084], device='cuda:0'), covar=tensor([0.0239, 0.0323, 0.0324, 0.0233, 0.0646, 0.0259, 0.0350, 0.1951], device='cuda:0'), in_proj_covar=tensor([0.0157, 0.0179, 0.0181, 0.0198, 0.0363, 0.0151, 0.0170, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:18:24,699 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.66 vs. limit=5.0 2023-03-08 21:18:41,449 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.385e+02 1.853e+02 2.270e+02 2.670e+02 9.199e+02, threshold=4.540e+02, percent-clipped=5.0 2023-03-08 21:18:46,823 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7832, 2.4817, 3.8664, 3.5978, 2.9146, 3.7051, 3.7694, 3.8220], device='cuda:0'), covar=tensor([0.0228, 0.1217, 0.0158, 0.0652, 0.1229, 0.0246, 0.0197, 0.0245], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0239, 0.0177, 0.0307, 0.0262, 0.0209, 0.0167, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:18:47,882 INFO [train2.py:809] (0/4) Epoch 20, batch 2750, loss[ctc_loss=0.11, att_loss=0.2547, loss=0.2258, over 14335.00 frames. utt_duration=397 frames, utt_pad_proportion=0.3096, over 145.00 utterances.], tot_loss[ctc_loss=0.07497, att_loss=0.2357, loss=0.2036, over 3261654.25 frames. utt_duration=1214 frames, utt_pad_proportion=0.0643, over 10758.86 utterances.], batch size: 145, lr: 5.37e-03, grad_scale: 8.0 2023-03-08 21:19:17,726 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78460.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:20:08,380 INFO [train2.py:809] (0/4) Epoch 20, batch 2800, loss[ctc_loss=0.06835, att_loss=0.2308, loss=0.1983, over 15944.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007633, over 41.00 utterances.], tot_loss[ctc_loss=0.07521, att_loss=0.236, loss=0.2038, over 3261698.76 frames. utt_duration=1219 frames, utt_pad_proportion=0.06467, over 10712.48 utterances.], batch size: 41, lr: 5.37e-03, grad_scale: 8.0 2023-03-08 21:20:16,249 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.20 vs. limit=2.0 2023-03-08 21:20:41,122 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0939, 3.7562, 3.7654, 3.2718, 3.7762, 3.8622, 3.8059, 2.7906], device='cuda:0'), covar=tensor([0.1025, 0.1074, 0.1715, 0.3380, 0.0759, 0.2074, 0.0767, 0.4161], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0184, 0.0195, 0.0251, 0.0155, 0.0256, 0.0173, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:20:55,770 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78521.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:21:08,247 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4471, 2.5471, 4.9107, 3.8201, 2.9089, 4.0601, 4.7291, 4.5313], device='cuda:0'), covar=tensor([0.0247, 0.1716, 0.0159, 0.0946, 0.1803, 0.0301, 0.0147, 0.0242], device='cuda:0'), in_proj_covar=tensor([0.0187, 0.0241, 0.0179, 0.0309, 0.0264, 0.0211, 0.0169, 0.0197], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:21:21,797 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.455e+02 2.103e+02 2.542e+02 3.045e+02 6.490e+02, threshold=5.083e+02, percent-clipped=4.0 2023-03-08 21:21:28,201 INFO [train2.py:809] (0/4) Epoch 20, batch 2850, loss[ctc_loss=0.07928, att_loss=0.2404, loss=0.2082, over 17322.00 frames. utt_duration=1006 frames, utt_pad_proportion=0.05137, over 69.00 utterances.], tot_loss[ctc_loss=0.07667, att_loss=0.2366, loss=0.2046, over 3246630.34 frames. utt_duration=1178 frames, utt_pad_proportion=0.07989, over 11041.15 utterances.], batch size: 69, lr: 5.36e-03, grad_scale: 16.0 2023-03-08 21:21:51,202 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78556.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:22:42,419 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78588.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 21:22:48,499 INFO [train2.py:809] (0/4) Epoch 20, batch 2900, loss[ctc_loss=0.09613, att_loss=0.2493, loss=0.2187, over 16316.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006901, over 45.00 utterances.], tot_loss[ctc_loss=0.07675, att_loss=0.2367, loss=0.2047, over 3252528.92 frames. utt_duration=1182 frames, utt_pad_proportion=0.07627, over 11023.12 utterances.], batch size: 45, lr: 5.36e-03, grad_scale: 16.0 2023-03-08 21:22:53,144 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.91 vs. limit=5.0 2023-03-08 21:23:26,673 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.85 vs. limit=5.0 2023-03-08 21:23:30,340 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=78617.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:24:02,855 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 1.965e+02 2.501e+02 3.006e+02 5.766e+02, threshold=5.002e+02, percent-clipped=1.0 2023-03-08 21:24:08,862 INFO [train2.py:809] (0/4) Epoch 20, batch 2950, loss[ctc_loss=0.0661, att_loss=0.2076, loss=0.1793, over 15502.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008578, over 36.00 utterances.], tot_loss[ctc_loss=0.07669, att_loss=0.2373, loss=0.2052, over 3267658.25 frames. utt_duration=1213 frames, utt_pad_proportion=0.06471, over 10792.90 utterances.], batch size: 36, lr: 5.36e-03, grad_scale: 16.0 2023-03-08 21:24:17,212 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8696, 5.1868, 5.4386, 5.3132, 5.3191, 5.8228, 5.1193, 5.8923], device='cuda:0'), covar=tensor([0.0815, 0.0811, 0.0810, 0.1238, 0.1839, 0.0906, 0.0800, 0.0769], device='cuda:0'), in_proj_covar=tensor([0.0869, 0.0508, 0.0592, 0.0656, 0.0866, 0.0610, 0.0481, 0.0601], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:24:45,047 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1399, 5.1491, 4.8655, 2.3193, 1.9863, 2.8250, 2.6720, 3.8254], device='cuda:0'), covar=tensor([0.0646, 0.0292, 0.0304, 0.5216, 0.5465, 0.2543, 0.3255, 0.1702], device='cuda:0'), in_proj_covar=tensor([0.0350, 0.0268, 0.0263, 0.0241, 0.0341, 0.0332, 0.0250, 0.0363], device='cuda:0'), out_proj_covar=tensor([1.4863e-04, 9.9626e-05, 1.1182e-04, 1.0349e-04, 1.4295e-04, 1.2996e-04, 1.0049e-04, 1.4743e-04], device='cuda:0') 2023-03-08 21:25:28,466 INFO [train2.py:809] (0/4) Epoch 20, batch 3000, loss[ctc_loss=0.07606, att_loss=0.2457, loss=0.2118, over 17321.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02176, over 59.00 utterances.], tot_loss[ctc_loss=0.07646, att_loss=0.2371, loss=0.205, over 3275346.13 frames. utt_duration=1229 frames, utt_pad_proportion=0.05913, over 10674.31 utterances.], batch size: 59, lr: 5.36e-03, grad_scale: 16.0 2023-03-08 21:25:28,469 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 21:25:42,048 INFO [train2.py:843] (0/4) Epoch 20, validation: ctc_loss=0.04025, att_loss=0.2341, loss=0.1953, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 21:25:42,049 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 21:26:41,144 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-08 21:26:46,749 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-03-08 21:26:55,740 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 1.875e+02 2.186e+02 2.641e+02 4.823e+02, threshold=4.373e+02, percent-clipped=0.0 2023-03-08 21:27:02,052 INFO [train2.py:809] (0/4) Epoch 20, batch 3050, loss[ctc_loss=0.09742, att_loss=0.254, loss=0.2227, over 17035.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009797, over 52.00 utterances.], tot_loss[ctc_loss=0.07672, att_loss=0.2373, loss=0.2052, over 3272777.92 frames. utt_duration=1244 frames, utt_pad_proportion=0.05522, over 10538.62 utterances.], batch size: 52, lr: 5.36e-03, grad_scale: 16.0 2023-03-08 21:27:33,396 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8136, 5.1236, 5.3611, 5.2272, 5.3028, 5.7559, 5.1289, 5.8768], device='cuda:0'), covar=tensor([0.0722, 0.0765, 0.0850, 0.1265, 0.1660, 0.0932, 0.0778, 0.0778], device='cuda:0'), in_proj_covar=tensor([0.0868, 0.0507, 0.0590, 0.0657, 0.0869, 0.0610, 0.0484, 0.0603], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:27:51,177 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7796, 4.5319, 4.6466, 4.7178, 5.2208, 4.7448, 4.6120, 2.6011], device='cuda:0'), covar=tensor([0.0190, 0.0331, 0.0294, 0.0264, 0.0680, 0.0189, 0.0288, 0.1861], device='cuda:0'), in_proj_covar=tensor([0.0161, 0.0185, 0.0185, 0.0202, 0.0370, 0.0154, 0.0173, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:28:22,265 INFO [train2.py:809] (0/4) Epoch 20, batch 3100, loss[ctc_loss=0.09642, att_loss=0.2515, loss=0.2205, over 17043.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.01024, over 53.00 utterances.], tot_loss[ctc_loss=0.07653, att_loss=0.2381, loss=0.2058, over 3282815.13 frames. utt_duration=1244 frames, utt_pad_proportion=0.05343, over 10567.63 utterances.], batch size: 53, lr: 5.36e-03, grad_scale: 16.0 2023-03-08 21:29:01,225 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78816.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:29:15,666 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7205, 5.9835, 5.4471, 5.7369, 5.7035, 5.0603, 5.3130, 5.1189], device='cuda:0'), covar=tensor([0.1275, 0.0869, 0.0831, 0.0784, 0.0837, 0.1448, 0.2331, 0.2397], device='cuda:0'), in_proj_covar=tensor([0.0512, 0.0597, 0.0451, 0.0450, 0.0423, 0.0455, 0.0607, 0.0520], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 21:29:36,348 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 1.977e+02 2.476e+02 2.944e+02 6.319e+02, threshold=4.952e+02, percent-clipped=6.0 2023-03-08 21:29:42,600 INFO [train2.py:809] (0/4) Epoch 20, batch 3150, loss[ctc_loss=0.07677, att_loss=0.2345, loss=0.203, over 15959.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006793, over 41.00 utterances.], tot_loss[ctc_loss=0.07607, att_loss=0.2377, loss=0.2053, over 3279343.83 frames. utt_duration=1257 frames, utt_pad_proportion=0.05097, over 10450.37 utterances.], batch size: 41, lr: 5.35e-03, grad_scale: 16.0 2023-03-08 21:29:59,065 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8427, 3.5124, 3.5366, 3.1232, 3.5349, 3.6357, 3.5870, 2.5073], device='cuda:0'), covar=tensor([0.1119, 0.1490, 0.2139, 0.3703, 0.1057, 0.3565, 0.1087, 0.5310], device='cuda:0'), in_proj_covar=tensor([0.0167, 0.0182, 0.0193, 0.0248, 0.0154, 0.0254, 0.0172, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:30:55,855 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=78888.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:31:01,821 INFO [train2.py:809] (0/4) Epoch 20, batch 3200, loss[ctc_loss=0.06238, att_loss=0.222, loss=0.1901, over 16119.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006191, over 42.00 utterances.], tot_loss[ctc_loss=0.0764, att_loss=0.238, loss=0.2057, over 3284429.48 frames. utt_duration=1250 frames, utt_pad_proportion=0.05158, over 10519.85 utterances.], batch size: 42, lr: 5.35e-03, grad_scale: 16.0 2023-03-08 21:31:34,107 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6676, 5.0751, 5.2481, 5.0673, 5.1380, 5.5700, 5.0573, 5.6868], device='cuda:0'), covar=tensor([0.0722, 0.0660, 0.0804, 0.1143, 0.1685, 0.0992, 0.0765, 0.0664], device='cuda:0'), in_proj_covar=tensor([0.0860, 0.0504, 0.0588, 0.0654, 0.0861, 0.0608, 0.0481, 0.0599], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:31:34,122 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=78912.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:32:12,154 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=78936.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:32:15,121 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 2.045e+02 2.242e+02 2.704e+02 5.175e+02, threshold=4.484e+02, percent-clipped=1.0 2023-03-08 21:32:17,041 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5076, 2.8634, 3.6206, 2.8445, 3.5752, 4.6112, 4.4629, 3.0938], device='cuda:0'), covar=tensor([0.0343, 0.1823, 0.1253, 0.1495, 0.1129, 0.0797, 0.0547, 0.1419], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0245, 0.0281, 0.0219, 0.0267, 0.0363, 0.0260, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:32:21,323 INFO [train2.py:809] (0/4) Epoch 20, batch 3250, loss[ctc_loss=0.06939, att_loss=0.2468, loss=0.2113, over 17027.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.00738, over 51.00 utterances.], tot_loss[ctc_loss=0.07593, att_loss=0.2372, loss=0.205, over 3283405.16 frames. utt_duration=1254 frames, utt_pad_proportion=0.05172, over 10489.12 utterances.], batch size: 51, lr: 5.35e-03, grad_scale: 16.0 2023-03-08 21:32:58,039 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=78965.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:33:40,325 INFO [train2.py:809] (0/4) Epoch 20, batch 3300, loss[ctc_loss=0.08459, att_loss=0.2428, loss=0.2112, over 16398.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007964, over 44.00 utterances.], tot_loss[ctc_loss=0.07507, att_loss=0.2363, loss=0.2041, over 3280970.12 frames. utt_duration=1274 frames, utt_pad_proportion=0.04778, over 10311.73 utterances.], batch size: 44, lr: 5.35e-03, grad_scale: 16.0 2023-03-08 21:34:35,245 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79026.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:34:54,084 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.427e+02 1.921e+02 2.242e+02 2.810e+02 6.510e+02, threshold=4.483e+02, percent-clipped=2.0 2023-03-08 21:35:00,358 INFO [train2.py:809] (0/4) Epoch 20, batch 3350, loss[ctc_loss=0.06293, att_loss=0.217, loss=0.1862, over 15892.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.008926, over 39.00 utterances.], tot_loss[ctc_loss=0.07498, att_loss=0.2363, loss=0.204, over 3276097.86 frames. utt_duration=1267 frames, utt_pad_proportion=0.05034, over 10359.03 utterances.], batch size: 39, lr: 5.35e-03, grad_scale: 16.0 2023-03-08 21:35:03,016 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.02 vs. limit=5.0 2023-03-08 21:36:21,444 INFO [train2.py:809] (0/4) Epoch 20, batch 3400, loss[ctc_loss=0.131, att_loss=0.2656, loss=0.2387, over 14976.00 frames. utt_duration=414.6 frames, utt_pad_proportion=0.2789, over 145.00 utterances.], tot_loss[ctc_loss=0.07548, att_loss=0.2368, loss=0.2046, over 3277029.13 frames. utt_duration=1236 frames, utt_pad_proportion=0.05695, over 10620.18 utterances.], batch size: 145, lr: 5.35e-03, grad_scale: 16.0 2023-03-08 21:36:59,670 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79116.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:37:33,667 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9299, 5.3429, 5.2053, 5.2831, 5.3511, 4.9637, 3.7153, 5.3000], device='cuda:0'), covar=tensor([0.0104, 0.0091, 0.0096, 0.0067, 0.0080, 0.0106, 0.0601, 0.0165], device='cuda:0'), in_proj_covar=tensor([0.0090, 0.0085, 0.0108, 0.0068, 0.0073, 0.0083, 0.0101, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:37:34,906 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 1.976e+02 2.387e+02 2.813e+02 1.211e+03, threshold=4.774e+02, percent-clipped=8.0 2023-03-08 21:37:41,716 INFO [train2.py:809] (0/4) Epoch 20, batch 3450, loss[ctc_loss=0.07847, att_loss=0.23, loss=0.1997, over 16001.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.008328, over 40.00 utterances.], tot_loss[ctc_loss=0.07516, att_loss=0.2367, loss=0.2044, over 3274074.69 frames. utt_duration=1243 frames, utt_pad_proportion=0.05497, over 10546.11 utterances.], batch size: 40, lr: 5.34e-03, grad_scale: 16.0 2023-03-08 21:38:16,621 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79164.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:39:01,281 INFO [train2.py:809] (0/4) Epoch 20, batch 3500, loss[ctc_loss=0.08056, att_loss=0.2501, loss=0.2162, over 17243.00 frames. utt_duration=1171 frames, utt_pad_proportion=0.02779, over 59.00 utterances.], tot_loss[ctc_loss=0.07524, att_loss=0.236, loss=0.2039, over 3269851.60 frames. utt_duration=1260 frames, utt_pad_proportion=0.05229, over 10393.95 utterances.], batch size: 59, lr: 5.34e-03, grad_scale: 16.0 2023-03-08 21:39:33,862 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79212.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:40:15,349 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.171e+02 1.933e+02 2.262e+02 2.788e+02 9.048e+02, threshold=4.523e+02, percent-clipped=2.0 2023-03-08 21:40:18,056 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5186, 4.5565, 4.6356, 4.6726, 5.1044, 4.6615, 4.5809, 2.6902], device='cuda:0'), covar=tensor([0.0241, 0.0318, 0.0306, 0.0327, 0.0824, 0.0191, 0.0305, 0.1660], device='cuda:0'), in_proj_covar=tensor([0.0159, 0.0183, 0.0185, 0.0201, 0.0366, 0.0153, 0.0172, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 21:40:22,326 INFO [train2.py:809] (0/4) Epoch 20, batch 3550, loss[ctc_loss=0.08037, att_loss=0.2298, loss=0.1999, over 16296.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006503, over 43.00 utterances.], tot_loss[ctc_loss=0.07483, att_loss=0.2363, loss=0.204, over 3277218.87 frames. utt_duration=1286 frames, utt_pad_proportion=0.04497, over 10207.79 utterances.], batch size: 43, lr: 5.34e-03, grad_scale: 16.0 2023-03-08 21:40:27,623 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.89 vs. limit=2.0 2023-03-08 21:40:51,170 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79260.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:41:31,913 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79285.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 21:41:42,772 INFO [train2.py:809] (0/4) Epoch 20, batch 3600, loss[ctc_loss=0.1402, att_loss=0.2764, loss=0.2492, over 14399.00 frames. utt_duration=396 frames, utt_pad_proportion=0.3101, over 146.00 utterances.], tot_loss[ctc_loss=0.07479, att_loss=0.2362, loss=0.2039, over 3274920.09 frames. utt_duration=1285 frames, utt_pad_proportion=0.04484, over 10207.73 utterances.], batch size: 146, lr: 5.34e-03, grad_scale: 16.0 2023-03-08 21:42:29,072 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79321.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:42:48,899 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79333.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:42:56,059 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.246e+02 1.889e+02 2.390e+02 3.056e+02 6.577e+02, threshold=4.779e+02, percent-clipped=3.0 2023-03-08 21:43:03,008 INFO [train2.py:809] (0/4) Epoch 20, batch 3650, loss[ctc_loss=0.06813, att_loss=0.24, loss=0.2056, over 16772.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006141, over 48.00 utterances.], tot_loss[ctc_loss=0.07517, att_loss=0.237, loss=0.2047, over 3284060.94 frames. utt_duration=1282 frames, utt_pad_proportion=0.04338, over 10256.72 utterances.], batch size: 48, lr: 5.34e-03, grad_scale: 16.0 2023-03-08 21:43:09,365 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79346.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 21:43:36,957 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79364.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:43:54,660 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79375.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 21:44:11,897 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1291, 5.4569, 5.3515, 5.3402, 5.3999, 5.4064, 5.1461, 4.8788], device='cuda:0'), covar=tensor([0.0968, 0.0461, 0.0285, 0.0525, 0.0298, 0.0300, 0.0322, 0.0335], device='cuda:0'), in_proj_covar=tensor([0.0524, 0.0355, 0.0343, 0.0353, 0.0415, 0.0428, 0.0353, 0.0393], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 21:44:21,500 INFO [train2.py:809] (0/4) Epoch 20, batch 3700, loss[ctc_loss=0.07548, att_loss=0.2257, loss=0.1956, over 15951.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006595, over 41.00 utterances.], tot_loss[ctc_loss=0.07565, att_loss=0.2374, loss=0.205, over 3286233.96 frames. utt_duration=1268 frames, utt_pad_proportion=0.04622, over 10378.68 utterances.], batch size: 41, lr: 5.34e-03, grad_scale: 16.0 2023-03-08 21:44:24,954 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79394.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:45:14,862 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79425.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:45:25,748 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8772, 5.2510, 4.9932, 5.0101, 5.2163, 4.8638, 3.6509, 5.1958], device='cuda:0'), covar=tensor([0.0098, 0.0098, 0.0124, 0.0100, 0.0088, 0.0119, 0.0666, 0.0164], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0086, 0.0109, 0.0069, 0.0074, 0.0085, 0.0103, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:45:32,652 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79436.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 21:45:35,343 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 1.959e+02 2.392e+02 2.997e+02 5.823e+02, threshold=4.784e+02, percent-clipped=2.0 2023-03-08 21:45:41,552 INFO [train2.py:809] (0/4) Epoch 20, batch 3750, loss[ctc_loss=0.05601, att_loss=0.2338, loss=0.1982, over 16340.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005594, over 45.00 utterances.], tot_loss[ctc_loss=0.07678, att_loss=0.2377, loss=0.2055, over 3267675.79 frames. utt_duration=1221 frames, utt_pad_proportion=0.06329, over 10715.89 utterances.], batch size: 45, lr: 5.33e-03, grad_scale: 16.0 2023-03-08 21:46:34,966 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2023-03-08 21:46:47,678 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9734, 4.9501, 4.7236, 2.8730, 4.6308, 4.5583, 4.2307, 2.7314], device='cuda:0'), covar=tensor([0.0083, 0.0101, 0.0262, 0.0943, 0.0109, 0.0217, 0.0279, 0.1319], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0101, 0.0101, 0.0109, 0.0083, 0.0110, 0.0098, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 21:47:01,107 INFO [train2.py:809] (0/4) Epoch 20, batch 3800, loss[ctc_loss=0.06781, att_loss=0.2246, loss=0.1933, over 15776.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008259, over 38.00 utterances.], tot_loss[ctc_loss=0.07705, att_loss=0.2373, loss=0.2053, over 3268094.57 frames. utt_duration=1209 frames, utt_pad_proportion=0.06571, over 10827.31 utterances.], batch size: 38, lr: 5.33e-03, grad_scale: 16.0 2023-03-08 21:47:10,604 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.90 vs. limit=2.0 2023-03-08 21:48:09,310 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1035, 5.3597, 5.6733, 5.4295, 5.5430, 6.0434, 5.3364, 6.0803], device='cuda:0'), covar=tensor([0.0698, 0.0785, 0.0753, 0.1386, 0.1836, 0.0861, 0.0648, 0.0714], device='cuda:0'), in_proj_covar=tensor([0.0852, 0.0501, 0.0581, 0.0653, 0.0855, 0.0605, 0.0480, 0.0594], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:48:11,182 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8821, 4.9590, 4.8093, 2.3047, 1.9393, 2.7800, 2.3867, 3.8319], device='cuda:0'), covar=tensor([0.0828, 0.0260, 0.0226, 0.4997, 0.5824, 0.2756, 0.3494, 0.1640], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0268, 0.0263, 0.0242, 0.0343, 0.0333, 0.0251, 0.0365], device='cuda:0'), out_proj_covar=tensor([1.4892e-04, 9.9243e-05, 1.1205e-04, 1.0303e-04, 1.4355e-04, 1.3034e-04, 1.0063e-04, 1.4808e-04], device='cuda:0') 2023-03-08 21:48:15,598 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 1.856e+02 2.318e+02 2.933e+02 8.070e+02, threshold=4.637e+02, percent-clipped=4.0 2023-03-08 21:48:21,941 INFO [train2.py:809] (0/4) Epoch 20, batch 3850, loss[ctc_loss=0.05868, att_loss=0.2219, loss=0.1893, over 16272.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007874, over 43.00 utterances.], tot_loss[ctc_loss=0.07634, att_loss=0.2373, loss=0.2051, over 3275283.84 frames. utt_duration=1224 frames, utt_pad_proportion=0.06025, over 10715.63 utterances.], batch size: 43, lr: 5.33e-03, grad_scale: 16.0 2023-03-08 21:49:18,051 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2362, 5.4809, 5.5094, 5.4228, 5.5438, 5.4983, 5.2658, 5.0193], device='cuda:0'), covar=tensor([0.0898, 0.0513, 0.0241, 0.0474, 0.0249, 0.0253, 0.0318, 0.0272], device='cuda:0'), in_proj_covar=tensor([0.0525, 0.0357, 0.0342, 0.0351, 0.0412, 0.0426, 0.0351, 0.0390], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 21:49:29,019 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79585.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:49:39,445 INFO [train2.py:809] (0/4) Epoch 20, batch 3900, loss[ctc_loss=0.06228, att_loss=0.2046, loss=0.1761, over 15766.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009013, over 38.00 utterances.], tot_loss[ctc_loss=0.07575, att_loss=0.2363, loss=0.2042, over 3269808.77 frames. utt_duration=1248 frames, utt_pad_proportion=0.05474, over 10496.65 utterances.], batch size: 38, lr: 5.33e-03, grad_scale: 16.0 2023-03-08 21:49:52,728 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-03-08 21:50:23,801 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79621.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:50:50,108 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 1.965e+02 2.336e+02 2.819e+02 5.688e+02, threshold=4.673e+02, percent-clipped=3.0 2023-03-08 21:50:54,905 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79641.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 21:50:56,229 INFO [train2.py:809] (0/4) Epoch 20, batch 3950, loss[ctc_loss=0.08458, att_loss=0.2409, loss=0.2096, over 16547.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.00497, over 45.00 utterances.], tot_loss[ctc_loss=0.07534, att_loss=0.2362, loss=0.204, over 3275409.31 frames. utt_duration=1261 frames, utt_pad_proportion=0.05031, over 10401.79 utterances.], batch size: 45, lr: 5.33e-03, grad_scale: 16.0 2023-03-08 21:51:02,703 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79646.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:51:19,567 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79657.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:51:38,137 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79669.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:51:44,420 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2918, 2.4804, 3.0635, 2.5393, 2.9859, 3.4442, 3.3723, 2.6491], device='cuda:0'), covar=tensor([0.0488, 0.1588, 0.1172, 0.1248, 0.1048, 0.1209, 0.0813, 0.1314], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0245, 0.0280, 0.0220, 0.0267, 0.0364, 0.0261, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 21:51:48,327 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-20.pt 2023-03-08 21:52:14,303 INFO [train2.py:809] (0/4) Epoch 21, batch 0, loss[ctc_loss=0.0573, att_loss=0.2095, loss=0.1791, over 15374.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01086, over 35.00 utterances.], tot_loss[ctc_loss=0.0573, att_loss=0.2095, loss=0.1791, over 15374.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01086, over 35.00 utterances.], batch size: 35, lr: 5.20e-03, grad_scale: 16.0 2023-03-08 21:52:14,305 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 21:52:26,405 INFO [train2.py:843] (0/4) Epoch 21, validation: ctc_loss=0.04229, att_loss=0.2351, loss=0.1965, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 21:52:26,406 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 21:52:29,128 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-08 21:52:47,079 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79689.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:53:33,479 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79718.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:53:36,487 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79720.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:53:46,007 INFO [train2.py:809] (0/4) Epoch 21, batch 50, loss[ctc_loss=0.1067, att_loss=0.2547, loss=0.2251, over 17421.00 frames. utt_duration=883.4 frames, utt_pad_proportion=0.07304, over 79.00 utterances.], tot_loss[ctc_loss=0.07334, att_loss=0.235, loss=0.2027, over 732984.77 frames. utt_duration=1297 frames, utt_pad_proportion=0.04311, over 2263.79 utterances.], batch size: 79, lr: 5.19e-03, grad_scale: 16.0 2023-03-08 21:53:54,347 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79731.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 21:54:05,022 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 1.975e+02 2.497e+02 3.178e+02 8.121e+02, threshold=4.994e+02, percent-clipped=4.0 2023-03-08 21:54:43,417 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=79762.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:55:05,925 INFO [train2.py:809] (0/4) Epoch 21, batch 100, loss[ctc_loss=0.1183, att_loss=0.2671, loss=0.2374, over 17289.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02447, over 59.00 utterances.], tot_loss[ctc_loss=0.07609, att_loss=0.235, loss=0.2032, over 1287754.75 frames. utt_duration=1268 frames, utt_pad_proportion=0.05497, over 4068.14 utterances.], batch size: 59, lr: 5.19e-03, grad_scale: 16.0 2023-03-08 21:56:04,049 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-08 21:56:22,726 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=79823.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:56:26,918 INFO [train2.py:809] (0/4) Epoch 21, batch 150, loss[ctc_loss=0.05536, att_loss=0.2081, loss=0.1775, over 15954.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006424, over 41.00 utterances.], tot_loss[ctc_loss=0.07552, att_loss=0.2357, loss=0.2037, over 1731621.73 frames. utt_duration=1266 frames, utt_pad_proportion=0.05037, over 5479.24 utterances.], batch size: 41, lr: 5.19e-03, grad_scale: 16.0 2023-03-08 21:56:46,155 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.523e+02 1.968e+02 2.419e+02 2.959e+02 7.390e+02, threshold=4.838e+02, percent-clipped=3.0 2023-03-08 21:57:47,562 INFO [train2.py:809] (0/4) Epoch 21, batch 200, loss[ctc_loss=0.06843, att_loss=0.2203, loss=0.19, over 16119.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.00596, over 42.00 utterances.], tot_loss[ctc_loss=0.0754, att_loss=0.236, loss=0.2039, over 2070944.81 frames. utt_duration=1228 frames, utt_pad_proportion=0.06102, over 6754.90 utterances.], batch size: 42, lr: 5.19e-03, grad_scale: 16.0 2023-03-08 21:58:21,602 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0083, 5.1240, 4.8865, 2.4847, 1.9619, 2.7110, 2.4314, 3.8987], device='cuda:0'), covar=tensor([0.0765, 0.0265, 0.0261, 0.4310, 0.5829, 0.2911, 0.3623, 0.1623], device='cuda:0'), in_proj_covar=tensor([0.0349, 0.0266, 0.0262, 0.0240, 0.0340, 0.0330, 0.0250, 0.0362], device='cuda:0'), out_proj_covar=tensor([1.4810e-04, 9.8582e-05, 1.1178e-04, 1.0273e-04, 1.4243e-04, 1.2917e-04, 1.0017e-04, 1.4704e-04], device='cuda:0') 2023-03-08 21:59:08,381 INFO [train2.py:809] (0/4) Epoch 21, batch 250, loss[ctc_loss=0.08709, att_loss=0.2463, loss=0.2144, over 17462.00 frames. utt_duration=1014 frames, utt_pad_proportion=0.04351, over 69.00 utterances.], tot_loss[ctc_loss=0.07586, att_loss=0.2369, loss=0.2047, over 2346015.15 frames. utt_duration=1215 frames, utt_pad_proportion=0.05989, over 7734.19 utterances.], batch size: 69, lr: 5.19e-03, grad_scale: 16.0 2023-03-08 21:59:28,152 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.404e+02 2.056e+02 2.417e+02 2.822e+02 7.152e+02, threshold=4.834e+02, percent-clipped=3.0 2023-03-08 21:59:33,212 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=79941.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 21:59:33,316 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79941.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 22:00:29,281 INFO [train2.py:809] (0/4) Epoch 21, batch 300, loss[ctc_loss=0.06092, att_loss=0.2163, loss=0.1852, over 16142.00 frames. utt_duration=1539 frames, utt_pad_proportion=0.005156, over 42.00 utterances.], tot_loss[ctc_loss=0.07532, att_loss=0.2363, loss=0.2041, over 2555680.51 frames. utt_duration=1238 frames, utt_pad_proportion=0.05195, over 8264.33 utterances.], batch size: 42, lr: 5.19e-03, grad_scale: 16.0 2023-03-08 22:00:49,776 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=79989.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 22:00:49,901 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=79989.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:01:06,876 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-08 22:01:07,987 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-80000.pt 2023-03-08 22:01:32,379 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80013.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:01:43,952 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80020.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:01:53,322 INFO [train2.py:809] (0/4) Epoch 21, batch 350, loss[ctc_loss=0.06577, att_loss=0.2254, loss=0.1935, over 16487.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.004885, over 46.00 utterances.], tot_loss[ctc_loss=0.075, att_loss=0.236, loss=0.2038, over 2716993.98 frames. utt_duration=1263 frames, utt_pad_proportion=0.04689, over 8613.15 utterances.], batch size: 46, lr: 5.18e-03, grad_scale: 16.0 2023-03-08 22:02:02,043 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80031.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 22:02:11,085 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80037.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:02:12,387 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 1.969e+02 2.385e+02 2.918e+02 9.167e+02, threshold=4.769e+02, percent-clipped=4.0 2023-03-08 22:03:01,883 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80068.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:03:14,462 INFO [train2.py:809] (0/4) Epoch 21, batch 400, loss[ctc_loss=0.05448, att_loss=0.2167, loss=0.1843, over 15500.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008658, over 36.00 utterances.], tot_loss[ctc_loss=0.0743, att_loss=0.2351, loss=0.2029, over 2833424.88 frames. utt_duration=1276 frames, utt_pad_proportion=0.04764, over 8894.17 utterances.], batch size: 36, lr: 5.18e-03, grad_scale: 16.0 2023-03-08 22:03:19,731 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80079.0, num_to_drop=1, layers_to_drop={1} 2023-03-08 22:04:22,281 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80118.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:04:34,496 INFO [train2.py:809] (0/4) Epoch 21, batch 450, loss[ctc_loss=0.0563, att_loss=0.2055, loss=0.1756, over 15503.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008402, over 36.00 utterances.], tot_loss[ctc_loss=0.07433, att_loss=0.2356, loss=0.2033, over 2935885.83 frames. utt_duration=1294 frames, utt_pad_proportion=0.04101, over 9083.12 utterances.], batch size: 36, lr: 5.18e-03, grad_scale: 16.0 2023-03-08 22:04:53,689 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 2.042e+02 2.548e+02 3.452e+02 1.189e+03, threshold=5.096e+02, percent-clipped=8.0 2023-03-08 22:05:55,598 INFO [train2.py:809] (0/4) Epoch 21, batch 500, loss[ctc_loss=0.06218, att_loss=0.237, loss=0.202, over 16467.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006584, over 46.00 utterances.], tot_loss[ctc_loss=0.07393, att_loss=0.2349, loss=0.2027, over 3008957.02 frames. utt_duration=1299 frames, utt_pad_proportion=0.04085, over 9274.74 utterances.], batch size: 46, lr: 5.18e-03, grad_scale: 16.0 2023-03-08 22:07:16,767 INFO [train2.py:809] (0/4) Epoch 21, batch 550, loss[ctc_loss=0.06901, att_loss=0.2331, loss=0.2003, over 16127.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006279, over 42.00 utterances.], tot_loss[ctc_loss=0.07446, att_loss=0.2352, loss=0.203, over 3072062.38 frames. utt_duration=1310 frames, utt_pad_proportion=0.03883, over 9392.52 utterances.], batch size: 42, lr: 5.18e-03, grad_scale: 16.0 2023-03-08 22:07:35,877 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.287e+01 1.925e+02 2.164e+02 2.477e+02 4.088e+02, threshold=4.327e+02, percent-clipped=0.0 2023-03-08 22:07:41,490 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80241.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:07:55,752 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.76 vs. limit=2.0 2023-03-08 22:08:00,795 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6660, 2.4316, 2.7010, 3.2792, 3.0593, 3.2189, 2.5534, 2.1935], device='cuda:0'), covar=tensor([0.0758, 0.1803, 0.0962, 0.0784, 0.0862, 0.0477, 0.1256, 0.1880], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0216, 0.0190, 0.0217, 0.0222, 0.0177, 0.0202, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 22:08:37,653 INFO [train2.py:809] (0/4) Epoch 21, batch 600, loss[ctc_loss=0.07952, att_loss=0.245, loss=0.2119, over 16541.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006047, over 45.00 utterances.], tot_loss[ctc_loss=0.07513, att_loss=0.2359, loss=0.2037, over 3116191.70 frames. utt_duration=1276 frames, utt_pad_proportion=0.04751, over 9783.43 utterances.], batch size: 45, lr: 5.18e-03, grad_scale: 16.0 2023-03-08 22:08:49,085 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80283.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:08:59,311 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80289.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:09:38,454 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80313.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:09:58,598 INFO [train2.py:809] (0/4) Epoch 21, batch 650, loss[ctc_loss=0.06857, att_loss=0.2171, loss=0.1874, over 16188.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.0065, over 41.00 utterances.], tot_loss[ctc_loss=0.07618, att_loss=0.2373, loss=0.2051, over 3153696.28 frames. utt_duration=1240 frames, utt_pad_proportion=0.05555, over 10187.06 utterances.], batch size: 41, lr: 5.17e-03, grad_scale: 16.0 2023-03-08 22:10:18,136 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.082e+02 1.849e+02 2.223e+02 2.876e+02 6.376e+02, threshold=4.446e+02, percent-clipped=5.0 2023-03-08 22:10:27,857 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80344.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:10:31,593 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-03-08 22:10:44,395 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4402, 1.6845, 2.0619, 2.2058, 2.1155, 2.2911, 1.7946, 2.4901], device='cuda:0'), covar=tensor([0.1292, 0.2391, 0.1993, 0.1325, 0.4389, 0.0933, 0.1452, 0.1180], device='cuda:0'), in_proj_covar=tensor([0.0116, 0.0123, 0.0116, 0.0108, 0.0121, 0.0105, 0.0127, 0.0097], device='cuda:0'), out_proj_covar=tensor([8.7214e-05, 9.4997e-05, 9.3097e-05, 8.4063e-05, 8.9565e-05, 8.4000e-05, 9.5310e-05, 7.7616e-05], device='cuda:0') 2023-03-08 22:10:55,568 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80361.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:11:15,289 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4635, 2.4195, 4.8059, 3.6749, 2.7664, 4.1450, 4.5095, 4.5255], device='cuda:0'), covar=tensor([0.0217, 0.1746, 0.0187, 0.0982, 0.1951, 0.0271, 0.0195, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0240, 0.0181, 0.0308, 0.0261, 0.0210, 0.0171, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 22:11:19,497 INFO [train2.py:809] (0/4) Epoch 21, batch 700, loss[ctc_loss=0.0855, att_loss=0.2494, loss=0.2166, over 17118.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01448, over 56.00 utterances.], tot_loss[ctc_loss=0.07648, att_loss=0.2376, loss=0.2053, over 3175667.39 frames. utt_duration=1195 frames, utt_pad_proportion=0.06747, over 10640.56 utterances.], batch size: 56, lr: 5.17e-03, grad_scale: 16.0 2023-03-08 22:11:46,959 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80393.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:12:28,201 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80418.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:12:40,236 INFO [train2.py:809] (0/4) Epoch 21, batch 750, loss[ctc_loss=0.04811, att_loss=0.2352, loss=0.1978, over 17433.00 frames. utt_duration=1108 frames, utt_pad_proportion=0.03122, over 63.00 utterances.], tot_loss[ctc_loss=0.07612, att_loss=0.2381, loss=0.2057, over 3205549.94 frames. utt_duration=1199 frames, utt_pad_proportion=0.06428, over 10707.88 utterances.], batch size: 63, lr: 5.17e-03, grad_scale: 16.0 2023-03-08 22:12:59,014 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 1.906e+02 2.332e+02 2.965e+02 6.816e+02, threshold=4.664e+02, percent-clipped=5.0 2023-03-08 22:13:25,124 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80454.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:13:39,524 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80463.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:13:44,235 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80466.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:13:59,489 INFO [train2.py:809] (0/4) Epoch 21, batch 800, loss[ctc_loss=0.08038, att_loss=0.2478, loss=0.2143, over 17064.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.00889, over 53.00 utterances.], tot_loss[ctc_loss=0.0758, att_loss=0.2373, loss=0.205, over 3219596.32 frames. utt_duration=1224 frames, utt_pad_proportion=0.0578, over 10532.42 utterances.], batch size: 53, lr: 5.17e-03, grad_scale: 16.0 2023-03-08 22:14:29,050 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80494.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:14:42,577 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.88 vs. limit=5.0 2023-03-08 22:15:05,149 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9269, 4.9682, 4.8090, 2.1626, 1.8882, 2.7858, 2.4398, 3.6654], device='cuda:0'), covar=tensor([0.0744, 0.0247, 0.0238, 0.5204, 0.6067, 0.2665, 0.3695, 0.1768], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0270, 0.0265, 0.0241, 0.0342, 0.0332, 0.0251, 0.0363], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 22:15:17,437 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80524.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:15:20,314 INFO [train2.py:809] (0/4) Epoch 21, batch 850, loss[ctc_loss=0.07057, att_loss=0.2356, loss=0.2026, over 17357.00 frames. utt_duration=1008 frames, utt_pad_proportion=0.04753, over 69.00 utterances.], tot_loss[ctc_loss=0.07532, att_loss=0.2364, loss=0.2042, over 3221198.34 frames. utt_duration=1219 frames, utt_pad_proportion=0.06145, over 10579.82 utterances.], batch size: 69, lr: 5.17e-03, grad_scale: 16.0 2023-03-08 22:15:40,967 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.109e+02 2.073e+02 2.361e+02 2.912e+02 1.019e+03, threshold=4.723e+02, percent-clipped=4.0 2023-03-08 22:16:07,004 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80555.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:16:40,197 INFO [train2.py:809] (0/4) Epoch 21, batch 900, loss[ctc_loss=0.0729, att_loss=0.22, loss=0.1906, over 16289.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006671, over 43.00 utterances.], tot_loss[ctc_loss=0.0746, att_loss=0.2359, loss=0.2036, over 3233269.48 frames. utt_duration=1238 frames, utt_pad_proportion=0.0569, over 10462.72 utterances.], batch size: 43, lr: 5.17e-03, grad_scale: 16.0 2023-03-08 22:18:00,929 INFO [train2.py:809] (0/4) Epoch 21, batch 950, loss[ctc_loss=0.06138, att_loss=0.214, loss=0.1835, over 15867.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01044, over 39.00 utterances.], tot_loss[ctc_loss=0.07517, att_loss=0.2358, loss=0.2037, over 3234479.82 frames. utt_duration=1233 frames, utt_pad_proportion=0.06029, over 10501.78 utterances.], batch size: 39, lr: 5.16e-03, grad_scale: 16.0 2023-03-08 22:18:22,509 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 1.900e+02 2.361e+02 2.857e+02 6.538e+02, threshold=4.723e+02, percent-clipped=3.0 2023-03-08 22:18:22,748 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80639.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:19:22,696 INFO [train2.py:809] (0/4) Epoch 21, batch 1000, loss[ctc_loss=0.08574, att_loss=0.2486, loss=0.216, over 17144.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01275, over 56.00 utterances.], tot_loss[ctc_loss=0.07408, att_loss=0.2351, loss=0.2029, over 3242927.18 frames. utt_duration=1252 frames, utt_pad_proportion=0.0558, over 10375.32 utterances.], batch size: 56, lr: 5.16e-03, grad_scale: 8.0 2023-03-08 22:19:40,995 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4739, 2.3311, 4.9634, 3.7694, 2.9184, 4.2369, 4.6591, 4.5405], device='cuda:0'), covar=tensor([0.0245, 0.1675, 0.0152, 0.0922, 0.1724, 0.0246, 0.0160, 0.0277], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0238, 0.0182, 0.0309, 0.0260, 0.0210, 0.0171, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 22:20:17,011 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7011, 5.9409, 5.4333, 5.6980, 5.6278, 5.0586, 5.2758, 5.1555], device='cuda:0'), covar=tensor([0.1237, 0.0970, 0.1057, 0.0914, 0.0918, 0.1636, 0.2716, 0.2611], device='cuda:0'), in_proj_covar=tensor([0.0509, 0.0593, 0.0448, 0.0445, 0.0418, 0.0458, 0.0598, 0.0514], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 22:20:44,014 INFO [train2.py:809] (0/4) Epoch 21, batch 1050, loss[ctc_loss=0.07317, att_loss=0.2341, loss=0.2019, over 16778.00 frames. utt_duration=679.5 frames, utt_pad_proportion=0.1485, over 99.00 utterances.], tot_loss[ctc_loss=0.07375, att_loss=0.2351, loss=0.2028, over 3243796.74 frames. utt_duration=1249 frames, utt_pad_proportion=0.05605, over 10397.12 utterances.], batch size: 99, lr: 5.16e-03, grad_scale: 4.0 2023-03-08 22:21:00,944 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80736.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:21:08,354 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.034e+02 1.786e+02 2.169e+02 2.535e+02 5.351e+02, threshold=4.338e+02, percent-clipped=1.0 2023-03-08 22:21:21,050 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80749.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:22:00,939 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=80773.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:22:05,379 INFO [train2.py:809] (0/4) Epoch 21, batch 1100, loss[ctc_loss=0.0383, att_loss=0.1968, loss=0.1651, over 14556.00 frames. utt_duration=1821 frames, utt_pad_proportion=0.04109, over 32.00 utterances.], tot_loss[ctc_loss=0.07368, att_loss=0.2351, loss=0.2028, over 3253398.80 frames. utt_duration=1261 frames, utt_pad_proportion=0.05207, over 10330.64 utterances.], batch size: 32, lr: 5.16e-03, grad_scale: 4.0 2023-03-08 22:22:39,408 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80797.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:23:16,375 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80819.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:23:27,019 INFO [train2.py:809] (0/4) Epoch 21, batch 1150, loss[ctc_loss=0.111, att_loss=0.2661, loss=0.2351, over 17409.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.04738, over 69.00 utterances.], tot_loss[ctc_loss=0.07532, att_loss=0.2369, loss=0.2046, over 3261088.06 frames. utt_duration=1223 frames, utt_pad_proportion=0.06154, over 10682.11 utterances.], batch size: 69, lr: 5.16e-03, grad_scale: 4.0 2023-03-08 22:23:40,094 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=80834.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:23:51,466 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 2.024e+02 2.478e+02 3.117e+02 8.099e+02, threshold=4.955e+02, percent-clipped=10.0 2023-03-08 22:23:58,296 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1684, 2.5046, 4.4301, 3.5884, 2.8468, 3.9055, 4.1514, 4.0979], device='cuda:0'), covar=tensor([0.0230, 0.1566, 0.0174, 0.1012, 0.1771, 0.0308, 0.0231, 0.0322], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0239, 0.0182, 0.0310, 0.0261, 0.0211, 0.0171, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 22:24:06,587 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=80850.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:24:48,752 INFO [train2.py:809] (0/4) Epoch 21, batch 1200, loss[ctc_loss=0.08563, att_loss=0.2495, loss=0.2167, over 17071.00 frames. utt_duration=1315 frames, utt_pad_proportion=0.007823, over 52.00 utterances.], tot_loss[ctc_loss=0.0749, att_loss=0.2364, loss=0.2041, over 3265635.39 frames. utt_duration=1243 frames, utt_pad_proportion=0.05648, over 10522.42 utterances.], batch size: 52, lr: 5.16e-03, grad_scale: 8.0 2023-03-08 22:24:49,307 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9608, 4.6976, 4.6611, 2.2255, 2.1428, 2.8222, 2.4044, 3.7527], device='cuda:0'), covar=tensor([0.0718, 0.0283, 0.0242, 0.5099, 0.5159, 0.2578, 0.3457, 0.1468], device='cuda:0'), in_proj_covar=tensor([0.0358, 0.0274, 0.0269, 0.0247, 0.0346, 0.0337, 0.0255, 0.0368], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 22:26:09,924 INFO [train2.py:809] (0/4) Epoch 21, batch 1250, loss[ctc_loss=0.06809, att_loss=0.2154, loss=0.186, over 15638.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.00874, over 37.00 utterances.], tot_loss[ctc_loss=0.07563, att_loss=0.2367, loss=0.2045, over 3257081.26 frames. utt_duration=1230 frames, utt_pad_proportion=0.0612, over 10606.21 utterances.], batch size: 37, lr: 5.16e-03, grad_scale: 8.0 2023-03-08 22:26:14,660 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-03-08 22:26:31,986 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=80939.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:26:34,820 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 2.023e+02 2.469e+02 3.077e+02 5.945e+02, threshold=4.937e+02, percent-clipped=1.0 2023-03-08 22:27:33,054 INFO [train2.py:809] (0/4) Epoch 21, batch 1300, loss[ctc_loss=0.0734, att_loss=0.2445, loss=0.2103, over 16609.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.00638, over 47.00 utterances.], tot_loss[ctc_loss=0.07401, att_loss=0.2356, loss=0.2033, over 3255340.78 frames. utt_duration=1231 frames, utt_pad_proportion=0.06155, over 10592.02 utterances.], batch size: 47, lr: 5.15e-03, grad_scale: 8.0 2023-03-08 22:27:50,963 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=80987.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:28:44,941 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2989, 2.5123, 4.6189, 3.6946, 2.8738, 4.0557, 4.3095, 4.3361], device='cuda:0'), covar=tensor([0.0253, 0.1614, 0.0200, 0.0945, 0.1753, 0.0295, 0.0230, 0.0279], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0238, 0.0182, 0.0309, 0.0260, 0.0210, 0.0171, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 22:28:46,380 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8955, 3.9509, 3.7285, 2.8202, 3.7827, 3.7727, 3.4642, 2.7089], device='cuda:0'), covar=tensor([0.0130, 0.0128, 0.0300, 0.0855, 0.0130, 0.0383, 0.0366, 0.1197], device='cuda:0'), in_proj_covar=tensor([0.0073, 0.0101, 0.0103, 0.0110, 0.0084, 0.0110, 0.0099, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 22:28:55,451 INFO [train2.py:809] (0/4) Epoch 21, batch 1350, loss[ctc_loss=0.06256, att_loss=0.2158, loss=0.1852, over 16139.00 frames. utt_duration=1539 frames, utt_pad_proportion=0.005371, over 42.00 utterances.], tot_loss[ctc_loss=0.07436, att_loss=0.2357, loss=0.2034, over 3254309.82 frames. utt_duration=1206 frames, utt_pad_proportion=0.06718, over 10803.61 utterances.], batch size: 42, lr: 5.15e-03, grad_scale: 8.0 2023-03-08 22:29:19,939 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 1.963e+02 2.296e+02 2.713e+02 5.032e+02, threshold=4.592e+02, percent-clipped=2.0 2023-03-08 22:29:33,230 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81049.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:30:15,541 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1552, 5.4491, 4.9585, 5.4769, 4.7978, 5.1121, 5.5644, 5.3237], device='cuda:0'), covar=tensor([0.0493, 0.0279, 0.0813, 0.0302, 0.0437, 0.0233, 0.0193, 0.0180], device='cuda:0'), in_proj_covar=tensor([0.0381, 0.0316, 0.0363, 0.0344, 0.0315, 0.0237, 0.0297, 0.0279], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 22:30:16,881 INFO [train2.py:809] (0/4) Epoch 21, batch 1400, loss[ctc_loss=0.04233, att_loss=0.204, loss=0.1717, over 15648.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.007964, over 37.00 utterances.], tot_loss[ctc_loss=0.07348, att_loss=0.2352, loss=0.2029, over 3255457.83 frames. utt_duration=1237 frames, utt_pad_proportion=0.06065, over 10542.08 utterances.], batch size: 37, lr: 5.15e-03, grad_scale: 8.0 2023-03-08 22:30:42,751 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81092.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:30:50,481 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81097.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:31:22,072 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81116.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:31:26,384 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81119.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:31:36,837 INFO [train2.py:809] (0/4) Epoch 21, batch 1450, loss[ctc_loss=0.0792, att_loss=0.2494, loss=0.2154, over 16967.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.006711, over 50.00 utterances.], tot_loss[ctc_loss=0.07315, att_loss=0.2351, loss=0.2027, over 3261570.11 frames. utt_duration=1265 frames, utt_pad_proportion=0.05272, over 10324.20 utterances.], batch size: 50, lr: 5.15e-03, grad_scale: 8.0 2023-03-08 22:31:41,559 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81129.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:32:00,575 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 1.791e+02 2.173e+02 2.721e+02 6.345e+02, threshold=4.347e+02, percent-clipped=3.0 2023-03-08 22:32:15,015 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81150.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:32:43,708 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81167.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:32:57,636 INFO [train2.py:809] (0/4) Epoch 21, batch 1500, loss[ctc_loss=0.06438, att_loss=0.2421, loss=0.2066, over 16765.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006629, over 48.00 utterances.], tot_loss[ctc_loss=0.0734, att_loss=0.2354, loss=0.203, over 3270676.57 frames. utt_duration=1267 frames, utt_pad_proportion=0.04888, over 10336.08 utterances.], batch size: 48, lr: 5.15e-03, grad_scale: 8.0 2023-03-08 22:32:59,560 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81177.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:33:33,283 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81198.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:34:07,042 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-03-08 22:34:19,140 INFO [train2.py:809] (0/4) Epoch 21, batch 1550, loss[ctc_loss=0.06473, att_loss=0.2124, loss=0.1829, over 15366.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01152, over 35.00 utterances.], tot_loss[ctc_loss=0.07373, att_loss=0.235, loss=0.2028, over 3273328.63 frames. utt_duration=1258 frames, utt_pad_proportion=0.0508, over 10419.16 utterances.], batch size: 35, lr: 5.15e-03, grad_scale: 8.0 2023-03-08 22:34:43,500 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.361e+02 1.828e+02 2.165e+02 2.736e+02 6.423e+02, threshold=4.329e+02, percent-clipped=4.0 2023-03-08 22:35:40,194 INFO [train2.py:809] (0/4) Epoch 21, batch 1600, loss[ctc_loss=0.07394, att_loss=0.2421, loss=0.2085, over 17007.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.009338, over 51.00 utterances.], tot_loss[ctc_loss=0.07364, att_loss=0.2353, loss=0.203, over 3285627.72 frames. utt_duration=1257 frames, utt_pad_proportion=0.04742, over 10470.55 utterances.], batch size: 51, lr: 5.14e-03, grad_scale: 8.0 2023-03-08 22:35:57,071 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0643, 4.9672, 4.8492, 1.9412, 1.9085, 2.6542, 2.2108, 3.7731], device='cuda:0'), covar=tensor([0.0687, 0.0258, 0.0235, 0.5854, 0.5936, 0.2928, 0.3930, 0.1655], device='cuda:0'), in_proj_covar=tensor([0.0351, 0.0271, 0.0263, 0.0242, 0.0339, 0.0330, 0.0250, 0.0359], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 22:36:16,918 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81299.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:36:20,811 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81301.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:36:23,766 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7275, 5.0119, 4.9223, 4.9065, 4.9751, 4.9358, 4.6569, 4.4803], device='cuda:0'), covar=tensor([0.1016, 0.0525, 0.0345, 0.0549, 0.0337, 0.0347, 0.0427, 0.0373], device='cuda:0'), in_proj_covar=tensor([0.0518, 0.0359, 0.0342, 0.0356, 0.0417, 0.0426, 0.0352, 0.0391], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 22:36:55,003 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9324, 5.2742, 4.7707, 5.2931, 4.6005, 4.9608, 5.3906, 5.1642], device='cuda:0'), covar=tensor([0.0607, 0.0292, 0.0811, 0.0351, 0.0486, 0.0286, 0.0214, 0.0194], device='cuda:0'), in_proj_covar=tensor([0.0380, 0.0314, 0.0360, 0.0343, 0.0314, 0.0236, 0.0295, 0.0280], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 22:37:01,166 INFO [train2.py:809] (0/4) Epoch 21, batch 1650, loss[ctc_loss=0.1355, att_loss=0.2724, loss=0.2451, over 13996.00 frames. utt_duration=387.7 frames, utt_pad_proportion=0.3258, over 145.00 utterances.], tot_loss[ctc_loss=0.07376, att_loss=0.236, loss=0.2035, over 3279243.45 frames. utt_duration=1233 frames, utt_pad_proportion=0.05598, over 10649.87 utterances.], batch size: 145, lr: 5.14e-03, grad_scale: 8.0 2023-03-08 22:37:25,609 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 1.957e+02 2.223e+02 2.631e+02 7.416e+02, threshold=4.446e+02, percent-clipped=3.0 2023-03-08 22:37:58,108 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81360.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:38:01,300 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81362.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:38:17,858 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.72 vs. limit=5.0 2023-03-08 22:38:20,167 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2063, 5.2172, 4.8014, 2.9865, 4.9520, 4.9425, 4.2477, 2.6731], device='cuda:0'), covar=tensor([0.0136, 0.0112, 0.0348, 0.1059, 0.0114, 0.0157, 0.0385, 0.1579], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0101, 0.0103, 0.0110, 0.0084, 0.0110, 0.0099, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 22:38:23,004 INFO [train2.py:809] (0/4) Epoch 21, batch 1700, loss[ctc_loss=0.06064, att_loss=0.2175, loss=0.1861, over 15990.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008256, over 40.00 utterances.], tot_loss[ctc_loss=0.07337, att_loss=0.2358, loss=0.2033, over 3279429.06 frames. utt_duration=1242 frames, utt_pad_proportion=0.05465, over 10576.35 utterances.], batch size: 40, lr: 5.14e-03, grad_scale: 8.0 2023-03-08 22:38:46,491 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-08 22:38:49,095 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81392.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:38:56,726 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8174, 5.2333, 4.9986, 5.1246, 5.2261, 4.8809, 3.6044, 5.2151], device='cuda:0'), covar=tensor([0.0124, 0.0112, 0.0148, 0.0092, 0.0103, 0.0133, 0.0710, 0.0220], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0087, 0.0111, 0.0070, 0.0075, 0.0086, 0.0104, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 22:39:44,356 INFO [train2.py:809] (0/4) Epoch 21, batch 1750, loss[ctc_loss=0.03833, att_loss=0.1977, loss=0.1659, over 15522.00 frames. utt_duration=1726 frames, utt_pad_proportion=0.007396, over 36.00 utterances.], tot_loss[ctc_loss=0.0733, att_loss=0.2356, loss=0.2031, over 3278479.24 frames. utt_duration=1226 frames, utt_pad_proportion=0.05803, over 10705.81 utterances.], batch size: 36, lr: 5.14e-03, grad_scale: 8.0 2023-03-08 22:39:49,350 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81429.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:40:06,975 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81440.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:40:08,460 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.362e+02 2.066e+02 2.468e+02 3.056e+02 6.142e+02, threshold=4.936e+02, percent-clipped=4.0 2023-03-08 22:40:59,673 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81472.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:41:05,534 INFO [train2.py:809] (0/4) Epoch 21, batch 1800, loss[ctc_loss=0.06845, att_loss=0.2378, loss=0.2039, over 16770.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.005745, over 48.00 utterances.], tot_loss[ctc_loss=0.07353, att_loss=0.2358, loss=0.2033, over 3280002.91 frames. utt_duration=1237 frames, utt_pad_proportion=0.05601, over 10616.67 utterances.], batch size: 48, lr: 5.14e-03, grad_scale: 8.0 2023-03-08 22:41:07,180 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81477.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:42:25,225 INFO [train2.py:809] (0/4) Epoch 21, batch 1850, loss[ctc_loss=0.08324, att_loss=0.2197, loss=0.1924, over 15359.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01135, over 35.00 utterances.], tot_loss[ctc_loss=0.07364, att_loss=0.2361, loss=0.2036, over 3284216.08 frames. utt_duration=1259 frames, utt_pad_proportion=0.04913, over 10445.44 utterances.], batch size: 35, lr: 5.14e-03, grad_scale: 8.0 2023-03-08 22:42:49,242 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 1.850e+02 2.304e+02 2.710e+02 5.976e+02, threshold=4.608e+02, percent-clipped=1.0 2023-03-08 22:43:45,866 INFO [train2.py:809] (0/4) Epoch 21, batch 1900, loss[ctc_loss=0.06634, att_loss=0.2408, loss=0.2059, over 16891.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006274, over 49.00 utterances.], tot_loss[ctc_loss=0.07462, att_loss=0.2365, loss=0.2041, over 3282601.63 frames. utt_duration=1248 frames, utt_pad_proportion=0.05235, over 10536.53 utterances.], batch size: 49, lr: 5.13e-03, grad_scale: 8.0 2023-03-08 22:44:38,922 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2068, 5.4221, 5.3987, 5.3241, 5.4694, 5.3919, 5.1525, 4.9394], device='cuda:0'), covar=tensor([0.0934, 0.0575, 0.0256, 0.0467, 0.0291, 0.0280, 0.0356, 0.0334], device='cuda:0'), in_proj_covar=tensor([0.0510, 0.0354, 0.0336, 0.0350, 0.0412, 0.0420, 0.0349, 0.0385], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 22:45:07,136 INFO [train2.py:809] (0/4) Epoch 21, batch 1950, loss[ctc_loss=0.08124, att_loss=0.2517, loss=0.2176, over 17092.00 frames. utt_duration=866.7 frames, utt_pad_proportion=0.09151, over 79.00 utterances.], tot_loss[ctc_loss=0.07395, att_loss=0.2359, loss=0.2035, over 3281876.07 frames. utt_duration=1267 frames, utt_pad_proportion=0.04839, over 10370.18 utterances.], batch size: 79, lr: 5.13e-03, grad_scale: 8.0 2023-03-08 22:45:31,399 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.479e+02 1.984e+02 2.266e+02 2.808e+02 6.592e+02, threshold=4.533e+02, percent-clipped=3.0 2023-03-08 22:45:54,431 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81655.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:45:58,064 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=81657.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:46:18,718 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81670.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:46:27,655 INFO [train2.py:809] (0/4) Epoch 21, batch 2000, loss[ctc_loss=0.05766, att_loss=0.2077, loss=0.1777, over 15476.00 frames. utt_duration=1721 frames, utt_pad_proportion=0.01045, over 36.00 utterances.], tot_loss[ctc_loss=0.07427, att_loss=0.2362, loss=0.2038, over 3269728.70 frames. utt_duration=1238 frames, utt_pad_proportion=0.0576, over 10578.29 utterances.], batch size: 36, lr: 5.13e-03, grad_scale: 8.0 2023-03-08 22:47:47,658 INFO [train2.py:809] (0/4) Epoch 21, batch 2050, loss[ctc_loss=0.08074, att_loss=0.2396, loss=0.2079, over 16605.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.005177, over 47.00 utterances.], tot_loss[ctc_loss=0.0745, att_loss=0.2362, loss=0.2039, over 3274232.54 frames. utt_duration=1234 frames, utt_pad_proportion=0.05729, over 10630.62 utterances.], batch size: 47, lr: 5.13e-03, grad_scale: 8.0 2023-03-08 22:47:55,824 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=81731.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 22:48:11,615 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 1.913e+02 2.441e+02 2.878e+02 1.219e+03, threshold=4.882e+02, percent-clipped=4.0 2023-03-08 22:48:22,170 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 22:48:26,132 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.50 vs. limit=2.0 2023-03-08 22:48:33,308 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-08 22:48:50,198 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1896, 3.8745, 3.8902, 3.3519, 3.8566, 3.9430, 3.8395, 2.7906], device='cuda:0'), covar=tensor([0.0908, 0.1258, 0.1654, 0.3514, 0.1047, 0.1972, 0.0853, 0.3979], device='cuda:0'), in_proj_covar=tensor([0.0169, 0.0184, 0.0194, 0.0247, 0.0154, 0.0256, 0.0174, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 22:49:02,499 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81772.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:49:08,390 INFO [train2.py:809] (0/4) Epoch 21, batch 2100, loss[ctc_loss=0.07641, att_loss=0.221, loss=0.1921, over 15362.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.0116, over 35.00 utterances.], tot_loss[ctc_loss=0.07523, att_loss=0.2364, loss=0.2041, over 3260456.75 frames. utt_duration=1197 frames, utt_pad_proportion=0.07074, over 10911.99 utterances.], batch size: 35, lr: 5.13e-03, grad_scale: 8.0 2023-03-08 22:49:52,964 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.21 vs. limit=5.0 2023-03-08 22:49:56,208 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-03-08 22:50:20,892 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=81820.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:50:30,145 INFO [train2.py:809] (0/4) Epoch 21, batch 2150, loss[ctc_loss=0.0752, att_loss=0.2362, loss=0.204, over 16269.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.00795, over 43.00 utterances.], tot_loss[ctc_loss=0.07488, att_loss=0.2363, loss=0.204, over 3263948.88 frames. utt_duration=1231 frames, utt_pad_proportion=0.06108, over 10616.90 utterances.], batch size: 43, lr: 5.13e-03, grad_scale: 8.0 2023-03-08 22:50:54,308 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.279e+02 1.964e+02 2.354e+02 2.825e+02 8.417e+02, threshold=4.708e+02, percent-clipped=3.0 2023-03-08 22:50:55,322 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.81 vs. limit=5.0 2023-03-08 22:51:20,956 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8899, 5.1129, 5.0654, 4.9983, 5.1587, 5.1137, 4.8065, 4.6183], device='cuda:0'), covar=tensor([0.1013, 0.0563, 0.0293, 0.0566, 0.0298, 0.0353, 0.0429, 0.0377], device='cuda:0'), in_proj_covar=tensor([0.0517, 0.0359, 0.0339, 0.0355, 0.0414, 0.0422, 0.0351, 0.0389], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 22:51:44,863 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9443, 3.9828, 3.8023, 2.9042, 3.8026, 3.8283, 3.5663, 2.7279], device='cuda:0'), covar=tensor([0.0149, 0.0139, 0.0306, 0.0825, 0.0139, 0.0356, 0.0321, 0.1245], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0102, 0.0104, 0.0110, 0.0084, 0.0112, 0.0099, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 22:51:50,727 INFO [train2.py:809] (0/4) Epoch 21, batch 2200, loss[ctc_loss=0.07819, att_loss=0.2347, loss=0.2034, over 16625.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005379, over 47.00 utterances.], tot_loss[ctc_loss=0.07446, att_loss=0.2358, loss=0.2036, over 3259729.31 frames. utt_duration=1231 frames, utt_pad_proportion=0.06266, over 10602.73 utterances.], batch size: 47, lr: 5.13e-03, grad_scale: 8.0 2023-03-08 22:52:24,786 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-08 22:52:59,587 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0541, 4.9458, 5.0707, 1.9447, 1.9756, 2.7288, 2.4976, 3.7327], device='cuda:0'), covar=tensor([0.0927, 0.0560, 0.0262, 0.5523, 0.6822, 0.3132, 0.3770, 0.2104], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0277, 0.0271, 0.0248, 0.0347, 0.0336, 0.0255, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 22:53:11,519 INFO [train2.py:809] (0/4) Epoch 21, batch 2250, loss[ctc_loss=0.08072, att_loss=0.2542, loss=0.2195, over 17351.00 frames. utt_duration=1178 frames, utt_pad_proportion=0.02199, over 59.00 utterances.], tot_loss[ctc_loss=0.07452, att_loss=0.2362, loss=0.2039, over 3264138.35 frames. utt_duration=1230 frames, utt_pad_proportion=0.06291, over 10624.43 utterances.], batch size: 59, lr: 5.12e-03, grad_scale: 8.0 2023-03-08 22:53:35,601 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 2.106e+02 2.465e+02 3.018e+02 5.655e+02, threshold=4.929e+02, percent-clipped=3.0 2023-03-08 22:53:58,771 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81955.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:54:02,629 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=81957.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:54:04,335 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4996, 3.0518, 2.6278, 2.9014, 3.1677, 3.0739, 2.4972, 3.0651], device='cuda:0'), covar=tensor([0.0969, 0.0441, 0.0857, 0.0597, 0.0597, 0.0540, 0.0821, 0.0422], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0217, 0.0227, 0.0199, 0.0279, 0.0240, 0.0200, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 22:54:32,969 INFO [train2.py:809] (0/4) Epoch 21, batch 2300, loss[ctc_loss=0.07437, att_loss=0.2425, loss=0.2089, over 17393.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03336, over 63.00 utterances.], tot_loss[ctc_loss=0.07462, att_loss=0.2359, loss=0.2037, over 3250055.27 frames. utt_duration=1193 frames, utt_pad_proportion=0.07534, over 10906.66 utterances.], batch size: 63, lr: 5.12e-03, grad_scale: 8.0 2023-03-08 22:54:50,486 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=81987.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:54:55,163 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7763, 5.2146, 4.9745, 5.1171, 5.2650, 4.8372, 3.8839, 5.1402], device='cuda:0'), covar=tensor([0.0119, 0.0113, 0.0150, 0.0093, 0.0101, 0.0115, 0.0602, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0088, 0.0112, 0.0070, 0.0076, 0.0086, 0.0105, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 22:55:12,700 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-82000.pt 2023-03-08 22:55:22,027 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82003.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:55:23,097 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-08 22:55:25,329 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82005.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:55:26,254 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-03-08 22:55:59,323 INFO [train2.py:809] (0/4) Epoch 21, batch 2350, loss[ctc_loss=0.05861, att_loss=0.2306, loss=0.1962, over 16954.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008064, over 50.00 utterances.], tot_loss[ctc_loss=0.07473, att_loss=0.2363, loss=0.204, over 3255125.11 frames. utt_duration=1205 frames, utt_pad_proportion=0.07088, over 10819.34 utterances.], batch size: 50, lr: 5.12e-03, grad_scale: 8.0 2023-03-08 22:55:59,522 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82026.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 22:56:09,226 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0070, 4.4232, 4.3281, 4.5998, 2.7327, 4.4239, 2.7274, 1.6213], device='cuda:0'), covar=tensor([0.0437, 0.0232, 0.0634, 0.0212, 0.1570, 0.0209, 0.1401, 0.1865], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0163, 0.0257, 0.0156, 0.0220, 0.0143, 0.0228, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 22:56:22,870 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.390e+02 1.965e+02 2.374e+02 2.963e+02 4.469e+02, threshold=4.748e+02, percent-clipped=0.0 2023-03-08 22:56:35,665 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82048.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:56:38,704 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7948, 6.0611, 5.4931, 5.8165, 5.7214, 5.1945, 5.4516, 5.2757], device='cuda:0'), covar=tensor([0.1249, 0.0814, 0.0891, 0.0759, 0.0798, 0.1544, 0.2260, 0.2100], device='cuda:0'), in_proj_covar=tensor([0.0520, 0.0607, 0.0457, 0.0455, 0.0427, 0.0464, 0.0607, 0.0526], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 22:56:47,292 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82055.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:57:02,780 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0424, 6.2784, 5.7578, 6.0447, 5.9592, 5.4399, 5.7536, 5.4719], device='cuda:0'), covar=tensor([0.1245, 0.0901, 0.0884, 0.0811, 0.0750, 0.1558, 0.2278, 0.2502], device='cuda:0'), in_proj_covar=tensor([0.0520, 0.0607, 0.0456, 0.0454, 0.0427, 0.0464, 0.0607, 0.0526], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-08 22:57:20,439 INFO [train2.py:809] (0/4) Epoch 21, batch 2400, loss[ctc_loss=0.1028, att_loss=0.2493, loss=0.22, over 17325.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02235, over 59.00 utterances.], tot_loss[ctc_loss=0.07468, att_loss=0.2362, loss=0.2039, over 3262203.99 frames. utt_duration=1216 frames, utt_pad_proportion=0.06682, over 10741.53 utterances.], batch size: 59, lr: 5.12e-03, grad_scale: 8.0 2023-03-08 22:58:25,036 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82116.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 22:58:40,651 INFO [train2.py:809] (0/4) Epoch 21, batch 2450, loss[ctc_loss=0.0534, att_loss=0.2189, loss=0.1858, over 12896.00 frames. utt_duration=1844 frames, utt_pad_proportion=0.1085, over 28.00 utterances.], tot_loss[ctc_loss=0.07448, att_loss=0.2362, loss=0.2039, over 3270328.55 frames. utt_duration=1250 frames, utt_pad_proportion=0.05656, over 10478.39 utterances.], batch size: 28, lr: 5.12e-03, grad_scale: 8.0 2023-03-08 22:59:04,489 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.920e+01 1.856e+02 2.212e+02 2.703e+02 4.819e+02, threshold=4.424e+02, percent-clipped=1.0 2023-03-08 23:00:00,568 INFO [train2.py:809] (0/4) Epoch 21, batch 2500, loss[ctc_loss=0.05623, att_loss=0.2357, loss=0.1998, over 16882.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.006861, over 49.00 utterances.], tot_loss[ctc_loss=0.07459, att_loss=0.2361, loss=0.2038, over 3270585.65 frames. utt_duration=1261 frames, utt_pad_proportion=0.05333, over 10383.07 utterances.], batch size: 49, lr: 5.12e-03, grad_scale: 8.0 2023-03-08 23:01:21,097 INFO [train2.py:809] (0/4) Epoch 21, batch 2550, loss[ctc_loss=0.08835, att_loss=0.2455, loss=0.214, over 17349.00 frames. utt_duration=1007 frames, utt_pad_proportion=0.0491, over 69.00 utterances.], tot_loss[ctc_loss=0.07505, att_loss=0.2365, loss=0.2042, over 3276017.66 frames. utt_duration=1253 frames, utt_pad_proportion=0.05369, over 10466.58 utterances.], batch size: 69, lr: 5.11e-03, grad_scale: 8.0 2023-03-08 23:01:45,119 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.468e+02 1.926e+02 2.375e+02 3.006e+02 1.037e+03, threshold=4.749e+02, percent-clipped=4.0 2023-03-08 23:02:12,224 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4875, 4.4017, 4.4528, 4.4180, 5.0599, 4.3718, 4.5279, 2.5413], device='cuda:0'), covar=tensor([0.0279, 0.0347, 0.0371, 0.0345, 0.0897, 0.0281, 0.0323, 0.1875], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0185, 0.0185, 0.0201, 0.0365, 0.0157, 0.0175, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:02:23,492 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 23:02:42,050 INFO [train2.py:809] (0/4) Epoch 21, batch 2600, loss[ctc_loss=0.06237, att_loss=0.2201, loss=0.1886, over 16397.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007107, over 44.00 utterances.], tot_loss[ctc_loss=0.07509, att_loss=0.2361, loss=0.2039, over 3265559.70 frames. utt_duration=1236 frames, utt_pad_proportion=0.06137, over 10584.41 utterances.], batch size: 44, lr: 5.11e-03, grad_scale: 8.0 2023-03-08 23:03:01,859 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6007, 4.5392, 4.5180, 4.5225, 5.1250, 4.3840, 4.6074, 2.5956], device='cuda:0'), covar=tensor([0.0222, 0.0334, 0.0324, 0.0314, 0.1016, 0.0282, 0.0302, 0.1918], device='cuda:0'), in_proj_covar=tensor([0.0163, 0.0185, 0.0185, 0.0202, 0.0366, 0.0158, 0.0175, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:03:04,938 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1078, 4.4700, 4.1038, 4.6524, 2.4122, 4.6059, 2.6006, 1.6749], device='cuda:0'), covar=tensor([0.0446, 0.0256, 0.0802, 0.0288, 0.1963, 0.0183, 0.1621, 0.1953], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0162, 0.0254, 0.0154, 0.0219, 0.0141, 0.0227, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:03:31,080 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6582, 3.4458, 3.4005, 2.9632, 3.4067, 3.4382, 3.4974, 2.4951], device='cuda:0'), covar=tensor([0.1220, 0.1247, 0.1776, 0.3948, 0.1257, 0.2783, 0.1009, 0.4139], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0186, 0.0197, 0.0251, 0.0155, 0.0256, 0.0175, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:04:03,377 INFO [train2.py:809] (0/4) Epoch 21, batch 2650, loss[ctc_loss=0.06983, att_loss=0.2253, loss=0.1942, over 16015.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.00697, over 40.00 utterances.], tot_loss[ctc_loss=0.07414, att_loss=0.2355, loss=0.2033, over 3263772.01 frames. utt_duration=1255 frames, utt_pad_proportion=0.05658, over 10413.02 utterances.], batch size: 40, lr: 5.11e-03, grad_scale: 8.0 2023-03-08 23:04:03,721 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82326.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:04:17,760 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1486, 3.8367, 3.2499, 3.4971, 4.0535, 3.6600, 3.1436, 4.2912], device='cuda:0'), covar=tensor([0.0965, 0.0496, 0.1085, 0.0715, 0.0736, 0.0741, 0.0841, 0.0514], device='cuda:0'), in_proj_covar=tensor([0.0200, 0.0215, 0.0225, 0.0199, 0.0277, 0.0240, 0.0199, 0.0284], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:04:27,517 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 1.898e+02 2.205e+02 2.899e+02 5.130e+02, threshold=4.409e+02, percent-clipped=1.0 2023-03-08 23:04:31,585 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82343.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:04:36,584 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82346.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:05:21,679 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82374.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:05:24,744 INFO [train2.py:809] (0/4) Epoch 21, batch 2700, loss[ctc_loss=0.0642, att_loss=0.2185, loss=0.1876, over 14110.00 frames. utt_duration=1822 frames, utt_pad_proportion=0.04998, over 31.00 utterances.], tot_loss[ctc_loss=0.07455, att_loss=0.2359, loss=0.2037, over 3258176.28 frames. utt_duration=1219 frames, utt_pad_proportion=0.06696, over 10708.07 utterances.], batch size: 31, lr: 5.11e-03, grad_scale: 8.0 2023-03-08 23:06:08,338 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8506, 5.0223, 5.0138, 4.9995, 5.1077, 5.0347, 4.7435, 4.5760], device='cuda:0'), covar=tensor([0.0917, 0.0560, 0.0307, 0.0517, 0.0302, 0.0323, 0.0439, 0.0373], device='cuda:0'), in_proj_covar=tensor([0.0513, 0.0358, 0.0340, 0.0351, 0.0414, 0.0419, 0.0349, 0.0389], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:06:16,105 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82407.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:06:22,220 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82411.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:06:33,768 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8415, 5.2208, 5.4364, 5.2358, 5.2973, 5.7956, 5.1792, 5.9044], device='cuda:0'), covar=tensor([0.0790, 0.0725, 0.0797, 0.1366, 0.1784, 0.0929, 0.0702, 0.0670], device='cuda:0'), in_proj_covar=tensor([0.0866, 0.0508, 0.0596, 0.0662, 0.0868, 0.0628, 0.0484, 0.0605], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 23:06:46,065 INFO [train2.py:809] (0/4) Epoch 21, batch 2750, loss[ctc_loss=0.07989, att_loss=0.2408, loss=0.2087, over 17071.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008629, over 53.00 utterances.], tot_loss[ctc_loss=0.075, att_loss=0.2365, loss=0.2042, over 3265733.37 frames. utt_duration=1224 frames, utt_pad_proportion=0.06236, over 10687.88 utterances.], batch size: 53, lr: 5.11e-03, grad_scale: 8.0 2023-03-08 23:07:10,790 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 1.903e+02 2.360e+02 2.836e+02 7.135e+02, threshold=4.721e+02, percent-clipped=4.0 2023-03-08 23:08:06,988 INFO [train2.py:809] (0/4) Epoch 21, batch 2800, loss[ctc_loss=0.07124, att_loss=0.241, loss=0.207, over 16878.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.006381, over 49.00 utterances.], tot_loss[ctc_loss=0.07551, att_loss=0.2365, loss=0.2043, over 3264878.12 frames. utt_duration=1217 frames, utt_pad_proportion=0.06467, over 10742.32 utterances.], batch size: 49, lr: 5.11e-03, grad_scale: 8.0 2023-03-08 23:08:08,649 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3223, 4.7502, 4.6331, 4.6856, 4.8422, 4.5544, 3.1452, 4.6732], device='cuda:0'), covar=tensor([0.0136, 0.0107, 0.0131, 0.0093, 0.0090, 0.0105, 0.0817, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0094, 0.0089, 0.0113, 0.0071, 0.0077, 0.0088, 0.0106, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 23:08:34,698 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82493.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 23:09:26,650 INFO [train2.py:809] (0/4) Epoch 21, batch 2850, loss[ctc_loss=0.08011, att_loss=0.2537, loss=0.219, over 17011.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.009091, over 51.00 utterances.], tot_loss[ctc_loss=0.0762, att_loss=0.2371, loss=0.2049, over 3266451.49 frames. utt_duration=1204 frames, utt_pad_proportion=0.06787, over 10865.93 utterances.], batch size: 51, lr: 5.11e-03, grad_scale: 8.0 2023-03-08 23:09:50,388 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 1.879e+02 2.361e+02 2.971e+02 6.221e+02, threshold=4.723e+02, percent-clipped=6.0 2023-03-08 23:10:11,718 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82554.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 23:10:35,777 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82569.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:10:46,927 INFO [train2.py:809] (0/4) Epoch 21, batch 2900, loss[ctc_loss=0.08621, att_loss=0.2546, loss=0.221, over 17309.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.0232, over 59.00 utterances.], tot_loss[ctc_loss=0.0756, att_loss=0.2365, loss=0.2043, over 3264617.60 frames. utt_duration=1231 frames, utt_pad_proportion=0.06171, over 10621.00 utterances.], batch size: 59, lr: 5.10e-03, grad_scale: 8.0 2023-03-08 23:11:01,843 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8723, 3.2162, 3.7973, 3.3009, 3.8163, 4.8862, 4.6761, 3.5611], device='cuda:0'), covar=tensor([0.0295, 0.1668, 0.1270, 0.1387, 0.0928, 0.0794, 0.0586, 0.1189], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0244, 0.0280, 0.0221, 0.0266, 0.0366, 0.0263, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 23:12:07,980 INFO [train2.py:809] (0/4) Epoch 21, batch 2950, loss[ctc_loss=0.08101, att_loss=0.2407, loss=0.2088, over 16402.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007676, over 44.00 utterances.], tot_loss[ctc_loss=0.07496, att_loss=0.2361, loss=0.2039, over 3263219.93 frames. utt_duration=1231 frames, utt_pad_proportion=0.06156, over 10617.97 utterances.], batch size: 44, lr: 5.10e-03, grad_scale: 8.0 2023-03-08 23:12:14,628 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=82630.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:12:31,985 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 1.787e+02 2.245e+02 2.585e+02 7.263e+02, threshold=4.491e+02, percent-clipped=3.0 2023-03-08 23:12:36,026 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82643.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:13:27,969 INFO [train2.py:809] (0/4) Epoch 21, batch 3000, loss[ctc_loss=0.06898, att_loss=0.2275, loss=0.1958, over 15784.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.00676, over 38.00 utterances.], tot_loss[ctc_loss=0.07447, att_loss=0.2358, loss=0.2035, over 3268855.35 frames. utt_duration=1260 frames, utt_pad_proportion=0.0532, over 10392.06 utterances.], batch size: 38, lr: 5.10e-03, grad_scale: 8.0 2023-03-08 23:13:27,972 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 23:13:41,813 INFO [train2.py:843] (0/4) Epoch 21, validation: ctc_loss=0.04141, att_loss=0.2346, loss=0.196, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 23:13:41,813 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 23:14:05,942 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82691.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:14:21,054 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7935, 2.1284, 2.6439, 2.7138, 2.8794, 2.5618, 2.5473, 3.2016], device='cuda:0'), covar=tensor([0.1466, 0.3248, 0.1971, 0.1473, 0.1808, 0.1371, 0.2541, 0.1040], device='cuda:0'), in_proj_covar=tensor([0.0115, 0.0123, 0.0117, 0.0107, 0.0121, 0.0105, 0.0127, 0.0097], device='cuda:0'), out_proj_covar=tensor([8.6987e-05, 9.5308e-05, 9.3663e-05, 8.3625e-05, 8.9828e-05, 8.3917e-05, 9.5531e-05, 7.7870e-05], device='cuda:0') 2023-03-08 23:14:24,611 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82702.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:14:38,805 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=82711.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:15:02,606 INFO [train2.py:809] (0/4) Epoch 21, batch 3050, loss[ctc_loss=0.06529, att_loss=0.2289, loss=0.1962, over 16275.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.00763, over 43.00 utterances.], tot_loss[ctc_loss=0.07382, att_loss=0.2353, loss=0.203, over 3266595.97 frames. utt_duration=1281 frames, utt_pad_proportion=0.04948, over 10209.06 utterances.], batch size: 43, lr: 5.10e-03, grad_scale: 16.0 2023-03-08 23:15:14,881 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-03-08 23:15:26,169 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 8.870e+01 1.786e+02 2.187e+02 2.749e+02 5.841e+02, threshold=4.374e+02, percent-clipped=4.0 2023-03-08 23:15:43,266 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1087, 3.7658, 3.1508, 3.3852, 3.8894, 3.6091, 3.0152, 4.2290], device='cuda:0'), covar=tensor([0.0960, 0.0466, 0.1080, 0.0674, 0.0694, 0.0729, 0.0854, 0.0378], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0218, 0.0227, 0.0200, 0.0279, 0.0242, 0.0200, 0.0284], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:15:55,320 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=82759.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:16:22,508 INFO [train2.py:809] (0/4) Epoch 21, batch 3100, loss[ctc_loss=0.0718, att_loss=0.2292, loss=0.1977, over 16378.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.008931, over 44.00 utterances.], tot_loss[ctc_loss=0.07409, att_loss=0.2359, loss=0.2036, over 3266122.45 frames. utt_duration=1247 frames, utt_pad_proportion=0.05811, over 10491.53 utterances.], batch size: 44, lr: 5.10e-03, grad_scale: 16.0 2023-03-08 23:17:04,236 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0638, 4.9882, 4.7325, 2.7929, 4.8129, 4.6441, 4.2202, 2.6375], device='cuda:0'), covar=tensor([0.0097, 0.0108, 0.0338, 0.1028, 0.0106, 0.0194, 0.0337, 0.1380], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0102, 0.0104, 0.0111, 0.0085, 0.0113, 0.0099, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-08 23:17:43,782 INFO [train2.py:809] (0/4) Epoch 21, batch 3150, loss[ctc_loss=0.07559, att_loss=0.2299, loss=0.199, over 16105.00 frames. utt_duration=1535 frames, utt_pad_proportion=0.007649, over 42.00 utterances.], tot_loss[ctc_loss=0.07467, att_loss=0.237, loss=0.2046, over 3275831.82 frames. utt_duration=1237 frames, utt_pad_proportion=0.05677, over 10609.29 utterances.], batch size: 42, lr: 5.10e-03, grad_scale: 8.0 2023-03-08 23:18:09,384 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.159e+02 2.205e+02 2.472e+02 3.065e+02 6.005e+02, threshold=4.943e+02, percent-clipped=5.0 2023-03-08 23:18:21,034 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82849.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 23:19:03,804 INFO [train2.py:809] (0/4) Epoch 21, batch 3200, loss[ctc_loss=0.07461, att_loss=0.2475, loss=0.2129, over 16966.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.00767, over 50.00 utterances.], tot_loss[ctc_loss=0.07478, att_loss=0.2371, loss=0.2047, over 3277830.31 frames. utt_duration=1217 frames, utt_pad_proportion=0.06101, over 10783.65 utterances.], batch size: 50, lr: 5.09e-03, grad_scale: 8.0 2023-03-08 23:20:23,179 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=82925.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:20:24,597 INFO [train2.py:809] (0/4) Epoch 21, batch 3250, loss[ctc_loss=0.06683, att_loss=0.2092, loss=0.1807, over 14581.00 frames. utt_duration=1824 frames, utt_pad_proportion=0.04451, over 32.00 utterances.], tot_loss[ctc_loss=0.07507, att_loss=0.2366, loss=0.2043, over 3270887.97 frames. utt_duration=1200 frames, utt_pad_proportion=0.06743, over 10919.13 utterances.], batch size: 32, lr: 5.09e-03, grad_scale: 8.0 2023-03-08 23:20:33,704 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.57 vs. limit=2.0 2023-03-08 23:20:50,592 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 1.807e+02 2.273e+02 2.911e+02 7.187e+02, threshold=4.546e+02, percent-clipped=3.0 2023-03-08 23:21:06,048 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0484, 5.1370, 4.8689, 2.1590, 2.0062, 2.8918, 2.6872, 3.8540], device='cuda:0'), covar=tensor([0.0768, 0.0309, 0.0253, 0.5584, 0.5684, 0.2567, 0.3296, 0.1742], device='cuda:0'), in_proj_covar=tensor([0.0360, 0.0277, 0.0271, 0.0248, 0.0347, 0.0335, 0.0253, 0.0367], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 23:21:22,247 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-08 23:21:24,853 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=82963.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:21:45,461 INFO [train2.py:809] (0/4) Epoch 21, batch 3300, loss[ctc_loss=0.06361, att_loss=0.2311, loss=0.1976, over 16952.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008473, over 50.00 utterances.], tot_loss[ctc_loss=0.07355, att_loss=0.2353, loss=0.2029, over 3269361.98 frames. utt_duration=1247 frames, utt_pad_proportion=0.0571, over 10501.11 utterances.], batch size: 50, lr: 5.09e-03, grad_scale: 8.0 2023-03-08 23:22:28,778 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83002.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:22:41,861 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.02 vs. limit=5.0 2023-03-08 23:23:04,015 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83024.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:23:07,445 INFO [train2.py:809] (0/4) Epoch 21, batch 3350, loss[ctc_loss=0.07516, att_loss=0.2506, loss=0.2155, over 17363.00 frames. utt_duration=1179 frames, utt_pad_proportion=0.02022, over 59.00 utterances.], tot_loss[ctc_loss=0.07365, att_loss=0.2355, loss=0.2031, over 3268679.99 frames. utt_duration=1230 frames, utt_pad_proportion=0.06065, over 10638.88 utterances.], batch size: 59, lr: 5.09e-03, grad_scale: 8.0 2023-03-08 23:23:33,599 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 1.819e+02 2.175e+02 2.592e+02 7.580e+02, threshold=4.350e+02, percent-clipped=2.0 2023-03-08 23:23:45,982 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83050.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:24:11,469 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6369, 2.4407, 2.5733, 2.6563, 2.7091, 2.6939, 2.2255, 3.1138], device='cuda:0'), covar=tensor([0.1517, 0.2381, 0.1810, 0.1239, 0.1790, 0.1132, 0.2071, 0.1057], device='cuda:0'), in_proj_covar=tensor([0.0117, 0.0124, 0.0121, 0.0108, 0.0123, 0.0106, 0.0128, 0.0099], device='cuda:0'), out_proj_covar=tensor([8.8513e-05, 9.6031e-05, 9.5817e-05, 8.4558e-05, 9.1557e-05, 8.4760e-05, 9.6630e-05, 7.8862e-05], device='cuda:0') 2023-03-08 23:24:21,588 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0778, 4.3730, 3.9653, 4.4763, 2.7231, 4.3426, 2.4128, 1.7261], device='cuda:0'), covar=tensor([0.0431, 0.0216, 0.0814, 0.0237, 0.1484, 0.0217, 0.1476, 0.1660], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0163, 0.0254, 0.0154, 0.0217, 0.0142, 0.0226, 0.0200], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:24:28,080 INFO [train2.py:809] (0/4) Epoch 21, batch 3400, loss[ctc_loss=0.07709, att_loss=0.2498, loss=0.2153, over 17288.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01172, over 55.00 utterances.], tot_loss[ctc_loss=0.07385, att_loss=0.236, loss=0.2036, over 3279157.22 frames. utt_duration=1226 frames, utt_pad_proportion=0.0595, over 10708.67 utterances.], batch size: 55, lr: 5.09e-03, grad_scale: 8.0 2023-03-08 23:25:28,381 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.35 vs. limit=5.0 2023-03-08 23:25:47,347 INFO [train2.py:809] (0/4) Epoch 21, batch 3450, loss[ctc_loss=0.06681, att_loss=0.2307, loss=0.198, over 17372.00 frames. utt_duration=1009 frames, utt_pad_proportion=0.0485, over 69.00 utterances.], tot_loss[ctc_loss=0.07335, att_loss=0.2356, loss=0.2032, over 3281375.82 frames. utt_duration=1262 frames, utt_pad_proportion=0.05037, over 10409.68 utterances.], batch size: 69, lr: 5.09e-03, grad_scale: 8.0 2023-03-08 23:26:13,623 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 1.851e+02 2.238e+02 2.723e+02 5.748e+02, threshold=4.475e+02, percent-clipped=3.0 2023-03-08 23:26:23,076 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5408, 2.7178, 5.0516, 3.9167, 3.0123, 4.2156, 4.7823, 4.5801], device='cuda:0'), covar=tensor([0.0244, 0.1585, 0.0171, 0.1028, 0.1713, 0.0271, 0.0152, 0.0266], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0239, 0.0184, 0.0308, 0.0262, 0.0211, 0.0172, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:26:24,486 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83149.0, num_to_drop=1, layers_to_drop={2} 2023-03-08 23:26:54,077 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.2431, 6.3524, 5.8680, 6.1488, 6.1260, 5.5803, 5.9658, 5.5906], device='cuda:0'), covar=tensor([0.0888, 0.0793, 0.0744, 0.0678, 0.0707, 0.1256, 0.1686, 0.2038], device='cuda:0'), in_proj_covar=tensor([0.0518, 0.0602, 0.0453, 0.0453, 0.0421, 0.0460, 0.0599, 0.0518], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 23:27:06,900 INFO [train2.py:809] (0/4) Epoch 21, batch 3500, loss[ctc_loss=0.1311, att_loss=0.2673, loss=0.24, over 16806.00 frames. utt_duration=680.6 frames, utt_pad_proportion=0.1472, over 99.00 utterances.], tot_loss[ctc_loss=0.07393, att_loss=0.2355, loss=0.2032, over 3268584.33 frames. utt_duration=1226 frames, utt_pad_proportion=0.06206, over 10681.19 utterances.], batch size: 99, lr: 5.09e-03, grad_scale: 8.0 2023-03-08 23:27:39,115 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83196.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:27:40,305 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83197.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 23:28:16,251 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-08 23:28:25,759 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83225.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:28:26,903 INFO [train2.py:809] (0/4) Epoch 21, batch 3550, loss[ctc_loss=0.06903, att_loss=0.2347, loss=0.2016, over 16878.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007693, over 49.00 utterances.], tot_loss[ctc_loss=0.07374, att_loss=0.2351, loss=0.2028, over 3262522.35 frames. utt_duration=1223 frames, utt_pad_proportion=0.06404, over 10680.99 utterances.], batch size: 49, lr: 5.08e-03, grad_scale: 8.0 2023-03-08 23:28:53,067 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 1.836e+02 2.204e+02 2.637e+02 6.921e+02, threshold=4.407e+02, percent-clipped=5.0 2023-03-08 23:29:16,521 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83257.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:29:41,888 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83273.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:29:42,257 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5340, 2.8452, 5.0956, 4.1168, 3.0128, 4.3726, 4.8889, 4.6389], device='cuda:0'), covar=tensor([0.0316, 0.1481, 0.0228, 0.0890, 0.1659, 0.0256, 0.0170, 0.0294], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0240, 0.0184, 0.0308, 0.0263, 0.0212, 0.0173, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:29:47,126 INFO [train2.py:809] (0/4) Epoch 21, batch 3600, loss[ctc_loss=0.05705, att_loss=0.2381, loss=0.2019, over 16889.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006303, over 49.00 utterances.], tot_loss[ctc_loss=0.07327, att_loss=0.2351, loss=0.2027, over 3262093.66 frames. utt_duration=1237 frames, utt_pad_proportion=0.06002, over 10562.10 utterances.], batch size: 49, lr: 5.08e-03, grad_scale: 8.0 2023-03-08 23:30:00,045 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83284.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:30:56,574 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83319.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:31:07,606 INFO [train2.py:809] (0/4) Epoch 21, batch 3650, loss[ctc_loss=0.07797, att_loss=0.2206, loss=0.1921, over 15383.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01046, over 35.00 utterances.], tot_loss[ctc_loss=0.07368, att_loss=0.2356, loss=0.2032, over 3267764.94 frames. utt_duration=1230 frames, utt_pad_proportion=0.05925, over 10636.52 utterances.], batch size: 35, lr: 5.08e-03, grad_scale: 8.0 2023-03-08 23:31:33,947 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 2.085e+02 2.475e+02 3.033e+02 6.314e+02, threshold=4.951e+02, percent-clipped=5.0 2023-03-08 23:31:39,070 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83345.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:31:55,815 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2081, 3.7459, 3.1941, 3.3913, 3.9920, 3.6355, 3.1329, 4.3674], device='cuda:0'), covar=tensor([0.0911, 0.0502, 0.1005, 0.0725, 0.0696, 0.0752, 0.0814, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0216, 0.0225, 0.0198, 0.0276, 0.0239, 0.0198, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:32:27,571 INFO [train2.py:809] (0/4) Epoch 21, batch 3700, loss[ctc_loss=0.0615, att_loss=0.2346, loss=0.2, over 16481.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006459, over 46.00 utterances.], tot_loss[ctc_loss=0.07392, att_loss=0.2355, loss=0.2032, over 3264274.02 frames. utt_duration=1211 frames, utt_pad_proportion=0.06556, over 10798.46 utterances.], batch size: 46, lr: 5.08e-03, grad_scale: 8.0 2023-03-08 23:33:49,259 INFO [train2.py:809] (0/4) Epoch 21, batch 3750, loss[ctc_loss=0.05913, att_loss=0.2228, loss=0.1901, over 16179.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006182, over 41.00 utterances.], tot_loss[ctc_loss=0.07373, att_loss=0.2351, loss=0.2028, over 3264290.18 frames. utt_duration=1220 frames, utt_pad_proportion=0.06372, over 10717.78 utterances.], batch size: 41, lr: 5.08e-03, grad_scale: 8.0 2023-03-08 23:33:49,703 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6005, 4.6304, 4.6696, 4.6793, 5.2787, 4.6077, 4.7062, 2.9722], device='cuda:0'), covar=tensor([0.0244, 0.0325, 0.0325, 0.0333, 0.0809, 0.0230, 0.0313, 0.1589], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0189, 0.0188, 0.0205, 0.0370, 0.0158, 0.0178, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:34:15,117 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 1.925e+02 2.407e+02 3.142e+02 5.755e+02, threshold=4.814e+02, percent-clipped=4.0 2023-03-08 23:35:09,001 INFO [train2.py:809] (0/4) Epoch 21, batch 3800, loss[ctc_loss=0.06296, att_loss=0.2048, loss=0.1765, over 15524.00 frames. utt_duration=1726 frames, utt_pad_proportion=0.006713, over 36.00 utterances.], tot_loss[ctc_loss=0.07441, att_loss=0.2361, loss=0.2038, over 3278262.49 frames. utt_duration=1237 frames, utt_pad_proportion=0.05557, over 10612.36 utterances.], batch size: 36, lr: 5.08e-03, grad_scale: 8.0 2023-03-08 23:36:29,154 INFO [train2.py:809] (0/4) Epoch 21, batch 3850, loss[ctc_loss=0.05796, att_loss=0.2085, loss=0.1784, over 14561.00 frames. utt_duration=1822 frames, utt_pad_proportion=0.03005, over 32.00 utterances.], tot_loss[ctc_loss=0.07397, att_loss=0.2359, loss=0.2035, over 3273623.87 frames. utt_duration=1218 frames, utt_pad_proportion=0.06068, over 10762.55 utterances.], batch size: 32, lr: 5.07e-03, grad_scale: 8.0 2023-03-08 23:36:49,445 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7472, 4.9814, 4.5843, 5.0689, 4.4250, 4.6975, 5.0892, 4.8849], device='cuda:0'), covar=tensor([0.0540, 0.0370, 0.0781, 0.0385, 0.0471, 0.0302, 0.0303, 0.0213], device='cuda:0'), in_proj_covar=tensor([0.0382, 0.0317, 0.0361, 0.0345, 0.0319, 0.0235, 0.0299, 0.0282], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 23:36:53,728 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 2.030e+02 2.457e+02 3.139e+02 8.463e+02, threshold=4.915e+02, percent-clipped=7.0 2023-03-08 23:37:09,094 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-03-08 23:37:09,491 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83552.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:37:28,348 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5994, 3.0194, 3.7236, 3.1868, 3.6697, 4.6655, 4.5174, 3.4656], device='cuda:0'), covar=tensor([0.0329, 0.1596, 0.1209, 0.1238, 0.0972, 0.0835, 0.0512, 0.1165], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0241, 0.0278, 0.0219, 0.0263, 0.0365, 0.0259, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 23:37:46,944 INFO [train2.py:809] (0/4) Epoch 21, batch 3900, loss[ctc_loss=0.05811, att_loss=0.2078, loss=0.1778, over 15478.00 frames. utt_duration=1721 frames, utt_pad_proportion=0.01008, over 36.00 utterances.], tot_loss[ctc_loss=0.07342, att_loss=0.236, loss=0.2035, over 3271914.28 frames. utt_duration=1231 frames, utt_pad_proportion=0.05729, over 10645.79 utterances.], batch size: 36, lr: 5.07e-03, grad_scale: 8.0 2023-03-08 23:38:17,238 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2434, 4.5402, 4.5121, 4.8600, 3.0449, 4.6504, 2.6937, 1.6453], device='cuda:0'), covar=tensor([0.0368, 0.0264, 0.0627, 0.0211, 0.1350, 0.0201, 0.1382, 0.1759], device='cuda:0'), in_proj_covar=tensor([0.0192, 0.0165, 0.0257, 0.0157, 0.0220, 0.0145, 0.0227, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:38:22,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-08 23:38:29,024 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-03-08 23:38:55,360 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83619.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:39:05,568 INFO [train2.py:809] (0/4) Epoch 21, batch 3950, loss[ctc_loss=0.06879, att_loss=0.2339, loss=0.2009, over 16883.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.006876, over 49.00 utterances.], tot_loss[ctc_loss=0.07347, att_loss=0.2354, loss=0.203, over 3268727.45 frames. utt_duration=1231 frames, utt_pad_proportion=0.06006, over 10630.86 utterances.], batch size: 49, lr: 5.07e-03, grad_scale: 8.0 2023-03-08 23:39:27,331 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=83640.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:39:30,215 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.299e+02 1.844e+02 2.165e+02 2.793e+02 5.808e+02, threshold=4.331e+02, percent-clipped=4.0 2023-03-08 23:39:55,518 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-21.pt 2023-03-08 23:40:16,718 INFO [train2.py:809] (0/4) Epoch 22, batch 0, loss[ctc_loss=0.06325, att_loss=0.2277, loss=0.1948, over 17347.00 frames. utt_duration=1178 frames, utt_pad_proportion=0.02106, over 59.00 utterances.], tot_loss[ctc_loss=0.06325, att_loss=0.2277, loss=0.1948, over 17347.00 frames. utt_duration=1178 frames, utt_pad_proportion=0.02106, over 59.00 utterances.], batch size: 59, lr: 4.95e-03, grad_scale: 8.0 2023-03-08 23:40:16,720 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-08 23:40:29,654 INFO [train2.py:843] (0/4) Epoch 22, validation: ctc_loss=0.04004, att_loss=0.2341, loss=0.1953, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-08 23:40:29,655 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-08 23:40:40,584 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4373, 4.6151, 4.5925, 4.5349, 4.6425, 4.6385, 4.3357, 4.1807], device='cuda:0'), covar=tensor([0.0945, 0.0584, 0.0407, 0.0652, 0.0342, 0.0381, 0.0468, 0.0408], device='cuda:0'), in_proj_covar=tensor([0.0516, 0.0362, 0.0342, 0.0356, 0.0418, 0.0429, 0.0353, 0.0392], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:40:41,954 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83667.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:41:48,593 INFO [train2.py:809] (0/4) Epoch 22, batch 50, loss[ctc_loss=0.06791, att_loss=0.2396, loss=0.2052, over 16871.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007958, over 49.00 utterances.], tot_loss[ctc_loss=0.07529, att_loss=0.2383, loss=0.2057, over 745387.97 frames. utt_duration=1133 frames, utt_pad_proportion=0.07345, over 2634.85 utterances.], batch size: 49, lr: 4.95e-03, grad_scale: 8.0 2023-03-08 23:42:40,552 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.187e+02 1.991e+02 2.354e+02 2.967e+02 6.077e+02, threshold=4.708e+02, percent-clipped=4.0 2023-03-08 23:42:43,902 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9747, 6.1730, 5.6645, 5.9436, 5.8667, 5.3015, 5.6201, 5.3897], device='cuda:0'), covar=tensor([0.1212, 0.0911, 0.0881, 0.0811, 0.0903, 0.1668, 0.2305, 0.2337], device='cuda:0'), in_proj_covar=tensor([0.0522, 0.0602, 0.0455, 0.0452, 0.0422, 0.0463, 0.0602, 0.0518], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003], device='cuda:0') 2023-03-08 23:43:01,761 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6712, 5.0285, 4.9670, 4.9552, 5.0787, 4.7635, 3.5915, 4.9859], device='cuda:0'), covar=tensor([0.0114, 0.0118, 0.0106, 0.0083, 0.0089, 0.0103, 0.0625, 0.0174], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0086, 0.0109, 0.0069, 0.0075, 0.0085, 0.0103, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 23:43:08,483 INFO [train2.py:809] (0/4) Epoch 22, batch 100, loss[ctc_loss=0.05993, att_loss=0.2392, loss=0.2034, over 17297.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01211, over 55.00 utterances.], tot_loss[ctc_loss=0.07522, att_loss=0.2365, loss=0.2042, over 1299329.67 frames. utt_duration=1183 frames, utt_pad_proportion=0.06841, over 4397.89 utterances.], batch size: 55, lr: 4.95e-03, grad_scale: 8.0 2023-03-08 23:43:10,916 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83760.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:43:48,887 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7447, 4.9650, 4.9493, 4.8703, 4.9977, 4.9643, 4.6757, 4.4725], device='cuda:0'), covar=tensor([0.0996, 0.0551, 0.0311, 0.0554, 0.0316, 0.0355, 0.0374, 0.0394], device='cuda:0'), in_proj_covar=tensor([0.0516, 0.0361, 0.0343, 0.0355, 0.0418, 0.0429, 0.0353, 0.0391], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:44:29,210 INFO [train2.py:809] (0/4) Epoch 22, batch 150, loss[ctc_loss=0.09936, att_loss=0.2558, loss=0.2245, over 17056.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008061, over 52.00 utterances.], tot_loss[ctc_loss=0.07501, att_loss=0.2357, loss=0.2036, over 1735864.03 frames. utt_duration=1208 frames, utt_pad_proportion=0.06318, over 5753.36 utterances.], batch size: 52, lr: 4.95e-03, grad_scale: 8.0 2023-03-08 23:44:48,402 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=83821.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:45:21,311 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 1.932e+02 2.277e+02 2.985e+02 7.139e+02, threshold=4.555e+02, percent-clipped=2.0 2023-03-08 23:45:37,566 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83852.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:45:37,622 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6075, 4.2382, 3.6031, 3.8129, 4.3459, 4.0186, 3.4787, 4.6072], device='cuda:0'), covar=tensor([0.0809, 0.0401, 0.0834, 0.0570, 0.0530, 0.0574, 0.0742, 0.0413], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0215, 0.0227, 0.0198, 0.0277, 0.0241, 0.0199, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:45:49,309 INFO [train2.py:809] (0/4) Epoch 22, batch 200, loss[ctc_loss=0.07107, att_loss=0.2253, loss=0.1945, over 16175.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006796, over 41.00 utterances.], tot_loss[ctc_loss=0.07472, att_loss=0.2363, loss=0.204, over 2073333.59 frames. utt_duration=1214 frames, utt_pad_proportion=0.06204, over 6840.26 utterances.], batch size: 41, lr: 4.95e-03, grad_scale: 8.0 2023-03-08 23:46:28,853 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7166, 5.0648, 5.3329, 5.1076, 5.2125, 5.6654, 5.1014, 5.7759], device='cuda:0'), covar=tensor([0.0859, 0.0891, 0.0816, 0.1321, 0.1896, 0.1066, 0.0860, 0.0686], device='cuda:0'), in_proj_covar=tensor([0.0881, 0.0517, 0.0599, 0.0669, 0.0878, 0.0631, 0.0490, 0.0609], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-08 23:46:53,869 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83900.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:46:55,844 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2661, 4.0196, 3.4160, 3.5239, 4.1744, 3.7755, 3.1317, 4.4991], device='cuda:0'), covar=tensor([0.0913, 0.0473, 0.0925, 0.0635, 0.0594, 0.0675, 0.0756, 0.0430], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0217, 0.0229, 0.0199, 0.0279, 0.0243, 0.0200, 0.0288], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:47:08,783 INFO [train2.py:809] (0/4) Epoch 22, batch 250, loss[ctc_loss=0.07437, att_loss=0.2178, loss=0.1891, over 15499.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008721, over 36.00 utterances.], tot_loss[ctc_loss=0.07407, att_loss=0.2356, loss=0.2033, over 2338185.45 frames. utt_duration=1234 frames, utt_pad_proportion=0.05674, over 7589.66 utterances.], batch size: 36, lr: 4.94e-03, grad_scale: 8.0 2023-03-08 23:47:57,194 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=83940.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:47:59,820 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.391e+02 1.880e+02 2.225e+02 2.668e+02 4.764e+02, threshold=4.450e+02, percent-clipped=1.0 2023-03-08 23:48:11,495 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83949.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:48:27,880 INFO [train2.py:809] (0/4) Epoch 22, batch 300, loss[ctc_loss=0.0704, att_loss=0.2411, loss=0.207, over 17407.00 frames. utt_duration=882.9 frames, utt_pad_proportion=0.07355, over 79.00 utterances.], tot_loss[ctc_loss=0.07428, att_loss=0.2356, loss=0.2034, over 2550177.00 frames. utt_duration=1238 frames, utt_pad_proportion=0.054, over 8249.47 utterances.], batch size: 79, lr: 4.94e-03, grad_scale: 8.0 2023-03-08 23:48:44,311 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.54 vs. limit=5.0 2023-03-08 23:48:46,386 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-08 23:48:55,408 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0885, 5.1533, 5.0183, 2.3567, 2.0702, 3.1133, 2.3108, 3.9539], device='cuda:0'), covar=tensor([0.0753, 0.0289, 0.0243, 0.5399, 0.5898, 0.2403, 0.4122, 0.1626], device='cuda:0'), in_proj_covar=tensor([0.0360, 0.0280, 0.0272, 0.0246, 0.0345, 0.0335, 0.0256, 0.0365], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-08 23:49:09,310 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=83985.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:49:13,753 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=83988.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:49:33,159 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-84000.pt 2023-03-08 23:49:51,797 INFO [train2.py:809] (0/4) Epoch 22, batch 350, loss[ctc_loss=0.1343, att_loss=0.2676, loss=0.2409, over 14177.00 frames. utt_duration=389.9 frames, utt_pad_proportion=0.3184, over 146.00 utterances.], tot_loss[ctc_loss=0.07426, att_loss=0.2356, loss=0.2033, over 2712264.22 frames. utt_duration=1244 frames, utt_pad_proportion=0.05359, over 8729.56 utterances.], batch size: 146, lr: 4.94e-03, grad_scale: 8.0 2023-03-08 23:49:53,700 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84010.0, num_to_drop=1, layers_to_drop={0} 2023-03-08 23:50:25,441 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.21 vs. limit=5.0 2023-03-08 23:50:41,422 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-08 23:50:43,094 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.569e+02 1.981e+02 2.404e+02 2.952e+02 5.790e+02, threshold=4.808e+02, percent-clipped=6.0 2023-03-08 23:50:50,320 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84046.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:51:11,737 INFO [train2.py:809] (0/4) Epoch 22, batch 400, loss[ctc_loss=0.08304, att_loss=0.2406, loss=0.2091, over 16283.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007082, over 43.00 utterances.], tot_loss[ctc_loss=0.07335, att_loss=0.2351, loss=0.2027, over 2836027.15 frames. utt_duration=1267 frames, utt_pad_proportion=0.04867, over 8960.99 utterances.], batch size: 43, lr: 4.94e-03, grad_scale: 8.0 2023-03-08 23:51:55,958 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3430, 5.5842, 5.2717, 5.6877, 5.1085, 5.2246, 5.7053, 5.5023], device='cuda:0'), covar=tensor([0.0492, 0.0272, 0.0594, 0.0237, 0.0351, 0.0243, 0.0196, 0.0143], device='cuda:0'), in_proj_covar=tensor([0.0385, 0.0319, 0.0362, 0.0348, 0.0321, 0.0237, 0.0300, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-08 23:52:12,372 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84097.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:52:31,752 INFO [train2.py:809] (0/4) Epoch 22, batch 450, loss[ctc_loss=0.07589, att_loss=0.2257, loss=0.1958, over 16163.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006554, over 41.00 utterances.], tot_loss[ctc_loss=0.07418, att_loss=0.2362, loss=0.2038, over 2936109.45 frames. utt_duration=1228 frames, utt_pad_proportion=0.05707, over 9574.69 utterances.], batch size: 41, lr: 4.94e-03, grad_scale: 8.0 2023-03-08 23:52:42,546 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84116.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:53:22,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 1.943e+02 2.393e+02 2.892e+02 4.484e+02, threshold=4.787e+02, percent-clipped=0.0 2023-03-08 23:53:44,913 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7020, 3.5150, 3.3778, 2.9560, 3.4045, 3.4181, 3.4732, 2.4689], device='cuda:0'), covar=tensor([0.1270, 0.1419, 0.2054, 0.3771, 0.1720, 0.3236, 0.1400, 0.4000], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0190, 0.0201, 0.0256, 0.0159, 0.0261, 0.0181, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-08 23:53:50,181 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84158.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:53:51,330 INFO [train2.py:809] (0/4) Epoch 22, batch 500, loss[ctc_loss=0.05674, att_loss=0.2196, loss=0.187, over 16404.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006638, over 44.00 utterances.], tot_loss[ctc_loss=0.07353, att_loss=0.2355, loss=0.2031, over 3009289.30 frames. utt_duration=1245 frames, utt_pad_proportion=0.05521, over 9680.99 utterances.], batch size: 44, lr: 4.94e-03, grad_scale: 8.0 2023-03-08 23:54:22,001 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5831, 2.3775, 2.1343, 2.4999, 2.7372, 2.8019, 2.4425, 3.2811], device='cuda:0'), covar=tensor([0.1612, 0.2396, 0.2636, 0.1387, 0.1549, 0.0982, 0.2103, 0.0943], device='cuda:0'), in_proj_covar=tensor([0.0118, 0.0124, 0.0123, 0.0109, 0.0123, 0.0106, 0.0128, 0.0100], device='cuda:0'), out_proj_covar=tensor([8.9076e-05, 9.6379e-05, 9.7094e-05, 8.5473e-05, 9.1514e-05, 8.5115e-05, 9.7062e-05, 7.9703e-05], device='cuda:0') 2023-03-08 23:54:34,880 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.67 vs. limit=5.0 2023-03-08 23:54:55,828 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84199.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:55:12,695 INFO [train2.py:809] (0/4) Epoch 22, batch 550, loss[ctc_loss=0.05787, att_loss=0.2079, loss=0.1779, over 15620.00 frames. utt_duration=1690 frames, utt_pad_proportion=0.00988, over 37.00 utterances.], tot_loss[ctc_loss=0.07323, att_loss=0.2349, loss=0.2026, over 3068872.78 frames. utt_duration=1254 frames, utt_pad_proportion=0.05284, over 9798.49 utterances.], batch size: 37, lr: 4.94e-03, grad_scale: 8.0 2023-03-08 23:55:47,280 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-08 23:55:55,720 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84236.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:56:03,700 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3288, 3.9153, 3.3194, 3.5415, 4.0721, 3.8230, 3.1001, 4.4711], device='cuda:0'), covar=tensor([0.0922, 0.0462, 0.1074, 0.0753, 0.0686, 0.0664, 0.0817, 0.0468], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0217, 0.0227, 0.0200, 0.0279, 0.0241, 0.0199, 0.0287], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-08 23:56:04,933 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.221e+02 1.935e+02 2.218e+02 2.633e+02 6.667e+02, threshold=4.436e+02, percent-clipped=1.0 2023-03-08 23:56:25,653 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.92 vs. limit=5.0 2023-03-08 23:56:33,888 INFO [train2.py:809] (0/4) Epoch 22, batch 600, loss[ctc_loss=0.05983, att_loss=0.2294, loss=0.1955, over 16549.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005075, over 45.00 utterances.], tot_loss[ctc_loss=0.07316, att_loss=0.2347, loss=0.2024, over 3111065.29 frames. utt_duration=1258 frames, utt_pad_proportion=0.05329, over 9906.95 utterances.], batch size: 45, lr: 4.93e-03, grad_scale: 8.0 2023-03-08 23:56:35,681 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84260.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:56:49,812 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5628, 3.0002, 3.7002, 2.8406, 3.5813, 4.6252, 4.4164, 3.3553], device='cuda:0'), covar=tensor([0.0337, 0.1612, 0.1160, 0.1493, 0.1099, 0.0807, 0.0627, 0.1122], device='cuda:0'), in_proj_covar=tensor([0.0245, 0.0241, 0.0280, 0.0219, 0.0263, 0.0366, 0.0259, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-08 23:57:35,181 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84297.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:57:49,018 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84305.0, num_to_drop=1, layers_to_drop={3} 2023-03-08 23:57:55,133 INFO [train2.py:809] (0/4) Epoch 22, batch 650, loss[ctc_loss=0.07477, att_loss=0.2431, loss=0.2094, over 17321.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02337, over 59.00 utterances.], tot_loss[ctc_loss=0.07431, att_loss=0.2356, loss=0.2033, over 3140617.65 frames. utt_duration=1204 frames, utt_pad_proportion=0.06774, over 10445.77 utterances.], batch size: 59, lr: 4.93e-03, grad_scale: 8.0 2023-03-08 23:58:38,264 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84336.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:58:45,779 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84341.0, num_to_drop=0, layers_to_drop=set() 2023-03-08 23:58:47,105 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.197e+02 2.036e+02 2.451e+02 3.016e+02 6.584e+02, threshold=4.901e+02, percent-clipped=3.0 2023-03-08 23:59:16,010 INFO [train2.py:809] (0/4) Epoch 22, batch 700, loss[ctc_loss=0.09649, att_loss=0.2565, loss=0.2245, over 16735.00 frames. utt_duration=684.6 frames, utt_pad_proportion=0.1421, over 98.00 utterances.], tot_loss[ctc_loss=0.07435, att_loss=0.2355, loss=0.2033, over 3164374.65 frames. utt_duration=1205 frames, utt_pad_proportion=0.06775, over 10515.07 utterances.], batch size: 98, lr: 4.93e-03, grad_scale: 8.0 2023-03-09 00:00:15,059 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.3884, 2.2026, 1.8956, 2.4980, 2.7096, 2.5251, 2.2009, 3.0043], device='cuda:0'), covar=tensor([0.1849, 0.2732, 0.2788, 0.1554, 0.1510, 0.1376, 0.2603, 0.1143], device='cuda:0'), in_proj_covar=tensor([0.0120, 0.0126, 0.0125, 0.0111, 0.0124, 0.0108, 0.0131, 0.0100], device='cuda:0'), out_proj_covar=tensor([9.0380e-05, 9.7568e-05, 9.8547e-05, 8.6752e-05, 9.2774e-05, 8.6838e-05, 9.8608e-05, 8.0430e-05], device='cuda:0') 2023-03-09 00:00:17,476 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84397.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:00:28,864 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1467, 4.4829, 4.5986, 4.4616, 5.0805, 4.3910, 4.5243, 2.4674], device='cuda:0'), covar=tensor([0.0349, 0.0309, 0.0267, 0.0296, 0.0873, 0.0256, 0.0325, 0.1890], device='cuda:0'), in_proj_covar=tensor([0.0166, 0.0192, 0.0189, 0.0208, 0.0372, 0.0159, 0.0182, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:00:36,270 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7592, 4.8817, 5.5297, 5.0539, 4.8636, 5.5780, 5.0217, 5.6249], device='cuda:0'), covar=tensor([0.1333, 0.1930, 0.1150, 0.2594, 0.3377, 0.1920, 0.1288, 0.1515], device='cuda:0'), in_proj_covar=tensor([0.0873, 0.0519, 0.0602, 0.0670, 0.0878, 0.0626, 0.0488, 0.0614], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:00:37,603 INFO [train2.py:809] (0/4) Epoch 22, batch 750, loss[ctc_loss=0.07791, att_loss=0.248, loss=0.214, over 16786.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005534, over 48.00 utterances.], tot_loss[ctc_loss=0.07425, att_loss=0.2359, loss=0.2036, over 3196146.45 frames. utt_duration=1210 frames, utt_pad_proportion=0.06433, over 10578.58 utterances.], batch size: 48, lr: 4.93e-03, grad_scale: 8.0 2023-03-09 00:00:48,909 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84416.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:01:29,379 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 1.936e+02 2.365e+02 2.996e+02 6.300e+02, threshold=4.730e+02, percent-clipped=3.0 2023-03-09 00:01:48,722 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84453.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:01:57,858 INFO [train2.py:809] (0/4) Epoch 22, batch 800, loss[ctc_loss=0.06739, att_loss=0.2369, loss=0.203, over 16693.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.00561, over 46.00 utterances.], tot_loss[ctc_loss=0.07469, att_loss=0.2364, loss=0.2041, over 3222466.66 frames. utt_duration=1210 frames, utt_pad_proportion=0.06068, over 10664.63 utterances.], batch size: 46, lr: 4.93e-03, grad_scale: 8.0 2023-03-09 00:02:03,609 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-09 00:02:06,042 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84464.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:02:29,525 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-09 00:03:17,418 INFO [train2.py:809] (0/4) Epoch 22, batch 850, loss[ctc_loss=0.05728, att_loss=0.2107, loss=0.18, over 13605.00 frames. utt_duration=1816 frames, utt_pad_proportion=0.08844, over 30.00 utterances.], tot_loss[ctc_loss=0.07371, att_loss=0.2355, loss=0.2031, over 3221048.41 frames. utt_duration=1232 frames, utt_pad_proportion=0.05839, over 10473.64 utterances.], batch size: 30, lr: 4.93e-03, grad_scale: 8.0 2023-03-09 00:03:42,546 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84525.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:04:07,926 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.340e+02 1.954e+02 2.376e+02 2.798e+02 7.444e+02, threshold=4.752e+02, percent-clipped=4.0 2023-03-09 00:04:18,267 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2344, 5.1032, 4.8379, 2.9336, 4.9573, 4.6700, 4.5900, 2.9167], device='cuda:0'), covar=tensor([0.0085, 0.0094, 0.0334, 0.1017, 0.0101, 0.0212, 0.0256, 0.1259], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0102, 0.0105, 0.0111, 0.0085, 0.0114, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:04:30,062 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84555.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:04:36,029 INFO [train2.py:809] (0/4) Epoch 22, batch 900, loss[ctc_loss=0.08291, att_loss=0.2494, loss=0.2161, over 17381.00 frames. utt_duration=1180 frames, utt_pad_proportion=0.02003, over 59.00 utterances.], tot_loss[ctc_loss=0.07389, att_loss=0.2355, loss=0.2032, over 3238607.10 frames. utt_duration=1243 frames, utt_pad_proportion=0.05382, over 10432.31 utterances.], batch size: 59, lr: 4.93e-03, grad_scale: 8.0 2023-03-09 00:05:18,237 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84586.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:05:22,770 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3096, 5.1795, 4.9484, 3.1803, 5.0450, 4.7261, 4.6111, 2.9144], device='cuda:0'), covar=tensor([0.0072, 0.0084, 0.0256, 0.0856, 0.0086, 0.0192, 0.0242, 0.1189], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0101, 0.0104, 0.0111, 0.0085, 0.0114, 0.0099, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:05:27,580 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84592.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:05:50,193 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84605.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:05:55,901 INFO [train2.py:809] (0/4) Epoch 22, batch 950, loss[ctc_loss=0.05524, att_loss=0.2185, loss=0.1858, over 15759.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009295, over 38.00 utterances.], tot_loss[ctc_loss=0.07356, att_loss=0.2352, loss=0.2029, over 3242251.15 frames. utt_duration=1226 frames, utt_pad_proportion=0.05933, over 10589.24 utterances.], batch size: 38, lr: 4.92e-03, grad_scale: 8.0 2023-03-09 00:06:05,565 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4694, 4.6656, 4.3024, 4.7176, 4.2666, 4.3718, 4.7462, 4.6060], device='cuda:0'), covar=tensor([0.0586, 0.0330, 0.0723, 0.0380, 0.0422, 0.0445, 0.0264, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0390, 0.0323, 0.0368, 0.0351, 0.0324, 0.0238, 0.0303, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 00:06:28,302 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.35 vs. limit=2.0 2023-03-09 00:06:33,723 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84633.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:06:46,125 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84641.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:06:47,277 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.223e+02 1.899e+02 2.395e+02 3.054e+02 5.479e+02, threshold=4.790e+02, percent-clipped=1.0 2023-03-09 00:07:06,385 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84653.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:07:15,607 INFO [train2.py:809] (0/4) Epoch 22, batch 1000, loss[ctc_loss=0.0867, att_loss=0.2493, loss=0.2168, over 17057.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008679, over 52.00 utterances.], tot_loss[ctc_loss=0.07412, att_loss=0.2362, loss=0.2038, over 3245840.12 frames. utt_duration=1195 frames, utt_pad_proportion=0.06677, over 10877.23 utterances.], batch size: 52, lr: 4.92e-03, grad_scale: 8.0 2023-03-09 00:07:51,112 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.69 vs. limit=2.0 2023-03-09 00:08:02,348 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84689.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:08:05,393 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7049, 5.9555, 5.4134, 5.7088, 5.6284, 5.1356, 5.4320, 5.1126], device='cuda:0'), covar=tensor([0.1288, 0.0912, 0.0855, 0.0793, 0.0911, 0.1467, 0.2275, 0.2257], device='cuda:0'), in_proj_covar=tensor([0.0524, 0.0615, 0.0461, 0.0457, 0.0429, 0.0466, 0.0608, 0.0526], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:08:06,996 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84692.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:08:10,927 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84694.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:08:34,842 INFO [train2.py:809] (0/4) Epoch 22, batch 1050, loss[ctc_loss=0.08315, att_loss=0.2267, loss=0.198, over 15939.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.008037, over 41.00 utterances.], tot_loss[ctc_loss=0.07479, att_loss=0.2362, loss=0.204, over 3245988.33 frames. utt_duration=1179 frames, utt_pad_proportion=0.07368, over 11023.69 utterances.], batch size: 41, lr: 4.92e-03, grad_scale: 8.0 2023-03-09 00:08:39,819 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7104, 4.7103, 4.7608, 4.6615, 5.2709, 4.6483, 4.7703, 2.8811], device='cuda:0'), covar=tensor([0.0219, 0.0316, 0.0274, 0.0343, 0.0683, 0.0227, 0.0287, 0.1561], device='cuda:0'), in_proj_covar=tensor([0.0165, 0.0191, 0.0189, 0.0206, 0.0368, 0.0158, 0.0181, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:09:25,764 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.058e+02 1.985e+02 2.384e+02 2.768e+02 5.690e+02, threshold=4.768e+02, percent-clipped=2.0 2023-03-09 00:09:45,224 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84753.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:09:54,285 INFO [train2.py:809] (0/4) Epoch 22, batch 1100, loss[ctc_loss=0.07685, att_loss=0.2383, loss=0.206, over 16792.00 frames. utt_duration=679.9 frames, utt_pad_proportion=0.1469, over 99.00 utterances.], tot_loss[ctc_loss=0.07503, att_loss=0.2357, loss=0.2036, over 3245899.50 frames. utt_duration=1170 frames, utt_pad_proportion=0.07713, over 11109.26 utterances.], batch size: 99, lr: 4.92e-03, grad_scale: 8.0 2023-03-09 00:10:50,238 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.06 vs. limit=5.0 2023-03-09 00:10:59,976 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84800.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:11:01,283 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84801.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:11:13,082 INFO [train2.py:809] (0/4) Epoch 22, batch 1150, loss[ctc_loss=0.06722, att_loss=0.2376, loss=0.2035, over 16480.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005695, over 46.00 utterances.], tot_loss[ctc_loss=0.07482, att_loss=0.2355, loss=0.2034, over 3252269.25 frames. utt_duration=1194 frames, utt_pad_proportion=0.07154, over 10911.57 utterances.], batch size: 46, lr: 4.92e-03, grad_scale: 16.0 2023-03-09 00:12:04,637 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.375e+02 1.971e+02 2.350e+02 2.955e+02 6.874e+02, threshold=4.701e+02, percent-clipped=3.0 2023-03-09 00:12:25,845 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84855.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:12:31,662 INFO [train2.py:809] (0/4) Epoch 22, batch 1200, loss[ctc_loss=0.06042, att_loss=0.2437, loss=0.2071, over 17304.00 frames. utt_duration=877.8 frames, utt_pad_proportion=0.08179, over 79.00 utterances.], tot_loss[ctc_loss=0.07462, att_loss=0.2357, loss=0.2035, over 3267696.31 frames. utt_duration=1195 frames, utt_pad_proportion=0.06689, over 10948.85 utterances.], batch size: 79, lr: 4.92e-03, grad_scale: 16.0 2023-03-09 00:12:35,051 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=84861.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:13:06,040 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84881.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:13:24,291 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84892.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:13:41,804 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84903.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:13:51,118 INFO [train2.py:809] (0/4) Epoch 22, batch 1250, loss[ctc_loss=0.09021, att_loss=0.2513, loss=0.2191, over 17383.00 frames. utt_duration=1009 frames, utt_pad_proportion=0.04508, over 69.00 utterances.], tot_loss[ctc_loss=0.07369, att_loss=0.2353, loss=0.203, over 3273636.46 frames. utt_duration=1215 frames, utt_pad_proportion=0.06104, over 10786.82 utterances.], batch size: 69, lr: 4.92e-03, grad_scale: 16.0 2023-03-09 00:13:55,034 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-03-09 00:14:10,782 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6064, 3.0175, 3.6774, 2.9563, 3.5119, 4.6843, 4.5058, 3.2588], device='cuda:0'), covar=tensor([0.0419, 0.1706, 0.1313, 0.1456, 0.1198, 0.0861, 0.0577, 0.1342], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0240, 0.0276, 0.0216, 0.0262, 0.0363, 0.0256, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:14:40,799 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=84940.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:14:43,743 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.170e+02 1.899e+02 2.318e+02 2.747e+02 6.253e+02, threshold=4.636e+02, percent-clipped=3.0 2023-03-09 00:14:46,349 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.13 vs. limit=5.0 2023-03-09 00:14:58,318 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=84951.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:15:09,885 INFO [train2.py:809] (0/4) Epoch 22, batch 1300, loss[ctc_loss=0.07411, att_loss=0.2473, loss=0.2127, over 17057.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009296, over 53.00 utterances.], tot_loss[ctc_loss=0.07273, att_loss=0.2345, loss=0.2021, over 3276863.95 frames. utt_duration=1240 frames, utt_pad_proportion=0.05524, over 10587.53 utterances.], batch size: 53, lr: 4.91e-03, grad_scale: 16.0 2023-03-09 00:15:56,204 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9139, 3.7851, 3.6933, 3.1915, 3.7681, 3.8206, 3.8401, 2.9217], device='cuda:0'), covar=tensor([0.0968, 0.0852, 0.1403, 0.2832, 0.0666, 0.2957, 0.0726, 0.3051], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0189, 0.0200, 0.0253, 0.0158, 0.0258, 0.0180, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:15:57,599 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=84989.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:16:02,879 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=84992.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:16:29,438 INFO [train2.py:809] (0/4) Epoch 22, batch 1350, loss[ctc_loss=0.09066, att_loss=0.2431, loss=0.2126, over 16976.00 frames. utt_duration=1360 frames, utt_pad_proportion=0.006925, over 50.00 utterances.], tot_loss[ctc_loss=0.07313, att_loss=0.2354, loss=0.203, over 3273148.13 frames. utt_duration=1210 frames, utt_pad_proportion=0.06327, over 10833.66 utterances.], batch size: 50, lr: 4.91e-03, grad_scale: 16.0 2023-03-09 00:16:34,322 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85012.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:17:18,549 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85040.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:17:18,756 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4773, 4.4932, 4.2571, 2.7538, 4.3242, 4.2148, 3.9707, 2.7939], device='cuda:0'), covar=tensor([0.0103, 0.0103, 0.0288, 0.1049, 0.0106, 0.0283, 0.0296, 0.1283], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0101, 0.0104, 0.0110, 0.0085, 0.0113, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:17:21,480 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 1.871e+02 2.280e+02 2.864e+02 6.552e+02, threshold=4.561e+02, percent-clipped=2.0 2023-03-09 00:17:21,700 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8013, 6.0990, 5.6045, 5.8451, 5.8029, 5.2690, 5.5450, 5.3371], device='cuda:0'), covar=tensor([0.1364, 0.0980, 0.0790, 0.0826, 0.0928, 0.1561, 0.2366, 0.2211], device='cuda:0'), in_proj_covar=tensor([0.0528, 0.0618, 0.0462, 0.0461, 0.0430, 0.0467, 0.0615, 0.0529], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:17:29,349 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-09 00:17:36,290 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0744, 5.0747, 4.9717, 2.1211, 1.9734, 2.9441, 2.3114, 3.8218], device='cuda:0'), covar=tensor([0.0717, 0.0277, 0.0210, 0.5381, 0.5966, 0.2489, 0.3954, 0.1674], device='cuda:0'), in_proj_covar=tensor([0.0354, 0.0276, 0.0266, 0.0244, 0.0339, 0.0330, 0.0253, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 00:17:42,412 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2323, 5.2308, 4.9822, 2.4519, 2.0756, 3.2796, 2.5741, 3.9974], device='cuda:0'), covar=tensor([0.0653, 0.0329, 0.0271, 0.4698, 0.5691, 0.2041, 0.3519, 0.1615], device='cuda:0'), in_proj_covar=tensor([0.0353, 0.0276, 0.0266, 0.0244, 0.0339, 0.0330, 0.0253, 0.0360], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 00:17:48,042 INFO [train2.py:809] (0/4) Epoch 22, batch 1400, loss[ctc_loss=0.08748, att_loss=0.252, loss=0.2191, over 17352.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03461, over 63.00 utterances.], tot_loss[ctc_loss=0.07356, att_loss=0.2355, loss=0.2031, over 3277185.95 frames. utt_duration=1221 frames, utt_pad_proportion=0.0577, over 10747.24 utterances.], batch size: 63, lr: 4.91e-03, grad_scale: 16.0 2023-03-09 00:18:51,216 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7242, 5.1611, 5.0799, 5.0993, 5.2653, 4.9127, 3.8342, 5.1310], device='cuda:0'), covar=tensor([0.0129, 0.0139, 0.0147, 0.0093, 0.0125, 0.0116, 0.0682, 0.0300], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0089, 0.0112, 0.0070, 0.0076, 0.0087, 0.0105, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:19:03,472 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85107.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:19:06,355 INFO [train2.py:809] (0/4) Epoch 22, batch 1450, loss[ctc_loss=0.0476, att_loss=0.2117, loss=0.1789, over 16137.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005648, over 42.00 utterances.], tot_loss[ctc_loss=0.07306, att_loss=0.235, loss=0.2026, over 3272114.80 frames. utt_duration=1243 frames, utt_pad_proportion=0.05251, over 10538.56 utterances.], batch size: 42, lr: 4.91e-03, grad_scale: 16.0 2023-03-09 00:19:16,257 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85115.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:19:59,353 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 1.922e+02 2.285e+02 2.782e+02 4.553e+02, threshold=4.570e+02, percent-clipped=0.0 2023-03-09 00:20:21,384 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85156.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:20:22,042 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-09 00:20:25,861 INFO [train2.py:809] (0/4) Epoch 22, batch 1500, loss[ctc_loss=0.0784, att_loss=0.2567, loss=0.221, over 16865.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007523, over 49.00 utterances.], tot_loss[ctc_loss=0.07307, att_loss=0.2345, loss=0.2022, over 3265020.24 frames. utt_duration=1234 frames, utt_pad_proportion=0.05781, over 10600.01 utterances.], batch size: 49, lr: 4.91e-03, grad_scale: 16.0 2023-03-09 00:20:35,817 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0378, 5.0435, 4.7884, 3.1090, 4.8079, 4.6794, 4.4293, 2.6162], device='cuda:0'), covar=tensor([0.0116, 0.0086, 0.0301, 0.0879, 0.0107, 0.0191, 0.0275, 0.1406], device='cuda:0'), in_proj_covar=tensor([0.0074, 0.0101, 0.0104, 0.0110, 0.0085, 0.0113, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:20:40,442 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85168.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 00:20:52,734 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85176.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:21:01,411 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85181.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:21:26,872 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.53 vs. limit=2.0 2023-03-09 00:21:32,982 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0875, 5.1623, 4.8891, 2.2746, 1.9957, 2.9532, 2.3722, 3.9069], device='cuda:0'), covar=tensor([0.0768, 0.0306, 0.0266, 0.5036, 0.5872, 0.2628, 0.3913, 0.1676], device='cuda:0'), in_proj_covar=tensor([0.0357, 0.0278, 0.0268, 0.0247, 0.0342, 0.0334, 0.0256, 0.0365], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 00:21:46,574 INFO [train2.py:809] (0/4) Epoch 22, batch 1550, loss[ctc_loss=0.05508, att_loss=0.2273, loss=0.1928, over 16270.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007163, over 43.00 utterances.], tot_loss[ctc_loss=0.07283, att_loss=0.2341, loss=0.2019, over 3254036.48 frames. utt_duration=1238 frames, utt_pad_proportion=0.06059, over 10525.36 utterances.], batch size: 43, lr: 4.91e-03, grad_scale: 16.0 2023-03-09 00:22:05,310 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-09 00:22:19,126 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85229.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:22:42,139 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.384e+02 1.952e+02 2.297e+02 2.764e+02 6.264e+02, threshold=4.595e+02, percent-clipped=3.0 2023-03-09 00:23:06,930 INFO [train2.py:809] (0/4) Epoch 22, batch 1600, loss[ctc_loss=0.06671, att_loss=0.2096, loss=0.181, over 15506.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008466, over 36.00 utterances.], tot_loss[ctc_loss=0.07267, att_loss=0.2347, loss=0.2023, over 3256926.11 frames. utt_duration=1234 frames, utt_pad_proportion=0.06225, over 10570.86 utterances.], batch size: 36, lr: 4.91e-03, grad_scale: 8.0 2023-03-09 00:23:17,285 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5983, 5.0186, 4.8788, 4.8996, 4.9996, 4.7136, 3.4928, 5.0004], device='cuda:0'), covar=tensor([0.0126, 0.0137, 0.0139, 0.0097, 0.0122, 0.0126, 0.0759, 0.0211], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0089, 0.0112, 0.0070, 0.0077, 0.0087, 0.0105, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:23:56,469 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85289.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:24:23,656 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85307.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:24:27,035 INFO [train2.py:809] (0/4) Epoch 22, batch 1650, loss[ctc_loss=0.07059, att_loss=0.242, loss=0.2077, over 17343.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03532, over 63.00 utterances.], tot_loss[ctc_loss=0.07229, att_loss=0.2345, loss=0.2021, over 3263675.02 frames. utt_duration=1241 frames, utt_pad_proportion=0.05918, over 10528.79 utterances.], batch size: 63, lr: 4.90e-03, grad_scale: 8.0 2023-03-09 00:25:01,088 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.64 vs. limit=2.0 2023-03-09 00:25:11,284 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85337.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:25:20,304 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 1.903e+02 2.201e+02 2.688e+02 4.353e+02, threshold=4.403e+02, percent-clipped=0.0 2023-03-09 00:25:44,525 INFO [train2.py:809] (0/4) Epoch 22, batch 1700, loss[ctc_loss=0.06067, att_loss=0.229, loss=0.1953, over 16556.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005267, over 45.00 utterances.], tot_loss[ctc_loss=0.07216, att_loss=0.2338, loss=0.2015, over 3264257.20 frames. utt_duration=1251 frames, utt_pad_proportion=0.05641, over 10449.14 utterances.], batch size: 45, lr: 4.90e-03, grad_scale: 8.0 2023-03-09 00:26:46,377 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-03-09 00:27:03,854 INFO [train2.py:809] (0/4) Epoch 22, batch 1750, loss[ctc_loss=0.07403, att_loss=0.2373, loss=0.2046, over 16397.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007903, over 44.00 utterances.], tot_loss[ctc_loss=0.07229, att_loss=0.2334, loss=0.2012, over 3261147.43 frames. utt_duration=1268 frames, utt_pad_proportion=0.05274, over 10303.21 utterances.], batch size: 44, lr: 4.90e-03, grad_scale: 8.0 2023-03-09 00:27:35,458 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85428.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:27:57,806 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6027, 3.3977, 3.3573, 2.9806, 3.3322, 3.4566, 3.4550, 2.5299], device='cuda:0'), covar=tensor([0.1187, 0.1291, 0.2386, 0.3349, 0.1394, 0.1565, 0.1038, 0.3710], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0187, 0.0198, 0.0250, 0.0157, 0.0257, 0.0181, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0001, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:27:58,932 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 1.824e+02 2.227e+02 2.751e+02 5.313e+02, threshold=4.454e+02, percent-clipped=1.0 2023-03-09 00:28:19,203 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85456.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:28:24,084 INFO [train2.py:809] (0/4) Epoch 22, batch 1800, loss[ctc_loss=0.06363, att_loss=0.2305, loss=0.1971, over 17359.00 frames. utt_duration=880.4 frames, utt_pad_proportion=0.07807, over 79.00 utterances.], tot_loss[ctc_loss=0.07163, att_loss=0.2333, loss=0.201, over 3270132.37 frames. utt_duration=1281 frames, utt_pad_proportion=0.04676, over 10225.16 utterances.], batch size: 79, lr: 4.90e-03, grad_scale: 8.0 2023-03-09 00:28:30,466 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85463.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 00:28:41,968 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.27 vs. limit=5.0 2023-03-09 00:28:44,313 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85471.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:29:01,204 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3270, 3.0118, 3.4649, 4.3740, 3.9149, 3.8448, 2.9246, 2.2981], device='cuda:0'), covar=tensor([0.0704, 0.1727, 0.0834, 0.0603, 0.0762, 0.0521, 0.1474, 0.2155], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0215, 0.0190, 0.0219, 0.0226, 0.0180, 0.0200, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:29:04,437 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85483.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:29:13,752 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85489.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:29:37,446 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85504.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:29:45,466 INFO [train2.py:809] (0/4) Epoch 22, batch 1850, loss[ctc_loss=0.06465, att_loss=0.2117, loss=0.1823, over 15501.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008071, over 36.00 utterances.], tot_loss[ctc_loss=0.07183, att_loss=0.2339, loss=0.2015, over 3274283.64 frames. utt_duration=1276 frames, utt_pad_proportion=0.0479, over 10276.46 utterances.], batch size: 36, lr: 4.90e-03, grad_scale: 8.0 2023-03-09 00:30:19,793 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85530.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:30:39,482 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.220e+02 1.961e+02 2.288e+02 2.776e+02 5.536e+02, threshold=4.576e+02, percent-clipped=1.0 2023-03-09 00:30:41,498 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85544.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:31:04,495 INFO [train2.py:809] (0/4) Epoch 22, batch 1900, loss[ctc_loss=0.07223, att_loss=0.2175, loss=0.1885, over 15948.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.00726, over 41.00 utterances.], tot_loss[ctc_loss=0.07186, att_loss=0.234, loss=0.2016, over 3274178.25 frames. utt_duration=1275 frames, utt_pad_proportion=0.04924, over 10287.51 utterances.], batch size: 41, lr: 4.90e-03, grad_scale: 8.0 2023-03-09 00:31:52,542 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7321, 6.0167, 5.5265, 5.7739, 5.7066, 5.1900, 5.3521, 5.2022], device='cuda:0'), covar=tensor([0.1366, 0.0952, 0.0899, 0.0862, 0.0905, 0.1620, 0.2565, 0.2140], device='cuda:0'), in_proj_covar=tensor([0.0527, 0.0620, 0.0460, 0.0459, 0.0434, 0.0467, 0.0615, 0.0526], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:31:55,920 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85591.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:32:21,694 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85607.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:32:24,411 INFO [train2.py:809] (0/4) Epoch 22, batch 1950, loss[ctc_loss=0.06433, att_loss=0.2254, loss=0.1932, over 15954.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.007057, over 41.00 utterances.], tot_loss[ctc_loss=0.07255, att_loss=0.2348, loss=0.2023, over 3276510.97 frames. utt_duration=1260 frames, utt_pad_proportion=0.05166, over 10412.66 utterances.], batch size: 41, lr: 4.90e-03, grad_scale: 8.0 2023-03-09 00:32:41,320 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8385, 5.2007, 5.3874, 5.2417, 5.3731, 5.8321, 5.1598, 5.8984], device='cuda:0'), covar=tensor([0.0736, 0.0721, 0.0814, 0.1304, 0.1795, 0.0782, 0.0683, 0.0651], device='cuda:0'), in_proj_covar=tensor([0.0866, 0.0515, 0.0596, 0.0660, 0.0873, 0.0625, 0.0483, 0.0608], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:32:51,806 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85625.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:33:19,614 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 1.791e+02 2.130e+02 2.666e+02 5.391e+02, threshold=4.261e+02, percent-clipped=1.0 2023-03-09 00:33:24,817 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2538, 3.0292, 3.3874, 4.4089, 3.8424, 3.8644, 2.9047, 2.3770], device='cuda:0'), covar=tensor([0.0819, 0.1837, 0.0888, 0.0518, 0.0863, 0.0538, 0.1540, 0.2184], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0216, 0.0192, 0.0221, 0.0228, 0.0182, 0.0202, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:33:39,318 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85655.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:33:45,243 INFO [train2.py:809] (0/4) Epoch 22, batch 2000, loss[ctc_loss=0.06012, att_loss=0.2191, loss=0.1873, over 16285.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007067, over 43.00 utterances.], tot_loss[ctc_loss=0.07243, att_loss=0.2354, loss=0.2028, over 3282908.12 frames. utt_duration=1232 frames, utt_pad_proportion=0.05658, over 10670.60 utterances.], batch size: 43, lr: 4.89e-03, grad_scale: 8.0 2023-03-09 00:34:29,085 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85686.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 00:35:05,235 INFO [train2.py:809] (0/4) Epoch 22, batch 2050, loss[ctc_loss=0.07994, att_loss=0.2437, loss=0.211, over 17235.00 frames. utt_duration=697.7 frames, utt_pad_proportion=0.1235, over 99.00 utterances.], tot_loss[ctc_loss=0.07218, att_loss=0.2353, loss=0.2027, over 3283909.60 frames. utt_duration=1223 frames, utt_pad_proportion=0.05786, over 10750.36 utterances.], batch size: 99, lr: 4.89e-03, grad_scale: 8.0 2023-03-09 00:36:01,001 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.043e+02 1.910e+02 2.299e+02 2.633e+02 5.710e+02, threshold=4.599e+02, percent-clipped=8.0 2023-03-09 00:36:27,082 INFO [train2.py:809] (0/4) Epoch 22, batch 2100, loss[ctc_loss=0.05512, att_loss=0.2282, loss=0.1936, over 16132.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005879, over 42.00 utterances.], tot_loss[ctc_loss=0.0731, att_loss=0.2362, loss=0.2036, over 3285284.53 frames. utt_duration=1221 frames, utt_pad_proportion=0.05852, over 10773.31 utterances.], batch size: 42, lr: 4.89e-03, grad_scale: 8.0 2023-03-09 00:36:33,829 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85763.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:36:47,327 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=85771.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:37:08,349 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85784.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:37:47,842 INFO [train2.py:809] (0/4) Epoch 22, batch 2150, loss[ctc_loss=0.0855, att_loss=0.2593, loss=0.2246, over 17293.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01245, over 55.00 utterances.], tot_loss[ctc_loss=0.07378, att_loss=0.2367, loss=0.2041, over 3290240.41 frames. utt_duration=1238 frames, utt_pad_proportion=0.05351, over 10640.53 utterances.], batch size: 55, lr: 4.89e-03, grad_scale: 8.0 2023-03-09 00:37:51,074 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85811.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:37:52,661 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0763, 6.2520, 5.6970, 5.9972, 5.9573, 5.4383, 5.7551, 5.4620], device='cuda:0'), covar=tensor([0.1069, 0.0924, 0.0928, 0.0853, 0.0913, 0.1558, 0.2340, 0.2577], device='cuda:0'), in_proj_covar=tensor([0.0527, 0.0616, 0.0460, 0.0460, 0.0430, 0.0466, 0.0616, 0.0529], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:38:04,882 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=85819.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:38:36,672 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85839.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:38:42,695 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 1.885e+02 2.205e+02 3.046e+02 7.448e+02, threshold=4.410e+02, percent-clipped=6.0 2023-03-09 00:39:07,837 INFO [train2.py:809] (0/4) Epoch 22, batch 2200, loss[ctc_loss=0.0708, att_loss=0.2162, loss=0.1871, over 16011.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.00637, over 40.00 utterances.], tot_loss[ctc_loss=0.07375, att_loss=0.2368, loss=0.2042, over 3283744.09 frames. utt_duration=1223 frames, utt_pad_proportion=0.05859, over 10754.99 utterances.], batch size: 40, lr: 4.89e-03, grad_scale: 8.0 2023-03-09 00:39:35,104 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85875.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:39:52,000 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85886.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:39:58,189 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1199, 5.4373, 5.3569, 5.3301, 5.4184, 5.4041, 5.0894, 4.8899], device='cuda:0'), covar=tensor([0.0827, 0.0422, 0.0269, 0.0417, 0.0245, 0.0272, 0.0351, 0.0278], device='cuda:0'), in_proj_covar=tensor([0.0521, 0.0364, 0.0344, 0.0359, 0.0423, 0.0428, 0.0357, 0.0394], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-09 00:40:28,362 INFO [train2.py:809] (0/4) Epoch 22, batch 2250, loss[ctc_loss=0.08171, att_loss=0.2511, loss=0.2172, over 16971.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007261, over 50.00 utterances.], tot_loss[ctc_loss=0.07468, att_loss=0.2375, loss=0.2049, over 3280630.91 frames. utt_duration=1212 frames, utt_pad_proportion=0.06371, over 10843.08 utterances.], batch size: 50, lr: 4.89e-03, grad_scale: 8.0 2023-03-09 00:40:50,242 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8909, 6.1372, 5.5522, 5.8416, 5.8001, 5.3698, 5.5892, 5.3108], device='cuda:0'), covar=tensor([0.1100, 0.0724, 0.0879, 0.0789, 0.0792, 0.1340, 0.2124, 0.2078], device='cuda:0'), in_proj_covar=tensor([0.0527, 0.0615, 0.0459, 0.0459, 0.0430, 0.0466, 0.0612, 0.0528], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:41:02,624 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85930.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 00:41:04,896 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-09 00:41:11,703 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85936.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:41:22,180 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 1.895e+02 2.304e+02 2.747e+02 7.782e+02, threshold=4.607e+02, percent-clipped=2.0 2023-03-09 00:41:44,920 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=85957.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:41:47,669 INFO [train2.py:809] (0/4) Epoch 22, batch 2300, loss[ctc_loss=0.06731, att_loss=0.2242, loss=0.1928, over 16285.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007097, over 43.00 utterances.], tot_loss[ctc_loss=0.07361, att_loss=0.2361, loss=0.2036, over 3270107.12 frames. utt_duration=1221 frames, utt_pad_proportion=0.06356, over 10726.43 utterances.], batch size: 43, lr: 4.89e-03, grad_scale: 8.0 2023-03-09 00:42:24,302 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=85981.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 00:42:40,389 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=85991.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 00:42:51,340 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7025, 5.0880, 4.8164, 5.0808, 5.1104, 4.7332, 3.7196, 5.0729], device='cuda:0'), covar=tensor([0.0120, 0.0122, 0.0159, 0.0100, 0.0117, 0.0129, 0.0641, 0.0197], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0090, 0.0112, 0.0070, 0.0077, 0.0087, 0.0105, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:42:54,502 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-86000.pt 2023-03-09 00:43:02,476 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86002.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:43:13,253 INFO [train2.py:809] (0/4) Epoch 22, batch 2350, loss[ctc_loss=0.0817, att_loss=0.2497, loss=0.2161, over 17410.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.0458, over 69.00 utterances.], tot_loss[ctc_loss=0.07271, att_loss=0.235, loss=0.2026, over 3263584.72 frames. utt_duration=1208 frames, utt_pad_proportion=0.06799, over 10818.13 utterances.], batch size: 69, lr: 4.88e-03, grad_scale: 8.0 2023-03-09 00:43:20,616 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5849, 5.0018, 4.7703, 5.0130, 5.0397, 4.6389, 3.5915, 4.9458], device='cuda:0'), covar=tensor([0.0120, 0.0111, 0.0136, 0.0083, 0.0102, 0.0125, 0.0668, 0.0216], device='cuda:0'), in_proj_covar=tensor([0.0092, 0.0090, 0.0112, 0.0070, 0.0077, 0.0087, 0.0105, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:43:28,490 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86018.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:44:06,999 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.206e+02 1.776e+02 2.270e+02 2.670e+02 5.895e+02, threshold=4.541e+02, percent-clipped=3.0 2023-03-09 00:44:21,738 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86052.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:44:32,410 INFO [train2.py:809] (0/4) Epoch 22, batch 2400, loss[ctc_loss=0.07676, att_loss=0.2386, loss=0.2062, over 16119.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006664, over 42.00 utterances.], tot_loss[ctc_loss=0.07285, att_loss=0.2353, loss=0.2028, over 3271182.80 frames. utt_duration=1227 frames, utt_pad_proportion=0.06027, over 10675.86 utterances.], batch size: 42, lr: 4.88e-03, grad_scale: 8.0 2023-03-09 00:44:39,812 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86063.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:44:49,041 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86069.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:44:51,319 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6612, 2.6216, 5.1194, 4.0609, 3.2213, 4.3312, 4.8420, 4.7321], device='cuda:0'), covar=tensor([0.0224, 0.1637, 0.0154, 0.0810, 0.1561, 0.0237, 0.0126, 0.0270], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0243, 0.0193, 0.0316, 0.0266, 0.0218, 0.0180, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:45:12,634 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86084.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:45:18,654 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2706, 4.3910, 4.7044, 4.4624, 5.2427, 4.6236, 4.5301, 2.5030], device='cuda:0'), covar=tensor([0.0336, 0.0436, 0.0301, 0.0415, 0.0707, 0.0234, 0.0355, 0.1836], device='cuda:0'), in_proj_covar=tensor([0.0168, 0.0193, 0.0192, 0.0209, 0.0370, 0.0162, 0.0183, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:45:50,868 INFO [train2.py:809] (0/4) Epoch 22, batch 2450, loss[ctc_loss=0.06321, att_loss=0.2324, loss=0.1985, over 16952.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008415, over 50.00 utterances.], tot_loss[ctc_loss=0.07329, att_loss=0.2352, loss=0.2028, over 3266692.98 frames. utt_duration=1218 frames, utt_pad_proportion=0.0645, over 10745.10 utterances.], batch size: 50, lr: 4.88e-03, grad_scale: 8.0 2023-03-09 00:45:58,779 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86113.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:46:07,819 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-03-09 00:46:26,013 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86130.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:46:28,883 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86132.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:46:39,883 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86139.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:46:45,846 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.355e+02 2.126e+02 2.601e+02 3.056e+02 8.215e+02, threshold=5.203e+02, percent-clipped=8.0 2023-03-09 00:47:11,848 INFO [train2.py:809] (0/4) Epoch 22, batch 2500, loss[ctc_loss=0.07101, att_loss=0.2382, loss=0.2048, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006751, over 43.00 utterances.], tot_loss[ctc_loss=0.07299, att_loss=0.2356, loss=0.2031, over 3280072.99 frames. utt_duration=1246 frames, utt_pad_proportion=0.05427, over 10545.23 utterances.], batch size: 43, lr: 4.88e-03, grad_scale: 8.0 2023-03-09 00:47:44,693 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9151, 5.1882, 4.7926, 5.3141, 4.6322, 4.9581, 5.3406, 5.1134], device='cuda:0'), covar=tensor([0.0568, 0.0301, 0.0740, 0.0300, 0.0398, 0.0261, 0.0213, 0.0187], device='cuda:0'), in_proj_covar=tensor([0.0388, 0.0318, 0.0367, 0.0350, 0.0322, 0.0236, 0.0303, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 00:47:50,496 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.99 vs. limit=5.0 2023-03-09 00:47:57,559 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86186.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:47:58,891 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86187.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:48:31,050 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.64 vs. limit=5.0 2023-03-09 00:48:33,441 INFO [train2.py:809] (0/4) Epoch 22, batch 2550, loss[ctc_loss=0.0811, att_loss=0.244, loss=0.2114, over 16627.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005185, over 47.00 utterances.], tot_loss[ctc_loss=0.07318, att_loss=0.2354, loss=0.203, over 3283062.19 frames. utt_duration=1252 frames, utt_pad_proportion=0.05049, over 10504.21 utterances.], batch size: 47, lr: 4.88e-03, grad_scale: 8.0 2023-03-09 00:49:10,400 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86231.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:49:15,067 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86234.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:49:29,161 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.366e+02 1.974e+02 2.356e+02 2.687e+02 6.031e+02, threshold=4.712e+02, percent-clipped=2.0 2023-03-09 00:49:34,284 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86246.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:49:54,969 INFO [train2.py:809] (0/4) Epoch 22, batch 2600, loss[ctc_loss=0.07973, att_loss=0.2475, loss=0.214, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007409, over 44.00 utterances.], tot_loss[ctc_loss=0.07267, att_loss=0.2354, loss=0.2028, over 3278970.60 frames. utt_duration=1238 frames, utt_pad_proportion=0.05503, over 10605.02 utterances.], batch size: 44, lr: 4.88e-03, grad_scale: 8.0 2023-03-09 00:50:07,331 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5105, 2.9269, 3.3435, 4.6359, 4.0743, 3.9150, 3.1135, 2.4267], device='cuda:0'), covar=tensor([0.0722, 0.1993, 0.1018, 0.0433, 0.0771, 0.0562, 0.1450, 0.2130], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0219, 0.0193, 0.0223, 0.0229, 0.0184, 0.0204, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:50:31,757 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86281.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:50:39,202 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86286.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 00:51:13,018 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86307.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:51:15,789 INFO [train2.py:809] (0/4) Epoch 22, batch 2650, loss[ctc_loss=0.07976, att_loss=0.2169, loss=0.1895, over 15387.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.01014, over 35.00 utterances.], tot_loss[ctc_loss=0.07173, att_loss=0.2341, loss=0.2016, over 3270111.73 frames. utt_duration=1263 frames, utt_pad_proportion=0.05172, over 10370.48 utterances.], batch size: 35, lr: 4.88e-03, grad_scale: 8.0 2023-03-09 00:51:23,125 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86313.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:51:35,546 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3860, 5.6813, 5.1437, 5.4814, 5.3398, 4.9215, 5.1240, 4.9650], device='cuda:0'), covar=tensor([0.1242, 0.0908, 0.0955, 0.0756, 0.0869, 0.1509, 0.2335, 0.2330], device='cuda:0'), in_proj_covar=tensor([0.0528, 0.0616, 0.0458, 0.0456, 0.0429, 0.0465, 0.0613, 0.0528], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 00:51:49,592 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86329.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:52:11,918 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 1.827e+02 2.241e+02 2.634e+02 5.171e+02, threshold=4.483e+02, percent-clipped=1.0 2023-03-09 00:52:36,602 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86358.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:52:38,735 INFO [train2.py:809] (0/4) Epoch 22, batch 2700, loss[ctc_loss=0.0689, att_loss=0.2335, loss=0.2006, over 16879.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007023, over 49.00 utterances.], tot_loss[ctc_loss=0.07132, att_loss=0.2337, loss=0.2013, over 3276411.14 frames. utt_duration=1291 frames, utt_pad_proportion=0.04396, over 10164.27 utterances.], batch size: 49, lr: 4.87e-03, grad_scale: 8.0 2023-03-09 00:53:09,924 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2361, 2.5245, 2.6700, 4.3916, 4.0320, 3.8268, 2.9351, 2.3385], device='cuda:0'), covar=tensor([0.0809, 0.2273, 0.1398, 0.0516, 0.0666, 0.0537, 0.1381, 0.2062], device='cuda:0'), in_proj_covar=tensor([0.0180, 0.0215, 0.0190, 0.0219, 0.0224, 0.0181, 0.0201, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:53:26,025 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.89 vs. limit=2.0 2023-03-09 00:53:48,171 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6853, 2.7394, 3.7872, 3.0258, 3.7236, 4.8600, 4.7076, 3.0847], device='cuda:0'), covar=tensor([0.0383, 0.1923, 0.0966, 0.1526, 0.0902, 0.0544, 0.0421, 0.1500], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0240, 0.0279, 0.0217, 0.0264, 0.0364, 0.0259, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 00:53:58,166 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86408.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:54:00,130 INFO [train2.py:809] (0/4) Epoch 22, batch 2750, loss[ctc_loss=0.05689, att_loss=0.2121, loss=0.181, over 15874.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.008703, over 39.00 utterances.], tot_loss[ctc_loss=0.07159, att_loss=0.2339, loss=0.2015, over 3275763.93 frames. utt_duration=1267 frames, utt_pad_proportion=0.04992, over 10357.43 utterances.], batch size: 39, lr: 4.87e-03, grad_scale: 8.0 2023-03-09 00:54:25,748 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86425.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:54:44,592 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0188, 3.8455, 3.1913, 3.4968, 4.0160, 3.6952, 3.1934, 4.3163], device='cuda:0'), covar=tensor([0.1051, 0.0477, 0.1082, 0.0711, 0.0655, 0.0709, 0.0819, 0.0414], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0218, 0.0228, 0.0202, 0.0282, 0.0243, 0.0201, 0.0289], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 00:54:49,456 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86440.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:54:53,694 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.448e+02 1.998e+02 2.389e+02 2.932e+02 8.437e+02, threshold=4.778e+02, percent-clipped=2.0 2023-03-09 00:54:57,204 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86445.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:55:06,220 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.91 vs. limit=2.0 2023-03-09 00:55:20,790 INFO [train2.py:809] (0/4) Epoch 22, batch 2800, loss[ctc_loss=0.09275, att_loss=0.2522, loss=0.2203, over 16931.00 frames. utt_duration=685.6 frames, utt_pad_proportion=0.1398, over 99.00 utterances.], tot_loss[ctc_loss=0.07162, att_loss=0.2339, loss=0.2014, over 3270015.78 frames. utt_duration=1277 frames, utt_pad_proportion=0.04865, over 10255.44 utterances.], batch size: 99, lr: 4.87e-03, grad_scale: 8.0 2023-03-09 00:55:46,307 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-03-09 00:56:01,625 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.92 vs. limit=2.0 2023-03-09 00:56:02,384 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5747, 3.0358, 2.6723, 2.9043, 3.1801, 3.0505, 2.5250, 3.0635], device='cuda:0'), covar=tensor([0.0881, 0.0447, 0.0797, 0.0595, 0.0702, 0.0626, 0.0797, 0.0444], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0217, 0.0227, 0.0201, 0.0281, 0.0243, 0.0201, 0.0289], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 00:56:28,116 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86501.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:56:36,525 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86506.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 00:56:40,824 INFO [train2.py:809] (0/4) Epoch 22, batch 2850, loss[ctc_loss=0.05492, att_loss=0.2216, loss=0.1883, over 15945.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007741, over 41.00 utterances.], tot_loss[ctc_loss=0.07208, att_loss=0.2347, loss=0.2022, over 3266399.84 frames. utt_duration=1239 frames, utt_pad_proportion=0.05946, over 10554.04 utterances.], batch size: 41, lr: 4.87e-03, grad_scale: 8.0 2023-03-09 00:57:16,054 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86531.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:57:34,135 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.710e+01 1.849e+02 2.202e+02 2.707e+02 3.690e+02, threshold=4.403e+02, percent-clipped=0.0 2023-03-09 00:58:01,269 INFO [train2.py:809] (0/4) Epoch 22, batch 2900, loss[ctc_loss=0.09548, att_loss=0.2544, loss=0.2226, over 17277.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01259, over 55.00 utterances.], tot_loss[ctc_loss=0.07268, att_loss=0.2354, loss=0.2028, over 3263066.71 frames. utt_duration=1213 frames, utt_pad_proportion=0.06645, over 10774.61 utterances.], batch size: 55, lr: 4.87e-03, grad_scale: 8.0 2023-03-09 00:58:33,022 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86579.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:58:34,871 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86580.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:58:44,199 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86586.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 00:59:09,657 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86602.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 00:59:20,605 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6422, 2.4711, 4.9477, 3.9555, 3.0250, 4.2237, 4.6030, 4.7544], device='cuda:0'), covar=tensor([0.0150, 0.1407, 0.0137, 0.0705, 0.1545, 0.0214, 0.0146, 0.0180], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0243, 0.0194, 0.0318, 0.0266, 0.0219, 0.0182, 0.0211], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 00:59:21,687 INFO [train2.py:809] (0/4) Epoch 22, batch 2950, loss[ctc_loss=0.07138, att_loss=0.2429, loss=0.2086, over 17114.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01521, over 56.00 utterances.], tot_loss[ctc_loss=0.07248, att_loss=0.2352, loss=0.2026, over 3266191.98 frames. utt_duration=1223 frames, utt_pad_proportion=0.06227, over 10694.56 utterances.], batch size: 56, lr: 4.87e-03, grad_scale: 8.0 2023-03-09 00:59:28,863 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86613.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:00:01,851 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86634.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 01:00:12,871 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86641.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:00:15,584 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.364e+02 1.951e+02 2.348e+02 3.080e+02 7.280e+02, threshold=4.697e+02, percent-clipped=5.0 2023-03-09 01:00:41,177 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86658.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:00:42,427 INFO [train2.py:809] (0/4) Epoch 22, batch 3000, loss[ctc_loss=0.0821, att_loss=0.2463, loss=0.2135, over 17410.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.04645, over 69.00 utterances.], tot_loss[ctc_loss=0.07211, att_loss=0.2345, loss=0.202, over 3267709.90 frames. utt_duration=1237 frames, utt_pad_proportion=0.05954, over 10582.89 utterances.], batch size: 69, lr: 4.87e-03, grad_scale: 8.0 2023-03-09 01:00:42,430 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 01:00:57,105 INFO [train2.py:843] (0/4) Epoch 22, validation: ctc_loss=0.03993, att_loss=0.2347, loss=0.1957, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 01:00:57,106 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 01:01:01,000 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86661.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:01:23,312 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.3213, 5.6118, 5.9533, 5.6533, 5.8392, 6.2793, 5.5130, 6.3226], device='cuda:0'), covar=tensor([0.0636, 0.0651, 0.0690, 0.1280, 0.1659, 0.0808, 0.0605, 0.0576], device='cuda:0'), in_proj_covar=tensor([0.0863, 0.0508, 0.0596, 0.0657, 0.0864, 0.0628, 0.0486, 0.0607], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 01:01:39,515 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.41 vs. limit=2.0 2023-03-09 01:01:45,295 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-09 01:02:11,029 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86706.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:02:14,480 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86708.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:02:16,474 INFO [train2.py:809] (0/4) Epoch 22, batch 3050, loss[ctc_loss=0.06452, att_loss=0.2367, loss=0.2022, over 17416.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03223, over 63.00 utterances.], tot_loss[ctc_loss=0.07135, att_loss=0.2336, loss=0.2011, over 3260497.07 frames. utt_duration=1253 frames, utt_pad_proportion=0.05666, over 10421.30 utterances.], batch size: 63, lr: 4.86e-03, grad_scale: 8.0 2023-03-09 01:02:43,439 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86725.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:02:51,091 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9976, 5.2948, 4.8474, 5.3468, 4.7241, 4.9675, 5.4166, 5.1457], device='cuda:0'), covar=tensor([0.0597, 0.0300, 0.0843, 0.0348, 0.0434, 0.0259, 0.0213, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0383, 0.0318, 0.0364, 0.0348, 0.0318, 0.0235, 0.0299, 0.0283], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 01:03:11,084 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 1.829e+02 2.203e+02 2.781e+02 6.356e+02, threshold=4.407e+02, percent-clipped=2.0 2023-03-09 01:03:31,224 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86756.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:03:36,291 INFO [train2.py:809] (0/4) Epoch 22, batch 3100, loss[ctc_loss=0.06551, att_loss=0.2255, loss=0.1935, over 16020.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006981, over 40.00 utterances.], tot_loss[ctc_loss=0.07194, att_loss=0.2344, loss=0.2019, over 3264921.91 frames. utt_duration=1249 frames, utt_pad_proportion=0.05633, over 10470.22 utterances.], batch size: 40, lr: 4.86e-03, grad_scale: 8.0 2023-03-09 01:03:59,909 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86773.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:04:35,399 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86796.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:04:35,661 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5441, 2.5111, 5.0208, 3.9999, 3.1335, 4.2850, 4.8435, 4.6936], device='cuda:0'), covar=tensor([0.0268, 0.1509, 0.0205, 0.0751, 0.1523, 0.0231, 0.0151, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0240, 0.0192, 0.0314, 0.0262, 0.0216, 0.0181, 0.0210], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:04:43,629 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86801.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 01:04:57,200 INFO [train2.py:809] (0/4) Epoch 22, batch 3150, loss[ctc_loss=0.06601, att_loss=0.222, loss=0.1908, over 16415.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.005508, over 44.00 utterances.], tot_loss[ctc_loss=0.07273, att_loss=0.2349, loss=0.2025, over 3270119.56 frames. utt_duration=1249 frames, utt_pad_proportion=0.05582, over 10484.86 utterances.], batch size: 44, lr: 4.86e-03, grad_scale: 8.0 2023-03-09 01:05:05,986 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86814.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:05:12,904 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0814, 5.0883, 4.7813, 2.7749, 4.7788, 4.6749, 4.3061, 2.7717], device='cuda:0'), covar=tensor([0.0117, 0.0108, 0.0294, 0.1120, 0.0116, 0.0216, 0.0323, 0.1326], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0102, 0.0106, 0.0111, 0.0086, 0.0115, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 01:05:51,843 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.403e+02 1.971e+02 2.330e+02 2.834e+02 6.258e+02, threshold=4.661e+02, percent-clipped=4.0 2023-03-09 01:06:18,406 INFO [train2.py:809] (0/4) Epoch 22, batch 3200, loss[ctc_loss=0.06192, att_loss=0.2166, loss=0.1857, over 15878.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009487, over 39.00 utterances.], tot_loss[ctc_loss=0.07246, att_loss=0.2349, loss=0.2024, over 3270327.86 frames. utt_duration=1237 frames, utt_pad_proportion=0.06042, over 10589.27 utterances.], batch size: 39, lr: 4.86e-03, grad_scale: 8.0 2023-03-09 01:06:45,803 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86875.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:06:50,321 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=86878.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:07:27,669 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=86902.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:07:39,432 INFO [train2.py:809] (0/4) Epoch 22, batch 3250, loss[ctc_loss=0.06548, att_loss=0.2435, loss=0.2079, over 16618.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005039, over 47.00 utterances.], tot_loss[ctc_loss=0.07239, att_loss=0.2347, loss=0.2023, over 3262072.32 frames. utt_duration=1217 frames, utt_pad_proportion=0.066, over 10738.85 utterances.], batch size: 47, lr: 4.86e-03, grad_scale: 8.0 2023-03-09 01:07:39,833 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2153, 3.7518, 3.2411, 3.4944, 4.0005, 3.6968, 3.1368, 4.2932], device='cuda:0'), covar=tensor([0.0885, 0.0503, 0.0986, 0.0624, 0.0649, 0.0645, 0.0776, 0.0377], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0218, 0.0228, 0.0203, 0.0282, 0.0244, 0.0201, 0.0291], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 01:08:22,562 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=86936.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:08:27,442 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=86939.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:08:33,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.070e+02 1.904e+02 2.296e+02 2.851e+02 6.880e+02, threshold=4.593e+02, percent-clipped=1.0 2023-03-09 01:08:44,088 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=86950.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:08:58,538 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0112, 5.3073, 4.8963, 5.3757, 4.7522, 4.9763, 5.4382, 5.2074], device='cuda:0'), covar=tensor([0.0579, 0.0310, 0.0791, 0.0324, 0.0451, 0.0282, 0.0228, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0386, 0.0320, 0.0365, 0.0351, 0.0322, 0.0237, 0.0301, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 01:08:59,935 INFO [train2.py:809] (0/4) Epoch 22, batch 3300, loss[ctc_loss=0.077, att_loss=0.2325, loss=0.2014, over 16413.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006351, over 44.00 utterances.], tot_loss[ctc_loss=0.07205, att_loss=0.2349, loss=0.2024, over 3265411.31 frames. utt_duration=1231 frames, utt_pad_proportion=0.0626, over 10623.30 utterances.], batch size: 44, lr: 4.86e-03, grad_scale: 8.0 2023-03-09 01:09:50,601 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8617, 5.1402, 5.1280, 5.0344, 5.1697, 5.1182, 4.7899, 4.5879], device='cuda:0'), covar=tensor([0.1022, 0.0560, 0.0310, 0.0558, 0.0316, 0.0340, 0.0460, 0.0369], device='cuda:0'), in_proj_covar=tensor([0.0521, 0.0361, 0.0344, 0.0357, 0.0422, 0.0429, 0.0356, 0.0395], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-09 01:10:08,592 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87002.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 01:10:20,440 INFO [train2.py:809] (0/4) Epoch 22, batch 3350, loss[ctc_loss=0.09006, att_loss=0.2527, loss=0.2201, over 17066.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008215, over 52.00 utterances.], tot_loss[ctc_loss=0.07197, att_loss=0.2355, loss=0.2028, over 3277786.11 frames. utt_duration=1232 frames, utt_pad_proportion=0.057, over 10657.82 utterances.], batch size: 52, lr: 4.86e-03, grad_scale: 8.0 2023-03-09 01:11:13,695 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 1.863e+02 2.152e+02 2.724e+02 4.782e+02, threshold=4.303e+02, percent-clipped=1.0 2023-03-09 01:11:40,079 INFO [train2.py:809] (0/4) Epoch 22, batch 3400, loss[ctc_loss=0.06055, att_loss=0.2264, loss=0.1932, over 16285.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006324, over 43.00 utterances.], tot_loss[ctc_loss=0.07203, att_loss=0.2356, loss=0.2029, over 3280698.67 frames. utt_duration=1226 frames, utt_pad_proportion=0.0572, over 10716.74 utterances.], batch size: 43, lr: 4.85e-03, grad_scale: 8.0 2023-03-09 01:11:47,031 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87063.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 01:11:54,596 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8856, 4.8743, 4.5723, 2.9672, 4.5899, 4.5721, 4.1111, 2.6390], device='cuda:0'), covar=tensor([0.0122, 0.0121, 0.0327, 0.0971, 0.0121, 0.0227, 0.0349, 0.1364], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0102, 0.0106, 0.0111, 0.0085, 0.0115, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 01:12:38,085 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87096.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:12:46,596 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87101.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:12:52,780 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1164, 3.7687, 3.1362, 3.3346, 3.9903, 3.6583, 2.8284, 4.2686], device='cuda:0'), covar=tensor([0.0985, 0.0515, 0.1086, 0.0740, 0.0687, 0.0699, 0.0977, 0.0441], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0217, 0.0227, 0.0202, 0.0281, 0.0243, 0.0201, 0.0290], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 01:12:59,255 INFO [train2.py:809] (0/4) Epoch 22, batch 3450, loss[ctc_loss=0.06843, att_loss=0.2179, loss=0.188, over 15791.00 frames. utt_duration=1664 frames, utt_pad_proportion=0.006787, over 38.00 utterances.], tot_loss[ctc_loss=0.0727, att_loss=0.2357, loss=0.2031, over 3280051.99 frames. utt_duration=1240 frames, utt_pad_proportion=0.05386, over 10597.23 utterances.], batch size: 38, lr: 4.85e-03, grad_scale: 8.0 2023-03-09 01:13:52,177 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.163e+02 1.972e+02 2.278e+02 2.813e+02 5.019e+02, threshold=4.557e+02, percent-clipped=5.0 2023-03-09 01:13:53,938 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87144.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:14:02,269 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87149.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:14:19,577 INFO [train2.py:809] (0/4) Epoch 22, batch 3500, loss[ctc_loss=0.06409, att_loss=0.2356, loss=0.2013, over 16626.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005484, over 47.00 utterances.], tot_loss[ctc_loss=0.07258, att_loss=0.2356, loss=0.203, over 3290253.80 frames. utt_duration=1255 frames, utt_pad_proportion=0.04737, over 10502.12 utterances.], batch size: 47, lr: 4.85e-03, grad_scale: 8.0 2023-03-09 01:14:37,035 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87170.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:15:39,617 INFO [train2.py:809] (0/4) Epoch 22, batch 3550, loss[ctc_loss=0.06469, att_loss=0.2318, loss=0.1984, over 16430.00 frames. utt_duration=1495 frames, utt_pad_proportion=0.006044, over 44.00 utterances.], tot_loss[ctc_loss=0.07202, att_loss=0.2351, loss=0.2025, over 3287233.14 frames. utt_duration=1261 frames, utt_pad_proportion=0.04725, over 10441.54 utterances.], batch size: 44, lr: 4.85e-03, grad_scale: 8.0 2023-03-09 01:15:51,489 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-09 01:16:12,837 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87230.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:16:18,879 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87234.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:16:22,077 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87236.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:16:32,594 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 1.973e+02 2.339e+02 2.801e+02 5.420e+02, threshold=4.678e+02, percent-clipped=3.0 2023-03-09 01:16:59,775 INFO [train2.py:809] (0/4) Epoch 22, batch 3600, loss[ctc_loss=0.0629, att_loss=0.2375, loss=0.2026, over 17537.00 frames. utt_duration=1018 frames, utt_pad_proportion=0.03954, over 69.00 utterances.], tot_loss[ctc_loss=0.07187, att_loss=0.2353, loss=0.2026, over 3281554.15 frames. utt_duration=1266 frames, utt_pad_proportion=0.04661, over 10379.61 utterances.], batch size: 69, lr: 4.85e-03, grad_scale: 16.0 2023-03-09 01:17:39,242 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87284.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:17:50,575 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87291.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:18:19,614 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-09 01:18:19,832 INFO [train2.py:809] (0/4) Epoch 22, batch 3650, loss[ctc_loss=0.08132, att_loss=0.2146, loss=0.1879, over 15364.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01114, over 35.00 utterances.], tot_loss[ctc_loss=0.07191, att_loss=0.2351, loss=0.2025, over 3277234.98 frames. utt_duration=1248 frames, utt_pad_proportion=0.05136, over 10516.13 utterances.], batch size: 35, lr: 4.85e-03, grad_scale: 16.0 2023-03-09 01:18:59,358 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.36 vs. limit=5.0 2023-03-09 01:19:13,098 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.232e+02 1.934e+02 2.315e+02 2.927e+02 6.600e+02, threshold=4.631e+02, percent-clipped=3.0 2023-03-09 01:19:38,652 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87358.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:19:40,005 INFO [train2.py:809] (0/4) Epoch 22, batch 3700, loss[ctc_loss=0.08133, att_loss=0.2382, loss=0.2068, over 16409.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007027, over 44.00 utterances.], tot_loss[ctc_loss=0.07205, att_loss=0.2355, loss=0.2028, over 3285766.90 frames. utt_duration=1234 frames, utt_pad_proportion=0.05186, over 10662.30 utterances.], batch size: 44, lr: 4.85e-03, grad_scale: 16.0 2023-03-09 01:19:44,110 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-03-09 01:19:44,742 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87362.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:19:53,066 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-03-09 01:20:59,237 INFO [train2.py:809] (0/4) Epoch 22, batch 3750, loss[ctc_loss=0.07001, att_loss=0.2545, loss=0.2176, over 17368.00 frames. utt_duration=1179 frames, utt_pad_proportion=0.02076, over 59.00 utterances.], tot_loss[ctc_loss=0.07211, att_loss=0.2351, loss=0.2025, over 3286089.52 frames. utt_duration=1242 frames, utt_pad_proportion=0.0504, over 10596.27 utterances.], batch size: 59, lr: 4.84e-03, grad_scale: 16.0 2023-03-09 01:21:21,505 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87423.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:21:36,133 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.93 vs. limit=5.0 2023-03-09 01:21:52,782 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.314e+02 1.951e+02 2.463e+02 3.155e+02 5.552e+02, threshold=4.927e+02, percent-clipped=3.0 2023-03-09 01:22:17,530 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-09 01:22:19,681 INFO [train2.py:809] (0/4) Epoch 22, batch 3800, loss[ctc_loss=0.04933, att_loss=0.2356, loss=0.1983, over 16895.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006171, over 49.00 utterances.], tot_loss[ctc_loss=0.07228, att_loss=0.2354, loss=0.2028, over 3277494.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.05759, over 10704.50 utterances.], batch size: 49, lr: 4.84e-03, grad_scale: 16.0 2023-03-09 01:22:23,099 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87461.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:22:23,964 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.86 vs. limit=5.0 2023-03-09 01:22:32,605 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1798, 4.4608, 4.5601, 4.7860, 2.7704, 4.2858, 2.8405, 2.0852], device='cuda:0'), covar=tensor([0.0456, 0.0262, 0.0599, 0.0217, 0.1573, 0.0260, 0.1438, 0.1511], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0170, 0.0261, 0.0160, 0.0222, 0.0152, 0.0233, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:22:35,690 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3301, 2.9450, 3.6556, 3.1903, 3.4906, 4.6192, 4.3985, 3.2935], device='cuda:0'), covar=tensor([0.0437, 0.1714, 0.1158, 0.1263, 0.1137, 0.0880, 0.0544, 0.1156], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0242, 0.0282, 0.0221, 0.0265, 0.0372, 0.0263, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 01:22:37,138 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87470.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:22:57,607 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4416, 4.9405, 4.8403, 4.8467, 4.9970, 4.6867, 3.2303, 4.8523], device='cuda:0'), covar=tensor([0.0173, 0.0180, 0.0174, 0.0128, 0.0135, 0.0150, 0.1014, 0.0391], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0089, 0.0113, 0.0071, 0.0077, 0.0088, 0.0106, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 01:23:34,629 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87505.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:23:40,698 INFO [train2.py:809] (0/4) Epoch 22, batch 3850, loss[ctc_loss=0.06801, att_loss=0.225, loss=0.1936, over 16128.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005483, over 42.00 utterances.], tot_loss[ctc_loss=0.07179, att_loss=0.2348, loss=0.2022, over 3280427.35 frames. utt_duration=1250 frames, utt_pad_proportion=0.05074, over 10508.39 utterances.], batch size: 42, lr: 4.84e-03, grad_scale: 16.0 2023-03-09 01:23:54,824 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87518.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:24:01,282 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87522.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:24:19,679 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87534.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:24:26,194 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-03-09 01:24:33,080 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 1.913e+02 2.224e+02 2.689e+02 7.925e+02, threshold=4.448e+02, percent-clipped=4.0 2023-03-09 01:24:57,881 INFO [train2.py:809] (0/4) Epoch 22, batch 3900, loss[ctc_loss=0.1232, att_loss=0.2675, loss=0.2386, over 13788.00 frames. utt_duration=379.4 frames, utt_pad_proportion=0.3391, over 146.00 utterances.], tot_loss[ctc_loss=0.07222, att_loss=0.2348, loss=0.2022, over 3270106.79 frames. utt_duration=1249 frames, utt_pad_proportion=0.05501, over 10489.19 utterances.], batch size: 146, lr: 4.84e-03, grad_scale: 16.0 2023-03-09 01:25:08,970 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87566.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 01:25:33,460 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87582.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:25:36,883 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87584.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:25:39,646 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87586.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:25:39,698 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8557, 5.2532, 4.7828, 5.3187, 4.6107, 4.9733, 5.3535, 5.0606], device='cuda:0'), covar=tensor([0.0696, 0.0243, 0.0785, 0.0288, 0.0440, 0.0274, 0.0225, 0.0209], device='cuda:0'), in_proj_covar=tensor([0.0388, 0.0321, 0.0367, 0.0352, 0.0323, 0.0238, 0.0304, 0.0287], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 01:25:44,716 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87589.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:26:09,219 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=87605.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:26:14,864 INFO [train2.py:809] (0/4) Epoch 22, batch 3950, loss[ctc_loss=0.05985, att_loss=0.2311, loss=0.1968, over 16761.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006323, over 48.00 utterances.], tot_loss[ctc_loss=0.07214, att_loss=0.2348, loss=0.2023, over 3272858.53 frames. utt_duration=1240 frames, utt_pad_proportion=0.05643, over 10570.00 utterances.], batch size: 48, lr: 4.84e-03, grad_scale: 16.0 2023-03-09 01:27:06,349 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-22.pt 2023-03-09 01:27:30,755 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.383e+02 1.991e+02 2.302e+02 2.874e+02 7.527e+02, threshold=4.603e+02, percent-clipped=2.0 2023-03-09 01:27:30,801 INFO [train2.py:809] (0/4) Epoch 23, batch 0, loss[ctc_loss=0.0656, att_loss=0.2172, loss=0.1869, over 14498.00 frames. utt_duration=1813 frames, utt_pad_proportion=0.03744, over 32.00 utterances.], tot_loss[ctc_loss=0.0656, att_loss=0.2172, loss=0.1869, over 14498.00 frames. utt_duration=1813 frames, utt_pad_proportion=0.03744, over 32.00 utterances.], batch size: 32, lr: 4.73e-03, grad_scale: 16.0 2023-03-09 01:27:30,803 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 01:27:42,674 INFO [train2.py:843] (0/4) Epoch 23, validation: ctc_loss=0.04039, att_loss=0.2346, loss=0.1958, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 01:27:42,675 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 01:27:46,045 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87645.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 01:27:53,554 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87650.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:28:05,997 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87658.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 01:28:18,395 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=87666.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 01:28:58,591 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-09 01:29:02,028 INFO [train2.py:809] (0/4) Epoch 23, batch 50, loss[ctc_loss=0.09385, att_loss=0.2597, loss=0.2265, over 16788.00 frames. utt_duration=686.9 frames, utt_pad_proportion=0.1392, over 98.00 utterances.], tot_loss[ctc_loss=0.07333, att_loss=0.2355, loss=0.2031, over 733261.25 frames. utt_duration=1198 frames, utt_pad_proportion=0.07171, over 2450.89 utterances.], batch size: 98, lr: 4.73e-03, grad_scale: 16.0 2023-03-09 01:29:22,819 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87706.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:29:41,129 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87718.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:29:54,234 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0477, 4.3394, 4.2495, 4.6300, 2.4717, 4.4445, 2.5370, 1.7262], device='cuda:0'), covar=tensor([0.0474, 0.0238, 0.0733, 0.0260, 0.1792, 0.0209, 0.1598, 0.1775], device='cuda:0'), in_proj_covar=tensor([0.0198, 0.0172, 0.0264, 0.0161, 0.0224, 0.0153, 0.0234, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:30:21,899 INFO [train2.py:809] (0/4) Epoch 23, batch 100, loss[ctc_loss=0.09303, att_loss=0.2475, loss=0.2166, over 16871.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007899, over 49.00 utterances.], tot_loss[ctc_loss=0.07288, att_loss=0.2357, loss=0.2032, over 1308110.50 frames. utt_duration=1237 frames, utt_pad_proportion=0.04996, over 4234.60 utterances.], batch size: 49, lr: 4.73e-03, grad_scale: 8.0 2023-03-09 01:30:23,394 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 2.012e+02 2.393e+02 2.818e+02 6.680e+02, threshold=4.785e+02, percent-clipped=2.0 2023-03-09 01:30:42,576 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2813, 2.9087, 3.1636, 4.2780, 3.8849, 3.8003, 2.9382, 2.3290], device='cuda:0'), covar=tensor([0.0780, 0.1856, 0.1015, 0.0597, 0.0888, 0.0484, 0.1483, 0.2144], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0217, 0.0192, 0.0224, 0.0229, 0.0183, 0.0205, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:31:41,407 INFO [train2.py:809] (0/4) Epoch 23, batch 150, loss[ctc_loss=0.05583, att_loss=0.2071, loss=0.1769, over 15766.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008903, over 38.00 utterances.], tot_loss[ctc_loss=0.07152, att_loss=0.2344, loss=0.2018, over 1740432.82 frames. utt_duration=1298 frames, utt_pad_proportion=0.03863, over 5367.88 utterances.], batch size: 38, lr: 4.73e-03, grad_scale: 8.0 2023-03-09 01:32:20,503 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87817.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:32:55,592 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.76 vs. limit=5.0 2023-03-09 01:33:02,591 INFO [train2.py:809] (0/4) Epoch 23, batch 200, loss[ctc_loss=0.06059, att_loss=0.2181, loss=0.1866, over 16524.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.00571, over 45.00 utterances.], tot_loss[ctc_loss=0.07058, att_loss=0.2351, loss=0.2022, over 2086027.54 frames. utt_duration=1291 frames, utt_pad_proportion=0.03961, over 6471.05 utterances.], batch size: 45, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:33:04,011 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.173e+02 1.980e+02 2.350e+02 2.635e+02 6.250e+02, threshold=4.701e+02, percent-clipped=5.0 2023-03-09 01:33:16,348 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.97 vs. limit=2.0 2023-03-09 01:33:31,410 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87861.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:34:11,186 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=87886.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:34:22,278 INFO [train2.py:809] (0/4) Epoch 23, batch 250, loss[ctc_loss=0.08888, att_loss=0.2547, loss=0.2215, over 17458.00 frames. utt_duration=885.4 frames, utt_pad_proportion=0.07289, over 79.00 utterances.], tot_loss[ctc_loss=0.06998, att_loss=0.2336, loss=0.2008, over 2334192.75 frames. utt_duration=1267 frames, utt_pad_proportion=0.04965, over 7379.47 utterances.], batch size: 79, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:35:27,224 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=87934.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:35:37,865 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87940.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:35:42,330 INFO [train2.py:809] (0/4) Epoch 23, batch 300, loss[ctc_loss=0.05358, att_loss=0.2045, loss=0.1743, over 15372.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01106, over 35.00 utterances.], tot_loss[ctc_loss=0.06987, att_loss=0.2339, loss=0.2011, over 2547949.10 frames. utt_duration=1302 frames, utt_pad_proportion=0.04, over 7837.00 utterances.], batch size: 35, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:35:43,864 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.187e+02 1.810e+02 2.179e+02 2.816e+02 7.106e+02, threshold=4.359e+02, percent-clipped=3.0 2023-03-09 01:35:45,670 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87945.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:36:10,731 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=87961.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 01:36:15,618 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1780, 2.8545, 3.1720, 4.2126, 3.8059, 3.7543, 2.8293, 2.1049], device='cuda:0'), covar=tensor([0.0862, 0.1896, 0.1005, 0.0782, 0.0960, 0.0536, 0.1668, 0.2433], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0215, 0.0191, 0.0223, 0.0227, 0.0182, 0.0204, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:36:55,194 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-03-09 01:37:02,478 INFO [train2.py:809] (0/4) Epoch 23, batch 350, loss[ctc_loss=0.06745, att_loss=0.2205, loss=0.1899, over 15950.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006626, over 41.00 utterances.], tot_loss[ctc_loss=0.07089, att_loss=0.2337, loss=0.2011, over 2696614.10 frames. utt_duration=1255 frames, utt_pad_proportion=0.05477, over 8603.52 utterances.], batch size: 41, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:37:14,382 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-88000.pt 2023-03-09 01:37:46,409 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88018.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:38:27,600 INFO [train2.py:809] (0/4) Epoch 23, batch 400, loss[ctc_loss=0.067, att_loss=0.2184, loss=0.1882, over 15793.00 frames. utt_duration=1664 frames, utt_pad_proportion=0.007427, over 38.00 utterances.], tot_loss[ctc_loss=0.0711, att_loss=0.2331, loss=0.2007, over 2821310.45 frames. utt_duration=1255 frames, utt_pad_proportion=0.05564, over 9004.62 utterances.], batch size: 38, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:38:29,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.200e+02 1.944e+02 2.395e+02 3.108e+02 5.272e+02, threshold=4.790e+02, percent-clipped=4.0 2023-03-09 01:39:04,772 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88066.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:39:48,105 INFO [train2.py:809] (0/4) Epoch 23, batch 450, loss[ctc_loss=0.07505, att_loss=0.2286, loss=0.1979, over 16397.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007918, over 44.00 utterances.], tot_loss[ctc_loss=0.07228, att_loss=0.2341, loss=0.2017, over 2918335.46 frames. utt_duration=1216 frames, utt_pad_proportion=0.067, over 9613.77 utterances.], batch size: 44, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:40:00,134 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88100.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:40:10,126 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 01:40:26,307 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88117.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:40:56,990 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-03-09 01:41:09,083 INFO [train2.py:809] (0/4) Epoch 23, batch 500, loss[ctc_loss=0.101, att_loss=0.2566, loss=0.2255, over 13537.00 frames. utt_duration=372.3 frames, utt_pad_proportion=0.3514, over 146.00 utterances.], tot_loss[ctc_loss=0.07203, att_loss=0.2347, loss=0.2022, over 2999340.17 frames. utt_duration=1206 frames, utt_pad_proportion=0.06824, over 9959.95 utterances.], batch size: 146, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:41:10,582 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 1.977e+02 2.397e+02 3.059e+02 5.388e+02, threshold=4.794e+02, percent-clipped=2.0 2023-03-09 01:41:37,481 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88161.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:41:37,541 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4879, 2.5309, 2.5621, 2.2650, 2.5054, 2.3633, 2.5831, 1.9260], device='cuda:0'), covar=tensor([0.1320, 0.1550, 0.2016, 0.3717, 0.1614, 0.2518, 0.1799, 0.4362], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0193, 0.0206, 0.0259, 0.0165, 0.0267, 0.0189, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:41:37,562 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88161.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:41:44,233 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88165.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:41:50,490 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6227, 4.8816, 4.5211, 4.9447, 4.4148, 4.6108, 4.9850, 4.7828], device='cuda:0'), covar=tensor([0.0605, 0.0315, 0.0715, 0.0412, 0.0417, 0.0389, 0.0271, 0.0221], device='cuda:0'), in_proj_covar=tensor([0.0384, 0.0320, 0.0362, 0.0351, 0.0321, 0.0238, 0.0303, 0.0288], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0005, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 01:42:29,385 INFO [train2.py:809] (0/4) Epoch 23, batch 550, loss[ctc_loss=0.1126, att_loss=0.2688, loss=0.2375, over 14150.00 frames. utt_duration=389.3 frames, utt_pad_proportion=0.3218, over 146.00 utterances.], tot_loss[ctc_loss=0.07296, att_loss=0.2351, loss=0.2027, over 3058835.51 frames. utt_duration=1189 frames, utt_pad_proportion=0.07219, over 10306.31 utterances.], batch size: 146, lr: 4.72e-03, grad_scale: 8.0 2023-03-09 01:42:41,105 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1134, 4.3749, 4.7495, 4.5492, 2.8981, 4.5186, 3.1650, 2.1061], device='cuda:0'), covar=tensor([0.0538, 0.0358, 0.0554, 0.0277, 0.1540, 0.0241, 0.1272, 0.1689], device='cuda:0'), in_proj_covar=tensor([0.0200, 0.0172, 0.0265, 0.0164, 0.0226, 0.0154, 0.0235, 0.0208], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:42:54,630 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88209.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:43:46,350 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88240.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:43:51,893 INFO [train2.py:809] (0/4) Epoch 23, batch 600, loss[ctc_loss=0.05293, att_loss=0.2064, loss=0.1757, over 15392.00 frames. utt_duration=1760 frames, utt_pad_proportion=0.009867, over 35.00 utterances.], tot_loss[ctc_loss=0.07148, att_loss=0.2339, loss=0.2014, over 3104161.25 frames. utt_duration=1242 frames, utt_pad_proportion=0.06028, over 10011.85 utterances.], batch size: 35, lr: 4.71e-03, grad_scale: 8.0 2023-03-09 01:43:53,455 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 1.886e+02 2.333e+02 2.841e+02 4.176e+02, threshold=4.666e+02, percent-clipped=0.0 2023-03-09 01:43:55,498 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88245.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:44:20,774 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88261.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 01:45:04,346 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88288.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:45:10,125 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=88291.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:45:12,842 INFO [train2.py:809] (0/4) Epoch 23, batch 650, loss[ctc_loss=0.06342, att_loss=0.2412, loss=0.2057, over 17151.00 frames. utt_duration=1227 frames, utt_pad_proportion=0.01323, over 56.00 utterances.], tot_loss[ctc_loss=0.07151, att_loss=0.2339, loss=0.2014, over 3140271.30 frames. utt_duration=1241 frames, utt_pad_proportion=0.0594, over 10132.09 utterances.], batch size: 56, lr: 4.71e-03, grad_scale: 8.0 2023-03-09 01:45:12,991 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88293.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:45:30,368 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2384, 5.1956, 4.7614, 2.9665, 5.0276, 4.9088, 4.4917, 2.5487], device='cuda:0'), covar=tensor([0.0151, 0.0125, 0.0411, 0.1289, 0.0111, 0.0192, 0.0399, 0.2103], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0102, 0.0106, 0.0111, 0.0086, 0.0115, 0.0100, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 01:45:38,312 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88309.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:46:20,237 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1153, 4.4047, 4.4261, 4.7025, 2.7785, 4.4598, 2.8429, 1.9745], device='cuda:0'), covar=tensor([0.0421, 0.0270, 0.0632, 0.0250, 0.1588, 0.0211, 0.1369, 0.1607], device='cuda:0'), in_proj_covar=tensor([0.0197, 0.0169, 0.0262, 0.0162, 0.0222, 0.0152, 0.0230, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:46:22,682 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-09 01:46:34,459 INFO [train2.py:809] (0/4) Epoch 23, batch 700, loss[ctc_loss=0.06659, att_loss=0.2264, loss=0.1944, over 16273.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.0076, over 43.00 utterances.], tot_loss[ctc_loss=0.07151, att_loss=0.2337, loss=0.2013, over 3165142.33 frames. utt_duration=1236 frames, utt_pad_proportion=0.06028, over 10252.75 utterances.], batch size: 43, lr: 4.71e-03, grad_scale: 8.0 2023-03-09 01:46:35,998 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.287e+02 1.968e+02 2.314e+02 2.837e+02 1.022e+03, threshold=4.627e+02, percent-clipped=4.0 2023-03-09 01:46:49,090 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=88352.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 01:47:56,424 INFO [train2.py:809] (0/4) Epoch 23, batch 750, loss[ctc_loss=0.05137, att_loss=0.2256, loss=0.1907, over 16266.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007285, over 43.00 utterances.], tot_loss[ctc_loss=0.07152, att_loss=0.2341, loss=0.2016, over 3190334.78 frames. utt_duration=1222 frames, utt_pad_proportion=0.06211, over 10452.19 utterances.], batch size: 43, lr: 4.71e-03, grad_scale: 8.0 2023-03-09 01:48:17,844 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1331, 3.8234, 3.8093, 3.2342, 3.8329, 3.9264, 3.8723, 2.9512], device='cuda:0'), covar=tensor([0.0961, 0.1267, 0.1722, 0.3297, 0.0999, 0.1450, 0.1020, 0.3457], device='cuda:0'), in_proj_covar=tensor([0.0186, 0.0193, 0.0206, 0.0260, 0.0165, 0.0267, 0.0189, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:49:15,196 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1155, 4.4465, 4.5256, 4.7510, 2.8701, 4.3763, 2.8407, 1.7480], device='cuda:0'), covar=tensor([0.0484, 0.0294, 0.0635, 0.0213, 0.1520, 0.0265, 0.1343, 0.1735], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0169, 0.0260, 0.0161, 0.0221, 0.0152, 0.0230, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:49:17,982 INFO [train2.py:809] (0/4) Epoch 23, batch 800, loss[ctc_loss=0.08494, att_loss=0.2515, loss=0.2182, over 17113.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.0154, over 56.00 utterances.], tot_loss[ctc_loss=0.07173, att_loss=0.2344, loss=0.2019, over 3209993.02 frames. utt_duration=1233 frames, utt_pad_proportion=0.05973, over 10427.96 utterances.], batch size: 56, lr: 4.71e-03, grad_scale: 8.0 2023-03-09 01:49:19,538 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.200e+02 1.999e+02 2.331e+02 2.822e+02 6.920e+02, threshold=4.662e+02, percent-clipped=4.0 2023-03-09 01:49:39,399 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88456.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:49:55,863 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5158, 1.9301, 2.0644, 2.1505, 2.0344, 2.2717, 1.8112, 2.4660], device='cuda:0'), covar=tensor([0.1304, 0.2523, 0.2054, 0.1444, 0.2251, 0.1320, 0.1614, 0.1415], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0130, 0.0127, 0.0117, 0.0130, 0.0112, 0.0136, 0.0107], device='cuda:0'), out_proj_covar=tensor([9.5376e-05, 1.0109e-04, 1.0136e-04, 9.1207e-05, 9.7475e-05, 9.0540e-05, 1.0311e-04, 8.5206e-05], device='cuda:0') 2023-03-09 01:50:00,458 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0420, 4.9843, 4.7461, 3.1365, 4.6980, 4.6608, 4.2840, 2.6906], device='cuda:0'), covar=tensor([0.0120, 0.0102, 0.0310, 0.0972, 0.0113, 0.0200, 0.0333, 0.1358], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0101, 0.0105, 0.0111, 0.0085, 0.0114, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 01:50:32,697 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-09 01:50:39,455 INFO [train2.py:809] (0/4) Epoch 23, batch 850, loss[ctc_loss=0.04012, att_loss=0.215, loss=0.18, over 16395.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007198, over 44.00 utterances.], tot_loss[ctc_loss=0.07143, att_loss=0.2346, loss=0.202, over 3231239.79 frames. utt_duration=1218 frames, utt_pad_proportion=0.06043, over 10626.00 utterances.], batch size: 44, lr: 4.71e-03, grad_scale: 8.0 2023-03-09 01:51:36,755 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5044, 2.3911, 4.8420, 3.9344, 2.9785, 4.2250, 4.4137, 4.5816], device='cuda:0'), covar=tensor([0.0166, 0.1361, 0.0150, 0.0718, 0.1553, 0.0212, 0.0175, 0.0200], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0244, 0.0194, 0.0319, 0.0266, 0.0218, 0.0184, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:51:41,315 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.42 vs. limit=5.0 2023-03-09 01:52:02,088 INFO [train2.py:809] (0/4) Epoch 23, batch 900, loss[ctc_loss=0.07018, att_loss=0.2356, loss=0.2025, over 16968.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007363, over 50.00 utterances.], tot_loss[ctc_loss=0.07069, att_loss=0.2344, loss=0.2017, over 3244318.38 frames. utt_duration=1247 frames, utt_pad_proportion=0.0525, over 10422.14 utterances.], batch size: 50, lr: 4.71e-03, grad_scale: 8.0 2023-03-09 01:52:03,749 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 1.915e+02 2.378e+02 3.006e+02 6.830e+02, threshold=4.756e+02, percent-clipped=6.0 2023-03-09 01:52:35,956 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-09 01:53:24,722 INFO [train2.py:809] (0/4) Epoch 23, batch 950, loss[ctc_loss=0.08297, att_loss=0.2412, loss=0.2095, over 17325.00 frames. utt_duration=1006 frames, utt_pad_proportion=0.0512, over 69.00 utterances.], tot_loss[ctc_loss=0.07127, att_loss=0.235, loss=0.2022, over 3247172.86 frames. utt_duration=1209 frames, utt_pad_proportion=0.06235, over 10754.30 utterances.], batch size: 69, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 01:54:46,091 INFO [train2.py:809] (0/4) Epoch 23, batch 1000, loss[ctc_loss=0.07846, att_loss=0.2396, loss=0.2073, over 17045.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008512, over 52.00 utterances.], tot_loss[ctc_loss=0.07087, att_loss=0.2347, loss=0.202, over 3261713.31 frames. utt_duration=1234 frames, utt_pad_proportion=0.05435, over 10588.79 utterances.], batch size: 52, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 01:54:48,312 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.066e+02 1.864e+02 2.157e+02 2.686e+02 4.602e+02, threshold=4.313e+02, percent-clipped=0.0 2023-03-09 01:54:53,439 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=88647.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 01:56:09,553 INFO [train2.py:809] (0/4) Epoch 23, batch 1050, loss[ctc_loss=0.05296, att_loss=0.2006, loss=0.1711, over 15345.00 frames. utt_duration=1755 frames, utt_pad_proportion=0.01287, over 35.00 utterances.], tot_loss[ctc_loss=0.07058, att_loss=0.234, loss=0.2013, over 3257554.03 frames. utt_duration=1233 frames, utt_pad_proportion=0.05775, over 10583.08 utterances.], batch size: 35, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 01:56:13,268 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7353, 3.2986, 3.8310, 3.3417, 3.7292, 4.7168, 4.5576, 3.5002], device='cuda:0'), covar=tensor([0.0348, 0.1491, 0.1276, 0.1244, 0.1040, 0.0997, 0.0658, 0.1187], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0245, 0.0285, 0.0223, 0.0268, 0.0376, 0.0266, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 01:57:01,074 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3158, 3.9025, 3.3600, 3.6557, 4.1299, 3.8495, 3.2784, 4.4161], device='cuda:0'), covar=tensor([0.0893, 0.0503, 0.0995, 0.0626, 0.0638, 0.0652, 0.0750, 0.0525], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0219, 0.0229, 0.0204, 0.0283, 0.0245, 0.0201, 0.0292], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 01:57:30,983 INFO [train2.py:809] (0/4) Epoch 23, batch 1100, loss[ctc_loss=0.06667, att_loss=0.2372, loss=0.2031, over 16963.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007032, over 50.00 utterances.], tot_loss[ctc_loss=0.07148, att_loss=0.235, loss=0.2023, over 3263777.02 frames. utt_duration=1203 frames, utt_pad_proportion=0.06436, over 10867.36 utterances.], batch size: 50, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 01:57:32,495 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 2.060e+02 2.487e+02 3.193e+02 8.233e+02, threshold=4.974e+02, percent-clipped=4.0 2023-03-09 01:57:52,268 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88756.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 01:58:48,560 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5421, 2.9188, 3.5271, 4.4975, 3.9922, 3.8960, 2.9774, 2.3190], device='cuda:0'), covar=tensor([0.0670, 0.1897, 0.0863, 0.0526, 0.0778, 0.0535, 0.1547, 0.2155], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0216, 0.0190, 0.0224, 0.0230, 0.0182, 0.0204, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 01:58:52,972 INFO [train2.py:809] (0/4) Epoch 23, batch 1150, loss[ctc_loss=0.08349, att_loss=0.2301, loss=0.2008, over 15643.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008988, over 37.00 utterances.], tot_loss[ctc_loss=0.07144, att_loss=0.2345, loss=0.2019, over 3266331.83 frames. utt_duration=1217 frames, utt_pad_proportion=0.06181, over 10747.95 utterances.], batch size: 37, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 01:58:56,125 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.33 vs. limit=5.0 2023-03-09 01:59:11,746 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88804.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:00:16,053 INFO [train2.py:809] (0/4) Epoch 23, batch 1200, loss[ctc_loss=0.08893, att_loss=0.2478, loss=0.216, over 17376.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03455, over 63.00 utterances.], tot_loss[ctc_loss=0.07094, att_loss=0.2342, loss=0.2015, over 3271078.95 frames. utt_duration=1247 frames, utt_pad_proportion=0.0535, over 10506.59 utterances.], batch size: 63, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 02:00:17,521 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 1.914e+02 2.277e+02 2.770e+02 5.405e+02, threshold=4.555e+02, percent-clipped=1.0 2023-03-09 02:01:37,928 INFO [train2.py:809] (0/4) Epoch 23, batch 1250, loss[ctc_loss=0.07234, att_loss=0.2391, loss=0.2058, over 16951.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008415, over 50.00 utterances.], tot_loss[ctc_loss=0.07083, att_loss=0.234, loss=0.2013, over 3272265.67 frames. utt_duration=1264 frames, utt_pad_proportion=0.04881, over 10363.43 utterances.], batch size: 50, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 02:02:41,993 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0054, 5.2668, 5.1670, 5.1019, 5.2899, 5.2222, 4.9094, 4.7016], device='cuda:0'), covar=tensor([0.1000, 0.0579, 0.0315, 0.0629, 0.0293, 0.0332, 0.0387, 0.0369], device='cuda:0'), in_proj_covar=tensor([0.0528, 0.0373, 0.0357, 0.0368, 0.0432, 0.0444, 0.0366, 0.0404], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 02:03:00,075 INFO [train2.py:809] (0/4) Epoch 23, batch 1300, loss[ctc_loss=0.06098, att_loss=0.2403, loss=0.2044, over 16897.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.006666, over 49.00 utterances.], tot_loss[ctc_loss=0.07141, att_loss=0.2344, loss=0.2018, over 3273510.98 frames. utt_duration=1245 frames, utt_pad_proportion=0.05423, over 10529.92 utterances.], batch size: 49, lr: 4.70e-03, grad_scale: 8.0 2023-03-09 02:03:01,704 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.195e+02 1.977e+02 2.384e+02 3.004e+02 6.014e+02, threshold=4.768e+02, percent-clipped=3.0 2023-03-09 02:03:06,841 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=88947.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 02:03:32,320 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-09 02:03:39,800 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-09 02:04:21,366 INFO [train2.py:809] (0/4) Epoch 23, batch 1350, loss[ctc_loss=0.06871, att_loss=0.2299, loss=0.1977, over 16297.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006381, over 43.00 utterances.], tot_loss[ctc_loss=0.07148, att_loss=0.2344, loss=0.2018, over 3274321.48 frames. utt_duration=1227 frames, utt_pad_proportion=0.05842, over 10690.14 utterances.], batch size: 43, lr: 4.69e-03, grad_scale: 8.0 2023-03-09 02:04:24,660 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=88995.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 02:05:07,439 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5815, 3.1415, 3.6953, 3.3159, 3.6225, 4.6266, 4.4786, 3.4873], device='cuda:0'), covar=tensor([0.0355, 0.1520, 0.1135, 0.1177, 0.1023, 0.0845, 0.0535, 0.1105], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0243, 0.0282, 0.0221, 0.0265, 0.0372, 0.0264, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 02:05:17,146 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89027.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:05:43,319 INFO [train2.py:809] (0/4) Epoch 23, batch 1400, loss[ctc_loss=0.08874, att_loss=0.2642, loss=0.2291, over 17097.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01481, over 56.00 utterances.], tot_loss[ctc_loss=0.07234, att_loss=0.2355, loss=0.2029, over 3281553.75 frames. utt_duration=1226 frames, utt_pad_proportion=0.05754, over 10717.32 utterances.], batch size: 56, lr: 4.69e-03, grad_scale: 8.0 2023-03-09 02:05:44,871 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.382e+02 1.935e+02 2.297e+02 2.917e+02 5.540e+02, threshold=4.593e+02, percent-clipped=1.0 2023-03-09 02:06:47,114 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9892, 4.2783, 4.4147, 4.6976, 2.8057, 4.3387, 2.8266, 1.8258], device='cuda:0'), covar=tensor([0.0472, 0.0317, 0.0686, 0.0221, 0.1658, 0.0230, 0.1451, 0.1787], device='cuda:0'), in_proj_covar=tensor([0.0196, 0.0169, 0.0260, 0.0160, 0.0223, 0.0153, 0.0232, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:06:55,148 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3503, 2.4566, 4.8958, 3.7681, 3.0235, 4.2064, 4.6846, 4.4196], device='cuda:0'), covar=tensor([0.0302, 0.1666, 0.0171, 0.0959, 0.1694, 0.0252, 0.0164, 0.0315], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0244, 0.0194, 0.0318, 0.0265, 0.0218, 0.0184, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:06:56,660 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89088.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:07:04,592 INFO [train2.py:809] (0/4) Epoch 23, batch 1450, loss[ctc_loss=0.06772, att_loss=0.2348, loss=0.2014, over 16981.00 frames. utt_duration=1360 frames, utt_pad_proportion=0.00614, over 50.00 utterances.], tot_loss[ctc_loss=0.07235, att_loss=0.2353, loss=0.2027, over 3279327.42 frames. utt_duration=1236 frames, utt_pad_proportion=0.05606, over 10626.55 utterances.], batch size: 50, lr: 4.69e-03, grad_scale: 8.0 2023-03-09 02:08:26,637 INFO [train2.py:809] (0/4) Epoch 23, batch 1500, loss[ctc_loss=0.046, att_loss=0.2035, loss=0.172, over 16277.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006797, over 43.00 utterances.], tot_loss[ctc_loss=0.07232, att_loss=0.2352, loss=0.2026, over 3277263.79 frames. utt_duration=1249 frames, utt_pad_proportion=0.05405, over 10508.08 utterances.], batch size: 43, lr: 4.69e-03, grad_scale: 8.0 2023-03-09 02:08:28,106 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.367e+02 1.977e+02 2.384e+02 2.734e+02 5.349e+02, threshold=4.768e+02, percent-clipped=3.0 2023-03-09 02:08:34,880 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4161, 4.4815, 4.1350, 2.7595, 4.2866, 4.2185, 3.7528, 2.3507], device='cuda:0'), covar=tensor([0.0167, 0.0133, 0.0358, 0.1222, 0.0128, 0.0325, 0.0422, 0.1793], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0101, 0.0105, 0.0110, 0.0085, 0.0113, 0.0099, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 02:08:47,270 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1319, 5.4239, 4.8721, 5.4903, 4.8882, 5.0769, 5.5049, 5.2608], device='cuda:0'), covar=tensor([0.0546, 0.0281, 0.0833, 0.0303, 0.0384, 0.0260, 0.0230, 0.0203], device='cuda:0'), in_proj_covar=tensor([0.0389, 0.0325, 0.0366, 0.0354, 0.0324, 0.0239, 0.0306, 0.0289], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 02:09:08,697 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89169.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:09:48,306 INFO [train2.py:809] (0/4) Epoch 23, batch 1550, loss[ctc_loss=0.06017, att_loss=0.2092, loss=0.1794, over 14525.00 frames. utt_duration=1817 frames, utt_pad_proportion=0.03454, over 32.00 utterances.], tot_loss[ctc_loss=0.07192, att_loss=0.2351, loss=0.2025, over 3285246.44 frames. utt_duration=1261 frames, utt_pad_proportion=0.04848, over 10429.81 utterances.], batch size: 32, lr: 4.69e-03, grad_scale: 8.0 2023-03-09 02:10:28,362 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6565, 2.7181, 3.8566, 3.5339, 3.0008, 3.6911, 3.6916, 3.6515], device='cuda:0'), covar=tensor([0.0339, 0.1158, 0.0191, 0.0723, 0.1209, 0.0289, 0.0237, 0.0354], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0244, 0.0194, 0.0318, 0.0264, 0.0219, 0.0185, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:10:49,443 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89230.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:11:10,191 INFO [train2.py:809] (0/4) Epoch 23, batch 1600, loss[ctc_loss=0.06521, att_loss=0.2343, loss=0.2005, over 16467.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.007226, over 46.00 utterances.], tot_loss[ctc_loss=0.07228, att_loss=0.235, loss=0.2025, over 3269289.89 frames. utt_duration=1225 frames, utt_pad_proportion=0.0639, over 10689.38 utterances.], batch size: 46, lr: 4.69e-03, grad_scale: 8.0 2023-03-09 02:11:11,787 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.099e+02 1.895e+02 2.282e+02 2.881e+02 5.348e+02, threshold=4.565e+02, percent-clipped=1.0 2023-03-09 02:12:31,879 INFO [train2.py:809] (0/4) Epoch 23, batch 1650, loss[ctc_loss=0.05245, att_loss=0.2214, loss=0.1876, over 16407.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006548, over 44.00 utterances.], tot_loss[ctc_loss=0.07141, att_loss=0.2343, loss=0.2017, over 3272750.88 frames. utt_duration=1257 frames, utt_pad_proportion=0.05489, over 10425.00 utterances.], batch size: 44, lr: 4.69e-03, grad_scale: 8.0 2023-03-09 02:13:54,204 INFO [train2.py:809] (0/4) Epoch 23, batch 1700, loss[ctc_loss=0.06367, att_loss=0.2159, loss=0.1855, over 16018.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.00666, over 40.00 utterances.], tot_loss[ctc_loss=0.07156, att_loss=0.2343, loss=0.2018, over 3268235.50 frames. utt_duration=1240 frames, utt_pad_proportion=0.05968, over 10556.29 utterances.], batch size: 40, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:13:55,725 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 1.978e+02 2.274e+02 2.807e+02 5.898e+02, threshold=4.548e+02, percent-clipped=3.0 2023-03-09 02:14:45,033 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89374.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:15:00,015 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89383.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:15:05,018 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6023, 2.1822, 2.5264, 2.8682, 2.6616, 2.4991, 2.4029, 2.9936], device='cuda:0'), covar=tensor([0.1786, 0.3339, 0.1951, 0.1519, 0.1836, 0.1714, 0.2508, 0.1164], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0129, 0.0125, 0.0116, 0.0131, 0.0113, 0.0135, 0.0107], device='cuda:0'), out_proj_covar=tensor([9.5784e-05, 1.0081e-04, 1.0026e-04, 9.0756e-05, 9.8042e-05, 9.1012e-05, 1.0299e-04, 8.5470e-05], device='cuda:0') 2023-03-09 02:15:15,728 INFO [train2.py:809] (0/4) Epoch 23, batch 1750, loss[ctc_loss=0.0903, att_loss=0.2496, loss=0.2177, over 16967.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007582, over 50.00 utterances.], tot_loss[ctc_loss=0.07193, att_loss=0.2346, loss=0.2021, over 3265981.85 frames. utt_duration=1233 frames, utt_pad_proportion=0.0611, over 10607.12 utterances.], batch size: 50, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:16:05,777 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7478, 3.3505, 3.8155, 3.4439, 3.7509, 4.7861, 4.6196, 3.6751], device='cuda:0'), covar=tensor([0.0312, 0.1384, 0.1074, 0.1094, 0.0863, 0.0842, 0.0490, 0.0975], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0246, 0.0284, 0.0223, 0.0269, 0.0376, 0.0267, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 02:16:08,992 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89425.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:16:24,734 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.98 vs. limit=5.0 2023-03-09 02:16:25,512 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89435.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:16:28,706 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4416, 2.3025, 2.0832, 2.7099, 2.5735, 2.1880, 2.4062, 2.7482], device='cuda:0'), covar=tensor([0.2040, 0.2909, 0.2049, 0.1435, 0.1942, 0.1866, 0.2221, 0.1823], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0128, 0.0124, 0.0115, 0.0131, 0.0113, 0.0135, 0.0107], device='cuda:0'), out_proj_covar=tensor([9.5371e-05, 1.0011e-04, 9.9605e-05, 9.0215e-05, 9.7585e-05, 9.0548e-05, 1.0230e-04, 8.5053e-05], device='cuda:0') 2023-03-09 02:16:31,827 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89439.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 02:16:37,776 INFO [train2.py:809] (0/4) Epoch 23, batch 1800, loss[ctc_loss=0.05632, att_loss=0.2288, loss=0.1943, over 16868.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.007611, over 49.00 utterances.], tot_loss[ctc_loss=0.07238, att_loss=0.2352, loss=0.2026, over 3270218.18 frames. utt_duration=1220 frames, utt_pad_proportion=0.06335, over 10737.87 utterances.], batch size: 49, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:16:39,280 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 1.934e+02 2.267e+02 2.788e+02 6.296e+02, threshold=4.533e+02, percent-clipped=2.0 2023-03-09 02:17:01,501 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2733, 3.7972, 3.2310, 3.5416, 4.0070, 3.7000, 3.2064, 4.4356], device='cuda:0'), covar=tensor([0.0863, 0.0489, 0.1067, 0.0641, 0.0643, 0.0709, 0.0726, 0.0406], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0217, 0.0228, 0.0203, 0.0280, 0.0243, 0.0199, 0.0292], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 02:17:30,328 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-09 02:17:38,944 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-03-09 02:17:49,765 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89486.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:18:00,960 INFO [train2.py:809] (0/4) Epoch 23, batch 1850, loss[ctc_loss=0.07867, att_loss=0.2399, loss=0.2077, over 17096.00 frames. utt_duration=692.5 frames, utt_pad_proportion=0.1322, over 99.00 utterances.], tot_loss[ctc_loss=0.07115, att_loss=0.2343, loss=0.2016, over 3262625.61 frames. utt_duration=1221 frames, utt_pad_proportion=0.06396, over 10697.67 utterances.], batch size: 99, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:18:01,344 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1702, 3.9360, 3.3877, 3.6293, 4.1233, 3.7589, 3.2669, 4.4523], device='cuda:0'), covar=tensor([0.1036, 0.0438, 0.1092, 0.0714, 0.0726, 0.0774, 0.0793, 0.0534], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0216, 0.0227, 0.0202, 0.0279, 0.0242, 0.0198, 0.0291], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 02:18:12,482 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89500.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 02:18:27,082 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89509.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:18:53,246 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89525.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:19:23,210 INFO [train2.py:809] (0/4) Epoch 23, batch 1900, loss[ctc_loss=0.1118, att_loss=0.2602, loss=0.2305, over 17056.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008236, over 52.00 utterances.], tot_loss[ctc_loss=0.0717, att_loss=0.2347, loss=0.2021, over 3264283.01 frames. utt_duration=1230 frames, utt_pad_proportion=0.06228, over 10627.60 utterances.], batch size: 52, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:19:24,724 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 1.857e+02 2.214e+02 2.613e+02 6.463e+02, threshold=4.427e+02, percent-clipped=1.0 2023-03-09 02:20:07,578 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89570.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:20:44,885 INFO [train2.py:809] (0/4) Epoch 23, batch 1950, loss[ctc_loss=0.05662, att_loss=0.2163, loss=0.1843, over 15758.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.008735, over 38.00 utterances.], tot_loss[ctc_loss=0.07207, att_loss=0.2346, loss=0.2021, over 3260820.63 frames. utt_duration=1237 frames, utt_pad_proportion=0.05982, over 10555.97 utterances.], batch size: 38, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:22:01,307 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7648, 3.4706, 3.3937, 2.9460, 3.4178, 3.4269, 3.5076, 2.6040], device='cuda:0'), covar=tensor([0.1164, 0.1251, 0.2392, 0.3781, 0.1896, 0.2741, 0.1050, 0.3525], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0189, 0.0203, 0.0256, 0.0162, 0.0263, 0.0188, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:22:06,987 INFO [train2.py:809] (0/4) Epoch 23, batch 2000, loss[ctc_loss=0.07308, att_loss=0.2474, loss=0.2125, over 16777.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005978, over 48.00 utterances.], tot_loss[ctc_loss=0.07209, att_loss=0.2344, loss=0.2019, over 3258160.98 frames. utt_duration=1240 frames, utt_pad_proportion=0.06046, over 10526.19 utterances.], batch size: 48, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:22:08,497 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 1.989e+02 2.307e+02 2.806e+02 7.990e+02, threshold=4.613e+02, percent-clipped=4.0 2023-03-09 02:22:34,292 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6656, 5.0564, 4.8261, 4.9640, 5.1271, 4.7495, 3.5375, 5.0824], device='cuda:0'), covar=tensor([0.0116, 0.0108, 0.0131, 0.0084, 0.0092, 0.0114, 0.0625, 0.0183], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0087, 0.0111, 0.0069, 0.0075, 0.0086, 0.0102, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 02:23:07,542 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89680.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:23:12,437 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89683.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:23:28,546 INFO [train2.py:809] (0/4) Epoch 23, batch 2050, loss[ctc_loss=0.06663, att_loss=0.2337, loss=0.2003, over 16125.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006387, over 42.00 utterances.], tot_loss[ctc_loss=0.07257, att_loss=0.2347, loss=0.2023, over 3255798.39 frames. utt_duration=1215 frames, utt_pad_proportion=0.0682, over 10728.62 utterances.], batch size: 42, lr: 4.68e-03, grad_scale: 8.0 2023-03-09 02:24:28,909 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89730.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:24:30,400 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=89731.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:24:47,318 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89741.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:24:50,039 INFO [train2.py:809] (0/4) Epoch 23, batch 2100, loss[ctc_loss=0.06464, att_loss=0.2289, loss=0.1961, over 16169.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007573, over 41.00 utterances.], tot_loss[ctc_loss=0.07309, att_loss=0.2355, loss=0.203, over 3262403.40 frames. utt_duration=1204 frames, utt_pad_proportion=0.06937, over 10849.49 utterances.], batch size: 41, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:24:51,496 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 1.974e+02 2.233e+02 2.769e+02 5.380e+02, threshold=4.466e+02, percent-clipped=5.0 2023-03-09 02:25:06,437 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9839, 5.2723, 4.8094, 5.3368, 4.7375, 4.9575, 5.3925, 5.1301], device='cuda:0'), covar=tensor([0.0576, 0.0324, 0.0832, 0.0355, 0.0400, 0.0237, 0.0219, 0.0212], device='cuda:0'), in_proj_covar=tensor([0.0390, 0.0326, 0.0367, 0.0354, 0.0324, 0.0240, 0.0308, 0.0292], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 02:25:52,105 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89781.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:26:11,805 INFO [train2.py:809] (0/4) Epoch 23, batch 2150, loss[ctc_loss=0.09268, att_loss=0.2563, loss=0.2236, over 16688.00 frames. utt_duration=675.9 frames, utt_pad_proportion=0.1487, over 99.00 utterances.], tot_loss[ctc_loss=0.0733, att_loss=0.2364, loss=0.2038, over 3271577.59 frames. utt_duration=1196 frames, utt_pad_proportion=0.06841, over 10959.68 utterances.], batch size: 99, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:26:15,133 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89795.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 02:27:04,685 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=89825.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:27:06,450 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5389, 2.7937, 5.0694, 4.1132, 3.1633, 4.3782, 4.8307, 4.7126], device='cuda:0'), covar=tensor([0.0305, 0.1371, 0.0216, 0.0837, 0.1644, 0.0265, 0.0184, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0242, 0.0195, 0.0318, 0.0265, 0.0219, 0.0186, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:27:33,615 INFO [train2.py:809] (0/4) Epoch 23, batch 2200, loss[ctc_loss=0.06717, att_loss=0.2218, loss=0.1909, over 16393.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.008069, over 44.00 utterances.], tot_loss[ctc_loss=0.07273, att_loss=0.2361, loss=0.2034, over 3275452.16 frames. utt_duration=1208 frames, utt_pad_proportion=0.06254, over 10860.90 utterances.], batch size: 44, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:27:34,943 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 1.973e+02 2.386e+02 3.013e+02 6.689e+02, threshold=4.771e+02, percent-clipped=5.0 2023-03-09 02:28:00,306 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=2.11 vs. limit=2.0 2023-03-09 02:28:09,093 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:28:09,288 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=89865.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:28:23,118 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=89873.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:28:54,729 INFO [train2.py:809] (0/4) Epoch 23, batch 2250, loss[ctc_loss=0.06291, att_loss=0.2371, loss=0.2022, over 16517.00 frames. utt_duration=1470 frames, utt_pad_proportion=0.007698, over 45.00 utterances.], tot_loss[ctc_loss=0.07278, att_loss=0.2357, loss=0.2031, over 3275814.42 frames. utt_duration=1215 frames, utt_pad_proportion=0.06095, over 10800.39 utterances.], batch size: 45, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:29:48,471 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=89926.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 02:29:55,413 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-03-09 02:30:14,731 INFO [train2.py:809] (0/4) Epoch 23, batch 2300, loss[ctc_loss=0.07236, att_loss=0.2467, loss=0.2118, over 16622.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005484, over 47.00 utterances.], tot_loss[ctc_loss=0.07389, att_loss=0.2364, loss=0.2039, over 3272909.72 frames. utt_duration=1185 frames, utt_pad_proportion=0.06898, over 11064.36 utterances.], batch size: 47, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:30:16,396 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 1.850e+02 2.227e+02 2.792e+02 5.012e+02, threshold=4.455e+02, percent-clipped=2.0 2023-03-09 02:31:00,713 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3726, 4.4672, 4.6047, 4.5284, 5.0095, 4.5286, 4.4459, 2.6983], device='cuda:0'), covar=tensor([0.0264, 0.0306, 0.0293, 0.0273, 0.0720, 0.0230, 0.0318, 0.1553], device='cuda:0'), in_proj_covar=tensor([0.0170, 0.0197, 0.0193, 0.0210, 0.0369, 0.0165, 0.0185, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:31:36,742 INFO [train2.py:809] (0/4) Epoch 23, batch 2350, loss[ctc_loss=0.08307, att_loss=0.2491, loss=0.2159, over 16613.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006171, over 47.00 utterances.], tot_loss[ctc_loss=0.07364, att_loss=0.2362, loss=0.2037, over 3269472.57 frames. utt_duration=1156 frames, utt_pad_proportion=0.07733, over 11326.81 utterances.], batch size: 47, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:31:49,153 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-90000.pt 2023-03-09 02:32:30,020 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4930, 2.6065, 4.9826, 4.0430, 3.2525, 4.4381, 4.8656, 4.6711], device='cuda:0'), covar=tensor([0.0308, 0.1521, 0.0229, 0.0858, 0.1482, 0.0233, 0.0183, 0.0285], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0243, 0.0195, 0.0317, 0.0263, 0.0218, 0.0186, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:32:42,537 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90030.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:32:52,445 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90036.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:33:03,246 INFO [train2.py:809] (0/4) Epoch 23, batch 2400, loss[ctc_loss=0.06083, att_loss=0.2083, loss=0.1788, over 15499.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008913, over 36.00 utterances.], tot_loss[ctc_loss=0.07296, att_loss=0.2358, loss=0.2033, over 3267583.93 frames. utt_duration=1183 frames, utt_pad_proportion=0.07184, over 11066.98 utterances.], batch size: 36, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:33:04,733 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 1.874e+02 2.405e+02 2.931e+02 6.452e+02, threshold=4.810e+02, percent-clipped=5.0 2023-03-09 02:34:00,211 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90078.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:34:05,180 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90081.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:34:24,277 INFO [train2.py:809] (0/4) Epoch 23, batch 2450, loss[ctc_loss=0.06195, att_loss=0.2115, loss=0.1816, over 16183.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006029, over 41.00 utterances.], tot_loss[ctc_loss=0.07232, att_loss=0.235, loss=0.2025, over 3274282.35 frames. utt_duration=1223 frames, utt_pad_proportion=0.06099, over 10726.43 utterances.], batch size: 41, lr: 4.67e-03, grad_scale: 16.0 2023-03-09 02:34:27,934 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90095.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 02:34:48,685 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-03-09 02:35:23,981 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90129.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:35:25,875 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90130.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:35:47,564 INFO [train2.py:809] (0/4) Epoch 23, batch 2500, loss[ctc_loss=0.06763, att_loss=0.2421, loss=0.2072, over 16633.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005005, over 47.00 utterances.], tot_loss[ctc_loss=0.0722, att_loss=0.2351, loss=0.2025, over 3274044.79 frames. utt_duration=1221 frames, utt_pad_proportion=0.06219, over 10737.15 utterances.], batch size: 47, lr: 4.66e-03, grad_scale: 16.0 2023-03-09 02:35:47,651 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90143.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 02:35:48,950 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.112e+02 2.045e+02 2.395e+02 2.937e+02 6.523e+02, threshold=4.790e+02, percent-clipped=3.0 2023-03-09 02:36:23,516 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90165.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:37:05,994 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90191.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:37:09,385 INFO [train2.py:809] (0/4) Epoch 23, batch 2550, loss[ctc_loss=0.05292, att_loss=0.2241, loss=0.1898, over 16117.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006864, over 42.00 utterances.], tot_loss[ctc_loss=0.07138, att_loss=0.2346, loss=0.2019, over 3277378.77 frames. utt_duration=1234 frames, utt_pad_proportion=0.05637, over 10637.05 utterances.], batch size: 42, lr: 4.66e-03, grad_scale: 16.0 2023-03-09 02:37:41,491 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90213.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:37:49,335 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-09 02:37:54,831 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90221.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 02:38:12,218 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90232.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:38:30,845 INFO [train2.py:809] (0/4) Epoch 23, batch 2600, loss[ctc_loss=0.07009, att_loss=0.2422, loss=0.2078, over 16878.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007082, over 49.00 utterances.], tot_loss[ctc_loss=0.07171, att_loss=0.235, loss=0.2023, over 3285815.94 frames. utt_duration=1238 frames, utt_pad_proportion=0.05343, over 10633.18 utterances.], batch size: 49, lr: 4.66e-03, grad_scale: 8.0 2023-03-09 02:38:34,186 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.085e+02 2.020e+02 2.371e+02 2.963e+02 6.277e+02, threshold=4.742e+02, percent-clipped=3.0 2023-03-09 02:38:41,846 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.29 vs. limit=2.0 2023-03-09 02:38:59,984 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1226, 4.2806, 4.2936, 4.6788, 2.5773, 4.5306, 2.8703, 1.5382], device='cuda:0'), covar=tensor([0.0460, 0.0301, 0.0675, 0.0193, 0.1766, 0.0187, 0.1329, 0.1861], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0173, 0.0266, 0.0164, 0.0227, 0.0156, 0.0235, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:39:32,875 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1330, 4.3263, 4.3995, 4.7335, 2.6946, 4.5291, 2.9573, 1.9178], device='cuda:0'), covar=tensor([0.0477, 0.0281, 0.0662, 0.0190, 0.1715, 0.0208, 0.1346, 0.1676], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0172, 0.0265, 0.0163, 0.0227, 0.0156, 0.0235, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:39:52,432 INFO [train2.py:809] (0/4) Epoch 23, batch 2650, loss[ctc_loss=0.05094, att_loss=0.2111, loss=0.1791, over 15897.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.00809, over 39.00 utterances.], tot_loss[ctc_loss=0.07102, att_loss=0.2345, loss=0.2018, over 3289445.76 frames. utt_duration=1268 frames, utt_pad_proportion=0.04576, over 10385.56 utterances.], batch size: 39, lr: 4.66e-03, grad_scale: 8.0 2023-03-09 02:39:52,893 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90293.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:39:59,149 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2579, 5.1216, 5.0068, 3.0135, 4.9690, 4.8492, 4.4878, 2.9548], device='cuda:0'), covar=tensor([0.0097, 0.0086, 0.0234, 0.0948, 0.0085, 0.0164, 0.0269, 0.1149], device='cuda:0'), in_proj_covar=tensor([0.0075, 0.0102, 0.0104, 0.0110, 0.0085, 0.0113, 0.0099, 0.0102], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 02:40:24,869 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4753, 3.1536, 3.2783, 4.5519, 4.0888, 4.0329, 3.0065, 2.4306], device='cuda:0'), covar=tensor([0.0719, 0.1907, 0.1041, 0.0627, 0.0800, 0.0452, 0.1557, 0.2223], device='cuda:0'), in_proj_covar=tensor([0.0184, 0.0219, 0.0191, 0.0224, 0.0228, 0.0182, 0.0205, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:40:37,351 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5720, 4.9609, 4.7917, 4.8861, 5.0643, 4.6911, 3.3957, 4.9315], device='cuda:0'), covar=tensor([0.0127, 0.0110, 0.0134, 0.0107, 0.0096, 0.0108, 0.0637, 0.0213], device='cuda:0'), in_proj_covar=tensor([0.0091, 0.0088, 0.0111, 0.0070, 0.0076, 0.0086, 0.0102, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 02:40:54,878 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4589, 4.5377, 4.8000, 4.8612, 3.1625, 4.7682, 3.2344, 2.3750], device='cuda:0'), covar=tensor([0.0373, 0.0267, 0.0554, 0.0206, 0.1446, 0.0178, 0.1165, 0.1541], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0171, 0.0263, 0.0163, 0.0225, 0.0154, 0.0233, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:41:03,332 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90336.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:41:14,427 INFO [train2.py:809] (0/4) Epoch 23, batch 2700, loss[ctc_loss=0.05765, att_loss=0.2142, loss=0.1829, over 15945.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.006813, over 41.00 utterances.], tot_loss[ctc_loss=0.07057, att_loss=0.234, loss=0.2013, over 3287114.40 frames. utt_duration=1277 frames, utt_pad_proportion=0.04423, over 10312.30 utterances.], batch size: 41, lr: 4.66e-03, grad_scale: 8.0 2023-03-09 02:41:17,513 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.381e+02 1.963e+02 2.237e+02 2.916e+02 6.361e+02, threshold=4.475e+02, percent-clipped=3.0 2023-03-09 02:41:26,997 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90351.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:41:35,734 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 02:42:20,705 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90384.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:42:35,957 INFO [train2.py:809] (0/4) Epoch 23, batch 2750, loss[ctc_loss=0.07462, att_loss=0.2347, loss=0.2027, over 16387.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.008296, over 44.00 utterances.], tot_loss[ctc_loss=0.06987, att_loss=0.2335, loss=0.2008, over 3282390.02 frames. utt_duration=1291 frames, utt_pad_proportion=0.04247, over 10184.63 utterances.], batch size: 44, lr: 4.66e-03, grad_scale: 8.0 2023-03-09 02:43:06,588 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90412.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:43:13,498 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.51 vs. limit=5.0 2023-03-09 02:43:58,168 INFO [train2.py:809] (0/4) Epoch 23, batch 2800, loss[ctc_loss=0.07045, att_loss=0.2482, loss=0.2127, over 17033.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.01016, over 52.00 utterances.], tot_loss[ctc_loss=0.0713, att_loss=0.2349, loss=0.2022, over 3283553.46 frames. utt_duration=1235 frames, utt_pad_proportion=0.05511, over 10648.74 utterances.], batch size: 52, lr: 4.66e-03, grad_scale: 8.0 2023-03-09 02:44:01,316 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 1.964e+02 2.366e+02 2.896e+02 7.050e+02, threshold=4.731e+02, percent-clipped=3.0 2023-03-09 02:44:23,912 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90459.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:44:40,256 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.40 vs. limit=2.0 2023-03-09 02:45:07,727 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90486.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:45:19,166 INFO [train2.py:809] (0/4) Epoch 23, batch 2850, loss[ctc_loss=0.06046, att_loss=0.2364, loss=0.2012, over 17031.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.00722, over 51.00 utterances.], tot_loss[ctc_loss=0.07125, att_loss=0.2352, loss=0.2024, over 3276542.86 frames. utt_duration=1229 frames, utt_pad_proportion=0.05872, over 10678.86 utterances.], batch size: 51, lr: 4.66e-03, grad_scale: 8.0 2023-03-09 02:45:24,405 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90496.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:46:03,418 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90520.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:46:04,883 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90521.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 02:46:40,205 INFO [train2.py:809] (0/4) Epoch 23, batch 2900, loss[ctc_loss=0.08286, att_loss=0.2529, loss=0.2189, over 17188.00 frames. utt_duration=871.8 frames, utt_pad_proportion=0.08715, over 79.00 utterances.], tot_loss[ctc_loss=0.07139, att_loss=0.2355, loss=0.2027, over 3280302.93 frames. utt_duration=1228 frames, utt_pad_proportion=0.05697, over 10697.53 utterances.], batch size: 79, lr: 4.65e-03, grad_scale: 8.0 2023-03-09 02:46:43,386 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 2.000e+02 2.282e+02 2.820e+02 8.053e+02, threshold=4.565e+02, percent-clipped=3.0 2023-03-09 02:46:50,677 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-09 02:46:53,196 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4365, 2.6577, 4.8865, 3.7813, 3.0001, 4.2550, 4.6184, 4.6228], device='cuda:0'), covar=tensor([0.0247, 0.1575, 0.0182, 0.0815, 0.1683, 0.0220, 0.0180, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0239, 0.0194, 0.0315, 0.0262, 0.0217, 0.0185, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:47:02,775 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90557.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:47:10,412 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8530, 3.7308, 3.0487, 3.2576, 3.8887, 3.6153, 2.7535, 4.0936], device='cuda:0'), covar=tensor([0.1206, 0.0593, 0.1197, 0.0801, 0.0830, 0.0737, 0.1028, 0.0575], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0221, 0.0231, 0.0206, 0.0285, 0.0246, 0.0202, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 02:47:21,110 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4359, 2.7278, 4.8789, 3.8151, 2.9956, 4.2143, 4.6532, 4.6207], device='cuda:0'), covar=tensor([0.0252, 0.1503, 0.0217, 0.0792, 0.1670, 0.0250, 0.0193, 0.0249], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0240, 0.0194, 0.0315, 0.0262, 0.0217, 0.0184, 0.0213], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:47:22,352 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90569.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:47:43,134 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2009, 5.2384, 5.0651, 2.3879, 2.1971, 2.9560, 2.4415, 4.0107], device='cuda:0'), covar=tensor([0.0719, 0.0266, 0.0205, 0.4705, 0.5153, 0.2442, 0.3480, 0.1567], device='cuda:0'), in_proj_covar=tensor([0.0355, 0.0279, 0.0268, 0.0242, 0.0339, 0.0332, 0.0255, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 02:47:50,349 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.46 vs. limit=5.0 2023-03-09 02:47:52,974 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90588.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:48:01,371 INFO [train2.py:809] (0/4) Epoch 23, batch 2950, loss[ctc_loss=0.05714, att_loss=0.2402, loss=0.2036, over 16632.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004931, over 47.00 utterances.], tot_loss[ctc_loss=0.07088, att_loss=0.2346, loss=0.2019, over 3272192.57 frames. utt_duration=1241 frames, utt_pad_proportion=0.05663, over 10556.18 utterances.], batch size: 47, lr: 4.65e-03, grad_scale: 8.0 2023-03-09 02:48:38,339 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9802, 5.2265, 5.2058, 5.1427, 5.2704, 5.1842, 4.8580, 4.6837], device='cuda:0'), covar=tensor([0.1030, 0.0608, 0.0314, 0.0521, 0.0306, 0.0364, 0.0440, 0.0371], device='cuda:0'), in_proj_covar=tensor([0.0530, 0.0369, 0.0357, 0.0365, 0.0433, 0.0439, 0.0366, 0.0403], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 02:49:21,807 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3254, 3.8896, 3.3497, 3.5682, 4.0806, 3.7993, 3.2035, 4.4708], device='cuda:0'), covar=tensor([0.0910, 0.0554, 0.1153, 0.0696, 0.0704, 0.0722, 0.0802, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0221, 0.0229, 0.0205, 0.0284, 0.0246, 0.0201, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 02:49:23,760 INFO [train2.py:809] (0/4) Epoch 23, batch 3000, loss[ctc_loss=0.05287, att_loss=0.2089, loss=0.1777, over 15621.00 frames. utt_duration=1690 frames, utt_pad_proportion=0.009911, over 37.00 utterances.], tot_loss[ctc_loss=0.07065, att_loss=0.2347, loss=0.2019, over 3280120.47 frames. utt_duration=1253 frames, utt_pad_proportion=0.05208, over 10486.59 utterances.], batch size: 37, lr: 4.65e-03, grad_scale: 8.0 2023-03-09 02:49:23,763 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 02:49:38,036 INFO [train2.py:843] (0/4) Epoch 23, validation: ctc_loss=0.03973, att_loss=0.234, loss=0.1952, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 02:49:38,037 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 02:49:41,307 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.311e+02 1.804e+02 2.183e+02 2.911e+02 8.359e+02, threshold=4.365e+02, percent-clipped=3.0 2023-03-09 02:51:00,738 INFO [train2.py:809] (0/4) Epoch 23, batch 3050, loss[ctc_loss=0.05274, att_loss=0.2362, loss=0.1995, over 16769.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006377, over 48.00 utterances.], tot_loss[ctc_loss=0.07011, att_loss=0.2344, loss=0.2016, over 3281907.85 frames. utt_duration=1242 frames, utt_pad_proportion=0.05261, over 10581.90 utterances.], batch size: 48, lr: 4.65e-03, grad_scale: 8.0 2023-03-09 02:51:24,183 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90707.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:51:30,639 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9768, 3.9964, 4.0898, 4.1304, 4.1717, 4.1489, 3.9245, 3.8538], device='cuda:0'), covar=tensor([0.0918, 0.0863, 0.0729, 0.0456, 0.0321, 0.0369, 0.0419, 0.0324], device='cuda:0'), in_proj_covar=tensor([0.0525, 0.0365, 0.0355, 0.0361, 0.0429, 0.0435, 0.0364, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-09 02:52:23,333 INFO [train2.py:809] (0/4) Epoch 23, batch 3100, loss[ctc_loss=0.06405, att_loss=0.2247, loss=0.1926, over 16186.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006853, over 41.00 utterances.], tot_loss[ctc_loss=0.0705, att_loss=0.2348, loss=0.2019, over 3285395.77 frames. utt_duration=1206 frames, utt_pad_proportion=0.05957, over 10909.94 utterances.], batch size: 41, lr: 4.65e-03, grad_scale: 4.0 2023-03-09 02:52:28,636 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.220e+02 1.776e+02 2.095e+02 2.519e+02 6.521e+02, threshold=4.191e+02, percent-clipped=1.0 2023-03-09 02:53:33,749 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90786.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:53:45,911 INFO [train2.py:809] (0/4) Epoch 23, batch 3150, loss[ctc_loss=0.05492, att_loss=0.22, loss=0.187, over 15978.00 frames. utt_duration=1560 frames, utt_pad_proportion=0.004838, over 41.00 utterances.], tot_loss[ctc_loss=0.07101, att_loss=0.2347, loss=0.202, over 3282491.99 frames. utt_duration=1215 frames, utt_pad_proportion=0.05899, over 10815.62 utterances.], batch size: 41, lr: 4.65e-03, grad_scale: 4.0 2023-03-09 02:54:21,931 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90815.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:54:43,386 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5710, 2.6655, 5.0393, 3.7969, 3.0698, 4.3889, 4.8587, 4.6663], device='cuda:0'), covar=tensor([0.0296, 0.1531, 0.0210, 0.1205, 0.1808, 0.0228, 0.0182, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0245, 0.0198, 0.0320, 0.0266, 0.0221, 0.0188, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:54:52,438 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90834.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:55:08,167 INFO [train2.py:809] (0/4) Epoch 23, batch 3200, loss[ctc_loss=0.05548, att_loss=0.2322, loss=0.1968, over 17013.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008195, over 51.00 utterances.], tot_loss[ctc_loss=0.07005, att_loss=0.2341, loss=0.2013, over 3279064.52 frames. utt_duration=1225 frames, utt_pad_proportion=0.05871, over 10719.29 utterances.], batch size: 51, lr: 4.65e-03, grad_scale: 8.0 2023-03-09 02:55:12,916 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 1.815e+02 2.300e+02 2.792e+02 4.062e+02, threshold=4.600e+02, percent-clipped=0.0 2023-03-09 02:55:18,198 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=90849.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:55:22,755 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=90852.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:56:21,728 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=90888.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:56:29,989 INFO [train2.py:809] (0/4) Epoch 23, batch 3250, loss[ctc_loss=0.07269, att_loss=0.2121, loss=0.1843, over 15370.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01053, over 35.00 utterances.], tot_loss[ctc_loss=0.06993, att_loss=0.234, loss=0.2012, over 3281406.21 frames. utt_duration=1247 frames, utt_pad_proportion=0.05328, over 10537.55 utterances.], batch size: 35, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 02:56:54,644 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-03-09 02:56:57,361 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=90910.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:57:39,143 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=90936.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:57:51,515 INFO [train2.py:809] (0/4) Epoch 23, batch 3300, loss[ctc_loss=0.06283, att_loss=0.2134, loss=0.1833, over 11858.00 frames. utt_duration=1826 frames, utt_pad_proportion=0.169, over 26.00 utterances.], tot_loss[ctc_loss=0.07033, att_loss=0.2343, loss=0.2015, over 3274820.59 frames. utt_duration=1231 frames, utt_pad_proportion=0.05727, over 10651.14 utterances.], batch size: 26, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 02:57:56,191 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 1.978e+02 2.311e+02 2.914e+02 8.897e+02, threshold=4.622e+02, percent-clipped=4.0 2023-03-09 02:58:19,290 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-03-09 02:58:34,070 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2798, 4.3546, 4.4101, 4.4547, 4.8870, 4.5117, 4.3650, 2.4365], device='cuda:0'), covar=tensor([0.0282, 0.0419, 0.0356, 0.0321, 0.0972, 0.0225, 0.0404, 0.1776], device='cuda:0'), in_proj_covar=tensor([0.0172, 0.0198, 0.0196, 0.0213, 0.0371, 0.0166, 0.0186, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 02:59:13,555 INFO [train2.py:809] (0/4) Epoch 23, batch 3350, loss[ctc_loss=0.08242, att_loss=0.2486, loss=0.2154, over 16884.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.007282, over 49.00 utterances.], tot_loss[ctc_loss=0.07031, att_loss=0.2345, loss=0.2017, over 3284873.81 frames. utt_duration=1248 frames, utt_pad_proportion=0.05023, over 10539.12 utterances.], batch size: 49, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 02:59:37,332 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91007.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 02:59:49,200 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9854, 4.2279, 4.3786, 4.5241, 2.7878, 4.3414, 2.5776, 1.7581], device='cuda:0'), covar=tensor([0.0441, 0.0272, 0.0607, 0.0228, 0.1651, 0.0234, 0.1568, 0.1843], device='cuda:0'), in_proj_covar=tensor([0.0197, 0.0170, 0.0260, 0.0162, 0.0222, 0.0153, 0.0230, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:00:35,857 INFO [train2.py:809] (0/4) Epoch 23, batch 3400, loss[ctc_loss=0.06621, att_loss=0.2393, loss=0.2047, over 16763.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.005997, over 48.00 utterances.], tot_loss[ctc_loss=0.07033, att_loss=0.235, loss=0.2021, over 3288089.10 frames. utt_duration=1230 frames, utt_pad_proportion=0.0538, over 10707.55 utterances.], batch size: 48, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 03:00:40,395 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 1.937e+02 2.293e+02 2.658e+02 6.821e+02, threshold=4.585e+02, percent-clipped=3.0 2023-03-09 03:00:55,178 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91055.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:01:09,434 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 03:01:30,611 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-03-09 03:01:57,224 INFO [train2.py:809] (0/4) Epoch 23, batch 3450, loss[ctc_loss=0.08038, att_loss=0.2371, loss=0.2058, over 16541.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.00542, over 45.00 utterances.], tot_loss[ctc_loss=0.0704, att_loss=0.235, loss=0.202, over 3291067.99 frames. utt_duration=1256 frames, utt_pad_proportion=0.04711, over 10496.98 utterances.], batch size: 45, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 03:02:33,669 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91115.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:03:19,291 INFO [train2.py:809] (0/4) Epoch 23, batch 3500, loss[ctc_loss=0.07347, att_loss=0.2453, loss=0.2109, over 17281.00 frames. utt_duration=1099 frames, utt_pad_proportion=0.03886, over 63.00 utterances.], tot_loss[ctc_loss=0.07043, att_loss=0.2352, loss=0.2022, over 3284067.92 frames. utt_duration=1256 frames, utt_pad_proportion=0.0494, over 10471.84 utterances.], batch size: 63, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 03:03:23,956 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 1.873e+02 2.204e+02 2.861e+02 4.686e+02, threshold=4.409e+02, percent-clipped=2.0 2023-03-09 03:03:33,673 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91152.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:03:51,229 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91163.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:03:51,388 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9255, 5.3370, 4.9020, 5.3610, 4.7961, 5.0894, 5.4293, 5.2220], device='cuda:0'), covar=tensor([0.0626, 0.0299, 0.0757, 0.0353, 0.0385, 0.0230, 0.0220, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0393, 0.0326, 0.0366, 0.0358, 0.0325, 0.0240, 0.0309, 0.0290], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 03:03:51,489 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5367, 3.0212, 3.6684, 3.2211, 3.6158, 4.6298, 4.4088, 3.3143], device='cuda:0'), covar=tensor([0.0405, 0.1684, 0.1183, 0.1180, 0.1117, 0.0836, 0.0619, 0.1259], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0246, 0.0284, 0.0223, 0.0271, 0.0376, 0.0265, 0.0236], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 03:04:05,640 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-09 03:04:40,874 INFO [train2.py:809] (0/4) Epoch 23, batch 3550, loss[ctc_loss=0.07663, att_loss=0.2362, loss=0.2043, over 16880.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006832, over 49.00 utterances.], tot_loss[ctc_loss=0.0704, att_loss=0.2349, loss=0.202, over 3283263.25 frames. utt_duration=1251 frames, utt_pad_proportion=0.05116, over 10509.89 utterances.], batch size: 49, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 03:04:52,820 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91200.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:05:01,525 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91205.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:06:04,019 INFO [train2.py:809] (0/4) Epoch 23, batch 3600, loss[ctc_loss=0.09513, att_loss=0.2505, loss=0.2194, over 16627.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.00517, over 47.00 utterances.], tot_loss[ctc_loss=0.07083, att_loss=0.2355, loss=0.2026, over 3289010.76 frames. utt_duration=1231 frames, utt_pad_proportion=0.05394, over 10697.16 utterances.], batch size: 47, lr: 4.64e-03, grad_scale: 8.0 2023-03-09 03:06:08,712 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.503e+02 1.921e+02 2.289e+02 2.747e+02 1.018e+03, threshold=4.578e+02, percent-clipped=6.0 2023-03-09 03:07:14,054 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91285.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:07:26,453 INFO [train2.py:809] (0/4) Epoch 23, batch 3650, loss[ctc_loss=0.06133, att_loss=0.2416, loss=0.2055, over 16877.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.00717, over 49.00 utterances.], tot_loss[ctc_loss=0.07088, att_loss=0.2352, loss=0.2023, over 3277363.93 frames. utt_duration=1202 frames, utt_pad_proportion=0.06338, over 10920.89 utterances.], batch size: 49, lr: 4.63e-03, grad_scale: 8.0 2023-03-09 03:08:21,529 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-09 03:08:47,987 INFO [train2.py:809] (0/4) Epoch 23, batch 3700, loss[ctc_loss=0.04494, att_loss=0.2037, loss=0.1719, over 15632.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009009, over 37.00 utterances.], tot_loss[ctc_loss=0.07125, att_loss=0.2352, loss=0.2024, over 3275191.03 frames. utt_duration=1187 frames, utt_pad_proportion=0.06692, over 11047.97 utterances.], batch size: 37, lr: 4.63e-03, grad_scale: 4.0 2023-03-09 03:08:53,192 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91346.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:08:54,349 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.187e+02 1.976e+02 2.254e+02 2.803e+02 7.124e+02, threshold=4.508e+02, percent-clipped=5.0 2023-03-09 03:09:36,138 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-09 03:10:10,208 INFO [train2.py:809] (0/4) Epoch 23, batch 3750, loss[ctc_loss=0.07468, att_loss=0.236, loss=0.2037, over 16971.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.006491, over 50.00 utterances.], tot_loss[ctc_loss=0.07148, att_loss=0.235, loss=0.2023, over 3268911.78 frames. utt_duration=1183 frames, utt_pad_proportion=0.07083, over 11068.77 utterances.], batch size: 50, lr: 4.63e-03, grad_scale: 4.0 2023-03-09 03:10:26,657 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9246, 5.3003, 5.3491, 5.3142, 5.3240, 5.3567, 5.0121, 4.8217], device='cuda:0'), covar=tensor([0.1338, 0.0635, 0.0319, 0.0479, 0.0468, 0.0396, 0.0466, 0.0403], device='cuda:0'), in_proj_covar=tensor([0.0524, 0.0364, 0.0353, 0.0360, 0.0427, 0.0435, 0.0362, 0.0399], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-09 03:11:04,097 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-09 03:11:32,632 INFO [train2.py:809] (0/4) Epoch 23, batch 3800, loss[ctc_loss=0.08122, att_loss=0.2133, loss=0.1869, over 15503.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008466, over 36.00 utterances.], tot_loss[ctc_loss=0.07118, att_loss=0.2342, loss=0.2016, over 3270676.91 frames. utt_duration=1205 frames, utt_pad_proportion=0.06508, over 10869.12 utterances.], batch size: 36, lr: 4.63e-03, grad_scale: 4.0 2023-03-09 03:11:38,947 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 1.926e+02 2.458e+02 3.092e+02 6.476e+02, threshold=4.915e+02, percent-clipped=5.0 2023-03-09 03:11:48,863 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7158, 3.3355, 3.3436, 2.8031, 3.2914, 3.4024, 3.3425, 2.3034], device='cuda:0'), covar=tensor([0.1156, 0.1447, 0.2056, 0.4302, 0.1580, 0.1829, 0.1298, 0.4702], device='cuda:0'), in_proj_covar=tensor([0.0187, 0.0194, 0.0207, 0.0261, 0.0168, 0.0271, 0.0193, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:12:41,281 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7747, 3.2396, 3.8252, 3.3305, 3.7787, 4.7762, 4.5964, 3.4624], device='cuda:0'), covar=tensor([0.0317, 0.1553, 0.1158, 0.1195, 0.0998, 0.0870, 0.0616, 0.1126], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0243, 0.0281, 0.0219, 0.0268, 0.0370, 0.0263, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 03:12:55,888 INFO [train2.py:809] (0/4) Epoch 23, batch 3850, loss[ctc_loss=0.0775, att_loss=0.242, loss=0.2091, over 16962.00 frames. utt_duration=687 frames, utt_pad_proportion=0.138, over 99.00 utterances.], tot_loss[ctc_loss=0.07066, att_loss=0.2337, loss=0.2011, over 3266943.16 frames. utt_duration=1204 frames, utt_pad_proportion=0.06609, over 10862.87 utterances.], batch size: 99, lr: 4.63e-03, grad_scale: 4.0 2023-03-09 03:13:15,002 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91505.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:14:12,257 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91542.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 03:14:13,404 INFO [train2.py:809] (0/4) Epoch 23, batch 3900, loss[ctc_loss=0.08881, att_loss=0.2557, loss=0.2223, over 16979.00 frames. utt_duration=1360 frames, utt_pad_proportion=0.006822, over 50.00 utterances.], tot_loss[ctc_loss=0.0708, att_loss=0.2337, loss=0.2011, over 3259479.29 frames. utt_duration=1235 frames, utt_pad_proportion=0.06001, over 10574.07 utterances.], batch size: 50, lr: 4.63e-03, grad_scale: 4.0 2023-03-09 03:14:19,548 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 1.985e+02 2.304e+02 2.861e+02 4.797e+02, threshold=4.607e+02, percent-clipped=0.0 2023-03-09 03:14:29,563 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91553.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:14:43,056 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-09 03:14:46,754 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6647, 5.0848, 4.9059, 5.0814, 5.0778, 4.7452, 3.6499, 5.0488], device='cuda:0'), covar=tensor([0.0119, 0.0115, 0.0177, 0.0071, 0.0142, 0.0115, 0.0638, 0.0168], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0090, 0.0112, 0.0070, 0.0076, 0.0087, 0.0103, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 03:14:55,533 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-09 03:15:32,014 INFO [train2.py:809] (0/4) Epoch 23, batch 3950, loss[ctc_loss=0.06023, att_loss=0.211, loss=0.1808, over 15996.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007946, over 40.00 utterances.], tot_loss[ctc_loss=0.0709, att_loss=0.2336, loss=0.2011, over 3253111.64 frames. utt_duration=1249 frames, utt_pad_proportion=0.05884, over 10431.31 utterances.], batch size: 40, lr: 4.63e-03, grad_scale: 4.0 2023-03-09 03:15:48,503 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91603.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 03:16:25,037 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-23.pt 2023-03-09 03:16:52,902 INFO [train2.py:809] (0/4) Epoch 24, batch 0, loss[ctc_loss=0.06081, att_loss=0.2094, loss=0.1797, over 14457.00 frames. utt_duration=1809 frames, utt_pad_proportion=0.03687, over 32.00 utterances.], tot_loss[ctc_loss=0.06081, att_loss=0.2094, loss=0.1797, over 14457.00 frames. utt_duration=1809 frames, utt_pad_proportion=0.03687, over 32.00 utterances.], batch size: 32, lr: 4.53e-03, grad_scale: 8.0 2023-03-09 03:16:52,904 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 03:17:00,606 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6618, 4.1545, 4.2070, 2.0396, 1.9678, 2.6873, 2.1696, 3.5223], device='cuda:0'), covar=tensor([0.0762, 0.0388, 0.0342, 0.4863, 0.5277, 0.2538, 0.3551, 0.1454], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0282, 0.0271, 0.0245, 0.0341, 0.0334, 0.0257, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 03:17:05,994 INFO [train2.py:843] (0/4) Epoch 24, validation: ctc_loss=0.04095, att_loss=0.2349, loss=0.1961, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 03:17:05,995 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 03:17:28,526 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91641.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:17:37,666 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.284e+02 2.024e+02 2.562e+02 3.130e+02 9.930e+02, threshold=5.124e+02, percent-clipped=6.0 2023-03-09 03:18:15,967 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 03:18:27,185 INFO [train2.py:809] (0/4) Epoch 24, batch 50, loss[ctc_loss=0.06828, att_loss=0.2364, loss=0.2028, over 17068.00 frames. utt_duration=1315 frames, utt_pad_proportion=0.007808, over 52.00 utterances.], tot_loss[ctc_loss=0.0675, att_loss=0.2323, loss=0.1994, over 739871.42 frames. utt_duration=1267 frames, utt_pad_proportion=0.04844, over 2339.42 utterances.], batch size: 52, lr: 4.53e-03, grad_scale: 8.0 2023-03-09 03:19:47,844 INFO [train2.py:809] (0/4) Epoch 24, batch 100, loss[ctc_loss=0.08627, att_loss=0.2571, loss=0.2229, over 16770.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006555, over 48.00 utterances.], tot_loss[ctc_loss=0.07126, att_loss=0.2359, loss=0.203, over 1299757.09 frames. utt_duration=1204 frames, utt_pad_proportion=0.0602, over 4324.07 utterances.], batch size: 48, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:20:20,176 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 1.937e+02 2.364e+02 2.777e+02 5.363e+02, threshold=4.728e+02, percent-clipped=1.0 2023-03-09 03:20:53,285 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.40 vs. limit=5.0 2023-03-09 03:21:08,901 INFO [train2.py:809] (0/4) Epoch 24, batch 150, loss[ctc_loss=0.04821, att_loss=0.1994, loss=0.1691, over 15996.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008008, over 40.00 utterances.], tot_loss[ctc_loss=0.06963, att_loss=0.2333, loss=0.2005, over 1736188.63 frames. utt_duration=1252 frames, utt_pad_proportion=0.04942, over 5555.44 utterances.], batch size: 40, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:22:30,996 INFO [train2.py:809] (0/4) Epoch 24, batch 200, loss[ctc_loss=0.05827, att_loss=0.2291, loss=0.1949, over 16895.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.006651, over 49.00 utterances.], tot_loss[ctc_loss=0.06888, att_loss=0.2329, loss=0.2001, over 2079631.75 frames. utt_duration=1269 frames, utt_pad_proportion=0.04589, over 6561.68 utterances.], batch size: 49, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:23:02,671 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.186e+02 1.924e+02 2.312e+02 2.679e+02 4.327e+02, threshold=4.624e+02, percent-clipped=0.0 2023-03-09 03:23:51,118 INFO [train2.py:809] (0/4) Epoch 24, batch 250, loss[ctc_loss=0.07214, att_loss=0.2313, loss=0.1995, over 16918.00 frames. utt_duration=685.1 frames, utt_pad_proportion=0.136, over 99.00 utterances.], tot_loss[ctc_loss=0.06881, att_loss=0.2331, loss=0.2003, over 2350837.13 frames. utt_duration=1291 frames, utt_pad_proportion=0.03911, over 7291.04 utterances.], batch size: 99, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:24:06,719 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4540, 2.7195, 4.9776, 3.8222, 3.0275, 4.2464, 4.7984, 4.5350], device='cuda:0'), covar=tensor([0.0315, 0.1470, 0.0192, 0.1162, 0.1717, 0.0261, 0.0210, 0.0341], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0241, 0.0196, 0.0317, 0.0263, 0.0220, 0.0188, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:24:14,501 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91891.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:24:25,435 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=91898.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 03:25:06,739 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.83 vs. limit=5.0 2023-03-09 03:25:11,835 INFO [train2.py:809] (0/4) Epoch 24, batch 300, loss[ctc_loss=0.08724, att_loss=0.2553, loss=0.2217, over 17342.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03193, over 63.00 utterances.], tot_loss[ctc_loss=0.06975, att_loss=0.2333, loss=0.2006, over 2553858.22 frames. utt_duration=1251 frames, utt_pad_proportion=0.04999, over 8174.73 utterances.], batch size: 63, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:25:35,428 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=91941.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:25:44,284 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.141e+02 1.878e+02 2.164e+02 2.654e+02 4.888e+02, threshold=4.328e+02, percent-clipped=1.0 2023-03-09 03:25:53,277 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=91952.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:25:57,835 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91955.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:26:32,685 INFO [train2.py:809] (0/4) Epoch 24, batch 350, loss[ctc_loss=0.064, att_loss=0.2453, loss=0.209, over 17324.00 frames. utt_duration=1101 frames, utt_pad_proportion=0.03723, over 63.00 utterances.], tot_loss[ctc_loss=0.07015, att_loss=0.2338, loss=0.201, over 2716902.25 frames. utt_duration=1255 frames, utt_pad_proportion=0.04966, over 8671.74 utterances.], batch size: 63, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:26:39,967 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=91981.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 03:26:52,901 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=91989.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:27:10,408 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-92000.pt 2023-03-09 03:27:40,047 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92016.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:27:58,111 INFO [train2.py:809] (0/4) Epoch 24, batch 400, loss[ctc_loss=0.09067, att_loss=0.2552, loss=0.2223, over 16863.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.008589, over 49.00 utterances.], tot_loss[ctc_loss=0.07018, att_loss=0.2347, loss=0.2018, over 2842451.34 frames. utt_duration=1236 frames, utt_pad_proportion=0.0541, over 9206.79 utterances.], batch size: 49, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:28:16,338 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2248, 5.5012, 5.4269, 5.4435, 5.5289, 5.5013, 5.1766, 4.9623], device='cuda:0'), covar=tensor([0.1023, 0.0494, 0.0269, 0.0446, 0.0278, 0.0303, 0.0368, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0527, 0.0366, 0.0354, 0.0361, 0.0428, 0.0438, 0.0362, 0.0402], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-09 03:28:22,879 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92042.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 03:28:30,203 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.202e+02 1.943e+02 2.339e+02 2.970e+02 6.262e+02, threshold=4.678e+02, percent-clipped=8.0 2023-03-09 03:29:04,443 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7372, 5.9823, 5.4576, 5.7576, 5.6748, 5.1827, 5.4031, 5.1336], device='cuda:0'), covar=tensor([0.1252, 0.0861, 0.0966, 0.0777, 0.0909, 0.1524, 0.2284, 0.2392], device='cuda:0'), in_proj_covar=tensor([0.0533, 0.0621, 0.0471, 0.0464, 0.0438, 0.0473, 0.0622, 0.0538], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 03:29:18,998 INFO [train2.py:809] (0/4) Epoch 24, batch 450, loss[ctc_loss=0.07853, att_loss=0.2481, loss=0.2142, over 17408.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03251, over 63.00 utterances.], tot_loss[ctc_loss=0.07064, att_loss=0.235, loss=0.2021, over 2933071.15 frames. utt_duration=1225 frames, utt_pad_proportion=0.05957, over 9589.05 utterances.], batch size: 63, lr: 4.52e-03, grad_scale: 8.0 2023-03-09 03:29:41,032 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8512, 2.2519, 2.5892, 2.6864, 2.8589, 2.4300, 2.5532, 3.0487], device='cuda:0'), covar=tensor([0.1388, 0.2532, 0.1590, 0.1254, 0.1168, 0.1492, 0.1784, 0.0993], device='cuda:0'), in_proj_covar=tensor([0.0126, 0.0129, 0.0125, 0.0116, 0.0132, 0.0115, 0.0139, 0.0110], device='cuda:0'), out_proj_covar=tensor([9.5961e-05, 1.0098e-04, 1.0121e-04, 9.1421e-05, 9.9260e-05, 9.1995e-05, 1.0485e-04, 8.7535e-05], device='cuda:0') 2023-03-09 03:30:40,859 INFO [train2.py:809] (0/4) Epoch 24, batch 500, loss[ctc_loss=0.07019, att_loss=0.1993, loss=0.1735, over 15873.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009923, over 39.00 utterances.], tot_loss[ctc_loss=0.07055, att_loss=0.2346, loss=0.2018, over 3012169.64 frames. utt_duration=1240 frames, utt_pad_proportion=0.05561, over 9728.82 utterances.], batch size: 39, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:31:13,738 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.285e+02 1.745e+02 2.110e+02 2.591e+02 7.932e+02, threshold=4.219e+02, percent-clipped=4.0 2023-03-09 03:31:15,870 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8376, 4.4813, 4.4298, 2.4307, 2.1894, 2.9706, 2.4289, 3.6949], device='cuda:0'), covar=tensor([0.0749, 0.0315, 0.0261, 0.3843, 0.4537, 0.2124, 0.3053, 0.1469], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0281, 0.0269, 0.0244, 0.0339, 0.0333, 0.0258, 0.0368], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 03:31:49,063 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9981, 5.3422, 4.9188, 5.4027, 4.7181, 5.0440, 5.4582, 5.2570], device='cuda:0'), covar=tensor([0.0601, 0.0316, 0.0718, 0.0344, 0.0429, 0.0254, 0.0218, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0389, 0.0322, 0.0361, 0.0354, 0.0325, 0.0237, 0.0305, 0.0286], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0005, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 03:32:01,720 INFO [train2.py:809] (0/4) Epoch 24, batch 550, loss[ctc_loss=0.06682, att_loss=0.2338, loss=0.2004, over 16560.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005072, over 45.00 utterances.], tot_loss[ctc_loss=0.0701, att_loss=0.2341, loss=0.2013, over 3071004.72 frames. utt_duration=1237 frames, utt_pad_proportion=0.05566, over 9941.32 utterances.], batch size: 45, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:32:35,896 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92198.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 03:33:22,323 INFO [train2.py:809] (0/4) Epoch 24, batch 600, loss[ctc_loss=0.06421, att_loss=0.216, loss=0.1857, over 16272.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007102, over 43.00 utterances.], tot_loss[ctc_loss=0.07027, att_loss=0.2346, loss=0.2017, over 3115632.81 frames. utt_duration=1230 frames, utt_pad_proportion=0.05877, over 10145.31 utterances.], batch size: 43, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:33:53,590 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92246.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 03:33:54,962 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.139e+02 1.830e+02 2.241e+02 2.667e+02 5.450e+02, threshold=4.482e+02, percent-clipped=6.0 2023-03-09 03:33:55,216 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92247.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:34:34,261 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-09 03:34:42,408 INFO [train2.py:809] (0/4) Epoch 24, batch 650, loss[ctc_loss=0.06507, att_loss=0.231, loss=0.1978, over 16949.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008342, over 50.00 utterances.], tot_loss[ctc_loss=0.07014, att_loss=0.2344, loss=0.2016, over 3160070.81 frames. utt_duration=1253 frames, utt_pad_proportion=0.05067, over 10103.45 utterances.], batch size: 50, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:34:48,282 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5607, 3.0357, 3.0201, 2.6775, 3.0084, 2.9751, 3.0795, 2.4121], device='cuda:0'), covar=tensor([0.1179, 0.1550, 0.2392, 0.3731, 0.1310, 0.1853, 0.1318, 0.3555], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0198, 0.0213, 0.0266, 0.0171, 0.0276, 0.0197, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:35:37,895 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92311.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:36:03,497 INFO [train2.py:809] (0/4) Epoch 24, batch 700, loss[ctc_loss=0.0583, att_loss=0.2084, loss=0.1784, over 15496.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.009105, over 36.00 utterances.], tot_loss[ctc_loss=0.06892, att_loss=0.2335, loss=0.2006, over 3186014.31 frames. utt_duration=1257 frames, utt_pad_proportion=0.04991, over 10153.30 utterances.], batch size: 36, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:36:20,211 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92337.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 03:36:32,071 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9010, 5.2015, 5.4343, 5.2684, 5.4008, 5.8820, 5.1463, 5.9643], device='cuda:0'), covar=tensor([0.0772, 0.0763, 0.0847, 0.1550, 0.1860, 0.0921, 0.0740, 0.0696], device='cuda:0'), in_proj_covar=tensor([0.0881, 0.0512, 0.0616, 0.0665, 0.0879, 0.0636, 0.0497, 0.0621], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 03:36:36,600 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 1.935e+02 2.344e+02 2.941e+02 5.112e+02, threshold=4.689e+02, percent-clipped=2.0 2023-03-09 03:36:42,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2023-03-09 03:37:25,731 INFO [train2.py:809] (0/4) Epoch 24, batch 750, loss[ctc_loss=0.06476, att_loss=0.2421, loss=0.2067, over 17025.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007657, over 51.00 utterances.], tot_loss[ctc_loss=0.06842, att_loss=0.2328, loss=0.1999, over 3201267.17 frames. utt_duration=1268 frames, utt_pad_proportion=0.04951, over 10111.68 utterances.], batch size: 51, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:38:04,648 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3830, 4.4266, 4.4375, 4.5047, 4.9699, 4.4441, 4.4438, 2.5838], device='cuda:0'), covar=tensor([0.0259, 0.0360, 0.0363, 0.0269, 0.0871, 0.0250, 0.0319, 0.1706], device='cuda:0'), in_proj_covar=tensor([0.0174, 0.0200, 0.0200, 0.0215, 0.0375, 0.0169, 0.0189, 0.0215], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:38:46,217 INFO [train2.py:809] (0/4) Epoch 24, batch 800, loss[ctc_loss=0.07438, att_loss=0.2468, loss=0.2123, over 16459.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.00702, over 46.00 utterances.], tot_loss[ctc_loss=0.06936, att_loss=0.2337, loss=0.2008, over 3221895.82 frames. utt_duration=1272 frames, utt_pad_proportion=0.04729, over 10141.10 utterances.], batch size: 46, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:39:19,379 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.406e+02 1.923e+02 2.416e+02 3.239e+02 1.370e+03, threshold=4.832e+02, percent-clipped=5.0 2023-03-09 03:40:02,749 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-09 03:40:07,917 INFO [train2.py:809] (0/4) Epoch 24, batch 850, loss[ctc_loss=0.07301, att_loss=0.2179, loss=0.1889, over 15967.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.005646, over 41.00 utterances.], tot_loss[ctc_loss=0.06894, att_loss=0.2332, loss=0.2004, over 3230767.97 frames. utt_duration=1266 frames, utt_pad_proportion=0.05055, over 10216.84 utterances.], batch size: 41, lr: 4.51e-03, grad_scale: 8.0 2023-03-09 03:41:29,528 INFO [train2.py:809] (0/4) Epoch 24, batch 900, loss[ctc_loss=0.05567, att_loss=0.211, loss=0.1799, over 15364.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.009583, over 35.00 utterances.], tot_loss[ctc_loss=0.06859, att_loss=0.2327, loss=0.1998, over 3237847.50 frames. utt_duration=1280 frames, utt_pad_proportion=0.04785, over 10129.82 utterances.], batch size: 35, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:42:02,358 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.067e+02 1.834e+02 2.155e+02 2.748e+02 5.069e+02, threshold=4.309e+02, percent-clipped=1.0 2023-03-09 03:42:02,704 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92547.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:42:16,017 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0566, 5.3464, 5.6225, 5.3929, 5.5472, 6.0700, 5.2671, 6.0769], device='cuda:0'), covar=tensor([0.0747, 0.0729, 0.0833, 0.1471, 0.1995, 0.0848, 0.0704, 0.0706], device='cuda:0'), in_proj_covar=tensor([0.0887, 0.0516, 0.0621, 0.0668, 0.0888, 0.0641, 0.0505, 0.0625], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 03:42:50,356 INFO [train2.py:809] (0/4) Epoch 24, batch 950, loss[ctc_loss=0.07942, att_loss=0.255, loss=0.2199, over 17064.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008374, over 52.00 utterances.], tot_loss[ctc_loss=0.06979, att_loss=0.234, loss=0.2011, over 3258312.99 frames. utt_duration=1272 frames, utt_pad_proportion=0.04593, over 10255.47 utterances.], batch size: 52, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:43:20,148 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92595.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:43:46,304 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92611.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:44:08,911 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92625.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:44:11,647 INFO [train2.py:809] (0/4) Epoch 24, batch 1000, loss[ctc_loss=0.06114, att_loss=0.2269, loss=0.1938, over 16179.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006228, over 41.00 utterances.], tot_loss[ctc_loss=0.06955, att_loss=0.2339, loss=0.201, over 3265621.89 frames. utt_duration=1281 frames, utt_pad_proportion=0.04364, over 10208.04 utterances.], batch size: 41, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:44:21,815 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2109, 5.1875, 4.8116, 3.1762, 5.0748, 4.8827, 4.4078, 2.6399], device='cuda:0'), covar=tensor([0.0127, 0.0114, 0.0381, 0.1084, 0.0108, 0.0228, 0.0367, 0.1620], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0103, 0.0106, 0.0112, 0.0087, 0.0115, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 03:44:28,679 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=92637.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 03:44:44,465 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.218e+02 1.788e+02 2.232e+02 2.616e+02 5.460e+02, threshold=4.464e+02, percent-clipped=2.0 2023-03-09 03:45:04,282 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92659.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:45:33,250 INFO [train2.py:809] (0/4) Epoch 24, batch 1050, loss[ctc_loss=0.1, att_loss=0.2565, loss=0.2252, over 14289.00 frames. utt_duration=393 frames, utt_pad_proportion=0.3166, over 146.00 utterances.], tot_loss[ctc_loss=0.06995, att_loss=0.2344, loss=0.2015, over 3267655.46 frames. utt_duration=1258 frames, utt_pad_proportion=0.04931, over 10406.12 utterances.], batch size: 146, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:45:47,549 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=92685.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 03:45:49,245 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92686.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:46:21,215 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.83 vs. limit=5.0 2023-03-09 03:46:43,747 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92720.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:46:54,334 INFO [train2.py:809] (0/4) Epoch 24, batch 1100, loss[ctc_loss=0.07684, att_loss=0.2517, loss=0.2167, over 17104.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01592, over 56.00 utterances.], tot_loss[ctc_loss=0.06912, att_loss=0.2336, loss=0.2007, over 3262036.96 frames. utt_duration=1271 frames, utt_pad_proportion=0.04815, over 10275.88 utterances.], batch size: 56, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:47:27,345 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.412e+02 1.778e+02 2.254e+02 2.467e+02 8.926e+02, threshold=4.507e+02, percent-clipped=3.0 2023-03-09 03:48:16,146 INFO [train2.py:809] (0/4) Epoch 24, batch 1150, loss[ctc_loss=0.07436, att_loss=0.2476, loss=0.2129, over 16479.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005815, over 46.00 utterances.], tot_loss[ctc_loss=0.06893, att_loss=0.2333, loss=0.2004, over 3264540.75 frames. utt_duration=1268 frames, utt_pad_proportion=0.04844, over 10311.01 utterances.], batch size: 46, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:48:23,644 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92781.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:49:34,640 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5310, 2.6057, 4.9090, 3.9692, 2.9762, 4.3183, 4.8033, 4.6736], device='cuda:0'), covar=tensor([0.0277, 0.1594, 0.0185, 0.0858, 0.1696, 0.0254, 0.0154, 0.0258], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0245, 0.0202, 0.0321, 0.0266, 0.0224, 0.0192, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:49:37,247 INFO [train2.py:809] (0/4) Epoch 24, batch 1200, loss[ctc_loss=0.0619, att_loss=0.2367, loss=0.2017, over 17314.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01063, over 55.00 utterances.], tot_loss[ctc_loss=0.06888, att_loss=0.2336, loss=0.2007, over 3274689.63 frames. utt_duration=1245 frames, utt_pad_proportion=0.05264, over 10530.59 utterances.], batch size: 55, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:49:44,640 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92831.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:50:10,381 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.270e+02 1.947e+02 2.306e+02 3.014e+02 1.019e+03, threshold=4.613e+02, percent-clipped=6.0 2023-03-09 03:50:59,083 INFO [train2.py:809] (0/4) Epoch 24, batch 1250, loss[ctc_loss=0.06241, att_loss=0.2348, loss=0.2003, over 16664.00 frames. utt_duration=1450 frames, utt_pad_proportion=0.007955, over 46.00 utterances.], tot_loss[ctc_loss=0.06904, att_loss=0.2336, loss=0.2007, over 3273142.51 frames. utt_duration=1252 frames, utt_pad_proportion=0.05113, over 10467.92 utterances.], batch size: 46, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:51:24,794 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92892.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:51:35,672 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92899.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:51:39,498 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.14 vs. limit=5.0 2023-03-09 03:52:08,598 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1008, 4.5119, 4.6207, 4.8687, 2.8108, 4.5608, 2.9397, 2.0394], device='cuda:0'), covar=tensor([0.0521, 0.0263, 0.0699, 0.0182, 0.1746, 0.0204, 0.1478, 0.1644], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0171, 0.0264, 0.0166, 0.0224, 0.0157, 0.0232, 0.0202], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:52:20,574 INFO [train2.py:809] (0/4) Epoch 24, batch 1300, loss[ctc_loss=0.04668, att_loss=0.2239, loss=0.1885, over 16529.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006096, over 45.00 utterances.], tot_loss[ctc_loss=0.07009, att_loss=0.2342, loss=0.2014, over 3270872.04 frames. utt_duration=1240 frames, utt_pad_proportion=0.05647, over 10561.62 utterances.], batch size: 45, lr: 4.50e-03, grad_scale: 8.0 2023-03-09 03:52:27,989 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92931.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:52:28,851 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-09 03:52:53,102 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.274e+02 1.940e+02 2.372e+02 2.718e+02 7.433e+02, threshold=4.745e+02, percent-clipped=1.0 2023-03-09 03:53:05,007 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8730, 4.8203, 4.8635, 4.8540, 5.3003, 4.9485, 4.6907, 2.3409], device='cuda:0'), covar=tensor([0.0144, 0.0187, 0.0185, 0.0199, 0.0766, 0.0130, 0.0220, 0.1897], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0200, 0.0198, 0.0214, 0.0371, 0.0168, 0.0187, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:53:14,569 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92960.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:53:39,342 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=92975.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:53:41,992 INFO [train2.py:809] (0/4) Epoch 24, batch 1350, loss[ctc_loss=0.06035, att_loss=0.2084, loss=0.1788, over 13664.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.06852, over 30.00 utterances.], tot_loss[ctc_loss=0.07032, att_loss=0.2343, loss=0.2015, over 3266969.23 frames. utt_duration=1207 frames, utt_pad_proportion=0.06582, over 10839.76 utterances.], batch size: 30, lr: 4.49e-03, grad_scale: 8.0 2023-03-09 03:53:49,238 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=92981.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:53:49,381 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6207, 3.1725, 3.7436, 3.3575, 3.6926, 4.7011, 4.5587, 3.5613], device='cuda:0'), covar=tensor([0.0365, 0.1692, 0.1178, 0.1149, 0.0939, 0.0843, 0.0534, 0.1035], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0246, 0.0283, 0.0220, 0.0267, 0.0372, 0.0265, 0.0233], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 03:54:07,101 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=92992.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:54:19,791 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1662, 5.4499, 5.4471, 5.3752, 5.4740, 5.4286, 5.0841, 4.8788], device='cuda:0'), covar=tensor([0.1020, 0.0483, 0.0255, 0.0454, 0.0295, 0.0304, 0.0362, 0.0346], device='cuda:0'), in_proj_covar=tensor([0.0526, 0.0365, 0.0350, 0.0362, 0.0427, 0.0435, 0.0361, 0.0398], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-09 03:54:31,608 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9644, 6.1913, 5.7158, 5.8947, 5.8932, 5.3421, 5.6995, 5.3948], device='cuda:0'), covar=tensor([0.1259, 0.0941, 0.0849, 0.0770, 0.0874, 0.1481, 0.1925, 0.2337], device='cuda:0'), in_proj_covar=tensor([0.0533, 0.0622, 0.0474, 0.0469, 0.0438, 0.0475, 0.0621, 0.0537], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 03:54:36,435 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93010.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:55:03,586 INFO [train2.py:809] (0/4) Epoch 24, batch 1400, loss[ctc_loss=0.06923, att_loss=0.2406, loss=0.2063, over 16483.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006338, over 46.00 utterances.], tot_loss[ctc_loss=0.06978, att_loss=0.2338, loss=0.201, over 3263242.55 frames. utt_duration=1213 frames, utt_pad_proportion=0.06644, over 10776.06 utterances.], batch size: 46, lr: 4.49e-03, grad_scale: 8.0 2023-03-09 03:55:19,500 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:55:19,531 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93036.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:55:33,886 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2169, 2.4805, 3.3801, 4.3015, 3.8327, 3.8754, 2.7393, 2.0725], device='cuda:0'), covar=tensor([0.0828, 0.2374, 0.0855, 0.0649, 0.0955, 0.0539, 0.1828, 0.2390], device='cuda:0'), in_proj_covar=tensor([0.0181, 0.0217, 0.0188, 0.0221, 0.0228, 0.0182, 0.0203, 0.0188], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:55:36,736 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 1.820e+02 2.183e+02 2.571e+02 5.095e+02, threshold=4.367e+02, percent-clipped=2.0 2023-03-09 03:56:16,549 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93071.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:56:24,023 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93076.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:56:25,389 INFO [train2.py:809] (0/4) Epoch 24, batch 1450, loss[ctc_loss=0.0686, att_loss=0.2403, loss=0.206, over 17131.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01429, over 56.00 utterances.], tot_loss[ctc_loss=0.06977, att_loss=0.2338, loss=0.201, over 3269171.69 frames. utt_duration=1215 frames, utt_pad_proportion=0.06401, over 10772.67 utterances.], batch size: 56, lr: 4.49e-03, grad_scale: 8.0 2023-03-09 03:56:57,949 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93097.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 03:57:26,903 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93115.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:57:28,413 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1883, 4.2820, 4.2362, 4.1198, 4.6421, 4.3210, 4.1688, 2.4010], device='cuda:0'), covar=tensor([0.0301, 0.0359, 0.0387, 0.0373, 0.0843, 0.0252, 0.0349, 0.1852], device='cuda:0'), in_proj_covar=tensor([0.0173, 0.0200, 0.0198, 0.0213, 0.0371, 0.0168, 0.0187, 0.0214], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 03:57:45,165 INFO [train2.py:809] (0/4) Epoch 24, batch 1500, loss[ctc_loss=0.06949, att_loss=0.2413, loss=0.2069, over 16778.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005608, over 48.00 utterances.], tot_loss[ctc_loss=0.0696, att_loss=0.2339, loss=0.201, over 3267177.08 frames. utt_duration=1211 frames, utt_pad_proportion=0.06608, over 10802.91 utterances.], batch size: 48, lr: 4.49e-03, grad_scale: 8.0 2023-03-09 03:58:17,655 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.288e+02 1.858e+02 2.215e+02 2.637e+02 6.796e+02, threshold=4.431e+02, percent-clipped=2.0 2023-03-09 03:59:04,767 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93176.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 03:59:05,934 INFO [train2.py:809] (0/4) Epoch 24, batch 1550, loss[ctc_loss=0.07028, att_loss=0.2334, loss=0.2007, over 15956.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006918, over 41.00 utterances.], tot_loss[ctc_loss=0.06908, att_loss=0.2327, loss=0.1999, over 3259345.89 frames. utt_duration=1242 frames, utt_pad_proportion=0.06075, over 10512.16 utterances.], batch size: 41, lr: 4.49e-03, grad_scale: 8.0 2023-03-09 03:59:23,105 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93187.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:00:26,997 INFO [train2.py:809] (0/4) Epoch 24, batch 1600, loss[ctc_loss=0.06704, att_loss=0.2361, loss=0.2023, over 17390.00 frames. utt_duration=882.3 frames, utt_pad_proportion=0.07617, over 79.00 utterances.], tot_loss[ctc_loss=0.06943, att_loss=0.2327, loss=0.2, over 3266172.39 frames. utt_duration=1245 frames, utt_pad_proportion=0.05762, over 10508.31 utterances.], batch size: 79, lr: 4.49e-03, grad_scale: 8.0 2023-03-09 04:00:46,034 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0833, 5.1349, 4.9442, 2.4331, 2.1668, 3.2165, 2.3494, 3.8358], device='cuda:0'), covar=tensor([0.0751, 0.0333, 0.0262, 0.4439, 0.5066, 0.2023, 0.3657, 0.1799], device='cuda:0'), in_proj_covar=tensor([0.0358, 0.0283, 0.0270, 0.0245, 0.0337, 0.0334, 0.0260, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 04:00:59,418 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.441e+02 2.026e+02 2.316e+02 3.152e+02 2.970e+03, threshold=4.633e+02, percent-clipped=9.0 2023-03-09 04:01:12,772 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93255.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:01:26,957 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.19 vs. limit=2.0 2023-03-09 04:01:47,572 INFO [train2.py:809] (0/4) Epoch 24, batch 1650, loss[ctc_loss=0.05659, att_loss=0.2162, loss=0.1843, over 16011.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007663, over 40.00 utterances.], tot_loss[ctc_loss=0.06912, att_loss=0.2324, loss=0.1997, over 3262289.08 frames. utt_duration=1248 frames, utt_pad_proportion=0.05773, over 10466.62 utterances.], batch size: 40, lr: 4.49e-03, grad_scale: 8.0 2023-03-09 04:01:54,062 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93281.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:02:05,029 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93287.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:03:09,900 INFO [train2.py:809] (0/4) Epoch 24, batch 1700, loss[ctc_loss=0.06908, att_loss=0.2342, loss=0.2012, over 17025.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007438, over 51.00 utterances.], tot_loss[ctc_loss=0.06904, att_loss=0.2323, loss=0.1997, over 3271695.71 frames. utt_duration=1275 frames, utt_pad_proportion=0.04927, over 10272.41 utterances.], batch size: 51, lr: 4.49e-03, grad_scale: 16.0 2023-03-09 04:03:13,047 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93329.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:03:16,315 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93331.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:03:42,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.289e+02 1.807e+02 2.162e+02 2.444e+02 4.017e+02, threshold=4.325e+02, percent-clipped=0.0 2023-03-09 04:04:14,240 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93366.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:04:29,842 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93376.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:04:31,042 INFO [train2.py:809] (0/4) Epoch 24, batch 1750, loss[ctc_loss=0.06581, att_loss=0.2102, loss=0.1813, over 15500.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008785, over 36.00 utterances.], tot_loss[ctc_loss=0.06823, att_loss=0.2314, loss=0.1988, over 3270454.42 frames. utt_duration=1305 frames, utt_pad_proportion=0.04256, over 10039.53 utterances.], batch size: 36, lr: 4.48e-03, grad_scale: 16.0 2023-03-09 04:04:56,247 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93392.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:04:56,868 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-09 04:05:28,888 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.98 vs. limit=2.0 2023-03-09 04:05:48,327 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93424.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:05:52,976 INFO [train2.py:809] (0/4) Epoch 24, batch 1800, loss[ctc_loss=0.07776, att_loss=0.2483, loss=0.2142, over 17392.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.04566, over 69.00 utterances.], tot_loss[ctc_loss=0.06871, att_loss=0.2322, loss=0.1995, over 3274426.18 frames. utt_duration=1275 frames, utt_pad_proportion=0.04738, over 10284.28 utterances.], batch size: 69, lr: 4.48e-03, grad_scale: 16.0 2023-03-09 04:06:25,986 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 1.826e+02 2.245e+02 2.637e+02 3.732e+02, threshold=4.489e+02, percent-clipped=0.0 2023-03-09 04:07:05,219 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=93471.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:07:14,514 INFO [train2.py:809] (0/4) Epoch 24, batch 1850, loss[ctc_loss=0.06197, att_loss=0.2293, loss=0.1959, over 16287.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006869, over 43.00 utterances.], tot_loss[ctc_loss=0.06864, att_loss=0.2323, loss=0.1996, over 3270677.72 frames. utt_duration=1288 frames, utt_pad_proportion=0.04521, over 10171.21 utterances.], batch size: 43, lr: 4.48e-03, grad_scale: 16.0 2023-03-09 04:07:31,221 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93487.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:07:47,715 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0379, 2.3201, 2.5983, 2.5984, 2.8627, 2.5974, 2.6000, 3.1124], device='cuda:0'), covar=tensor([0.1095, 0.2228, 0.1544, 0.1391, 0.1268, 0.1076, 0.1659, 0.1019], device='cuda:0'), in_proj_covar=tensor([0.0127, 0.0131, 0.0128, 0.0119, 0.0134, 0.0115, 0.0139, 0.0113], device='cuda:0'), out_proj_covar=tensor([9.6804e-05, 1.0241e-04, 1.0283e-04, 9.3392e-05, 1.0037e-04, 9.2422e-05, 1.0577e-04, 8.9538e-05], device='cuda:0') 2023-03-09 04:08:35,372 INFO [train2.py:809] (0/4) Epoch 24, batch 1900, loss[ctc_loss=0.05815, att_loss=0.2374, loss=0.2016, over 16944.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008619, over 50.00 utterances.], tot_loss[ctc_loss=0.06859, att_loss=0.2326, loss=0.1998, over 3274325.63 frames. utt_duration=1302 frames, utt_pad_proportion=0.04034, over 10073.41 utterances.], batch size: 50, lr: 4.48e-03, grad_scale: 16.0 2023-03-09 04:08:48,811 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93535.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:09:07,453 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.393e+02 1.868e+02 2.241e+02 2.979e+02 6.717e+02, threshold=4.482e+02, percent-clipped=6.0 2023-03-09 04:09:21,285 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93555.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:09:55,676 INFO [train2.py:809] (0/4) Epoch 24, batch 1950, loss[ctc_loss=0.04311, att_loss=0.2281, loss=0.1911, over 16469.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006538, over 46.00 utterances.], tot_loss[ctc_loss=0.06816, att_loss=0.232, loss=0.1993, over 3267737.16 frames. utt_duration=1304 frames, utt_pad_proportion=0.04139, over 10037.83 utterances.], batch size: 46, lr: 4.48e-03, grad_scale: 16.0 2023-03-09 04:10:12,608 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93587.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:10:38,923 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93603.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:10:52,560 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5050, 2.7775, 4.9066, 3.9672, 3.1239, 4.3202, 4.7735, 4.6485], device='cuda:0'), covar=tensor([0.0293, 0.1442, 0.0234, 0.0856, 0.1587, 0.0228, 0.0198, 0.0311], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0242, 0.0203, 0.0319, 0.0265, 0.0222, 0.0192, 0.0218], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:11:17,165 INFO [train2.py:809] (0/4) Epoch 24, batch 2000, loss[ctc_loss=0.1024, att_loss=0.2551, loss=0.2246, over 16950.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007675, over 50.00 utterances.], tot_loss[ctc_loss=0.06827, att_loss=0.2321, loss=0.1993, over 3270307.55 frames. utt_duration=1288 frames, utt_pad_proportion=0.04477, over 10166.03 utterances.], batch size: 50, lr: 4.48e-03, grad_scale: 16.0 2023-03-09 04:11:24,565 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93631.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:11:31,206 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93635.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:11:50,186 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 1.872e+02 2.240e+02 2.588e+02 7.651e+02, threshold=4.480e+02, percent-clipped=4.0 2023-03-09 04:12:22,044 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93666.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:12:38,936 INFO [train2.py:809] (0/4) Epoch 24, batch 2050, loss[ctc_loss=0.07013, att_loss=0.2383, loss=0.2047, over 16270.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.005951, over 43.00 utterances.], tot_loss[ctc_loss=0.06907, att_loss=0.2334, loss=0.2005, over 3277719.72 frames. utt_duration=1254 frames, utt_pad_proportion=0.05178, over 10468.00 utterances.], batch size: 43, lr: 4.48e-03, grad_scale: 16.0 2023-03-09 04:12:42,984 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93679.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:13:03,659 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93692.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:13:39,435 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93714.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:14:00,050 INFO [train2.py:809] (0/4) Epoch 24, batch 2100, loss[ctc_loss=0.05655, att_loss=0.2131, loss=0.1818, over 16019.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.006629, over 40.00 utterances.], tot_loss[ctc_loss=0.06866, att_loss=0.2335, loss=0.2005, over 3282640.45 frames. utt_duration=1271 frames, utt_pad_proportion=0.04705, over 10338.96 utterances.], batch size: 40, lr: 4.48e-03, grad_scale: 8.0 2023-03-09 04:14:21,498 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93740.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:14:34,153 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.365e+02 1.856e+02 2.273e+02 2.574e+02 6.013e+02, threshold=4.546e+02, percent-clipped=4.0 2023-03-09 04:15:01,025 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-03-09 04:15:09,551 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93770.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:15:10,985 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=93771.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:15:19,887 INFO [train2.py:809] (0/4) Epoch 24, batch 2150, loss[ctc_loss=0.07212, att_loss=0.2388, loss=0.2055, over 16970.00 frames. utt_duration=680.4 frames, utt_pad_proportion=0.1388, over 100.00 utterances.], tot_loss[ctc_loss=0.06866, att_loss=0.233, loss=0.2001, over 3278328.12 frames. utt_duration=1254 frames, utt_pad_proportion=0.05339, over 10469.61 utterances.], batch size: 100, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:16:28,699 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=93819.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:16:41,185 INFO [train2.py:809] (0/4) Epoch 24, batch 2200, loss[ctc_loss=0.06394, att_loss=0.2373, loss=0.2026, over 16937.00 frames. utt_duration=685.6 frames, utt_pad_proportion=0.1398, over 99.00 utterances.], tot_loss[ctc_loss=0.06877, att_loss=0.2331, loss=0.2003, over 3274642.35 frames. utt_duration=1241 frames, utt_pad_proportion=0.05597, over 10565.98 utterances.], batch size: 99, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:16:49,227 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=93831.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:17:16,339 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 1.790e+02 2.187e+02 2.751e+02 8.387e+02, threshold=4.374e+02, percent-clipped=4.0 2023-03-09 04:18:02,869 INFO [train2.py:809] (0/4) Epoch 24, batch 2250, loss[ctc_loss=0.04742, att_loss=0.213, loss=0.1799, over 15860.00 frames. utt_duration=1628 frames, utt_pad_proportion=0.008974, over 39.00 utterances.], tot_loss[ctc_loss=0.06849, att_loss=0.2329, loss=0.2, over 3267922.14 frames. utt_duration=1279 frames, utt_pad_proportion=0.04829, over 10233.81 utterances.], batch size: 39, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:18:26,723 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1495, 2.8323, 3.5666, 2.8563, 3.3833, 4.3128, 4.1802, 2.9603], device='cuda:0'), covar=tensor([0.0427, 0.1749, 0.1155, 0.1351, 0.1114, 0.0991, 0.0635, 0.1376], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0246, 0.0283, 0.0217, 0.0266, 0.0368, 0.0265, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 04:19:24,950 INFO [train2.py:809] (0/4) Epoch 24, batch 2300, loss[ctc_loss=0.05756, att_loss=0.2189, loss=0.1867, over 16187.00 frames. utt_duration=1581 frames, utt_pad_proportion=0.00586, over 41.00 utterances.], tot_loss[ctc_loss=0.06837, att_loss=0.2334, loss=0.2004, over 3262275.65 frames. utt_duration=1280 frames, utt_pad_proportion=0.04752, over 10206.67 utterances.], batch size: 41, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:19:50,399 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9319, 4.2313, 4.4770, 4.4615, 2.9190, 4.3373, 2.7641, 1.7288], device='cuda:0'), covar=tensor([0.0580, 0.0291, 0.0683, 0.0279, 0.1474, 0.0270, 0.1423, 0.1743], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0171, 0.0260, 0.0165, 0.0220, 0.0157, 0.0229, 0.0201], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:20:00,413 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.133e+02 1.894e+02 2.401e+02 2.825e+02 6.941e+02, threshold=4.802e+02, percent-clipped=5.0 2023-03-09 04:20:47,686 INFO [train2.py:809] (0/4) Epoch 24, batch 2350, loss[ctc_loss=0.0768, att_loss=0.2272, loss=0.1971, over 15474.00 frames. utt_duration=1720 frames, utt_pad_proportion=0.0102, over 36.00 utterances.], tot_loss[ctc_loss=0.06897, att_loss=0.2334, loss=0.2005, over 3261891.24 frames. utt_duration=1244 frames, utt_pad_proportion=0.05594, over 10503.42 utterances.], batch size: 36, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:20:54,647 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=93981.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:21:25,551 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-94000.pt 2023-03-09 04:22:13,389 INFO [train2.py:809] (0/4) Epoch 24, batch 2400, loss[ctc_loss=0.07121, att_loss=0.233, loss=0.2006, over 16280.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006843, over 43.00 utterances.], tot_loss[ctc_loss=0.06838, att_loss=0.233, loss=0.2001, over 3258296.58 frames. utt_duration=1269 frames, utt_pad_proportion=0.0517, over 10284.73 utterances.], batch size: 43, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:22:13,661 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0721, 5.3497, 4.9491, 5.4057, 4.8039, 5.1065, 5.5031, 5.2512], device='cuda:0'), covar=tensor([0.0557, 0.0254, 0.0688, 0.0288, 0.0393, 0.0242, 0.0187, 0.0204], device='cuda:0'), in_proj_covar=tensor([0.0386, 0.0321, 0.0362, 0.0350, 0.0322, 0.0238, 0.0304, 0.0285], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0005, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 04:22:38,905 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94042.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:22:47,937 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.339e+02 1.881e+02 2.121e+02 2.677e+02 6.185e+02, threshold=4.243e+02, percent-clipped=1.0 2023-03-09 04:23:05,252 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6769, 5.0614, 4.8999, 4.9808, 5.1493, 4.7548, 3.6513, 5.0878], device='cuda:0'), covar=tensor([0.0133, 0.0121, 0.0136, 0.0095, 0.0114, 0.0130, 0.0667, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0093, 0.0089, 0.0111, 0.0070, 0.0076, 0.0087, 0.0103, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 04:23:34,729 INFO [train2.py:809] (0/4) Epoch 24, batch 2450, loss[ctc_loss=0.06685, att_loss=0.2151, loss=0.1854, over 15371.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.0111, over 35.00 utterances.], tot_loss[ctc_loss=0.06803, att_loss=0.2329, loss=0.1999, over 3267357.95 frames. utt_duration=1273 frames, utt_pad_proportion=0.04721, over 10276.56 utterances.], batch size: 35, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:23:38,220 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9223, 3.6619, 3.6336, 3.2047, 3.7378, 3.7625, 3.7753, 2.7960], device='cuda:0'), covar=tensor([0.1025, 0.1050, 0.1552, 0.2604, 0.0758, 0.1949, 0.0673, 0.3102], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0199, 0.0214, 0.0266, 0.0173, 0.0274, 0.0198, 0.0227], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:24:55,081 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94126.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:24:56,468 INFO [train2.py:809] (0/4) Epoch 24, batch 2500, loss[ctc_loss=0.06189, att_loss=0.2428, loss=0.2066, over 17312.00 frames. utt_duration=877.9 frames, utt_pad_proportion=0.07588, over 79.00 utterances.], tot_loss[ctc_loss=0.068, att_loss=0.233, loss=0.2, over 3276016.40 frames. utt_duration=1264 frames, utt_pad_proportion=0.04752, over 10383.14 utterances.], batch size: 79, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:25:31,026 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.204e+02 1.808e+02 2.198e+02 2.566e+02 4.413e+02, threshold=4.397e+02, percent-clipped=1.0 2023-03-09 04:26:13,833 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0891, 4.2408, 4.4788, 4.5882, 2.6678, 4.4856, 2.8501, 1.6849], device='cuda:0'), covar=tensor([0.0452, 0.0341, 0.0620, 0.0231, 0.1654, 0.0245, 0.1357, 0.1737], device='cuda:0'), in_proj_covar=tensor([0.0203, 0.0174, 0.0264, 0.0167, 0.0224, 0.0160, 0.0232, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:26:18,035 INFO [train2.py:809] (0/4) Epoch 24, batch 2550, loss[ctc_loss=0.07312, att_loss=0.2368, loss=0.204, over 16964.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.006754, over 50.00 utterances.], tot_loss[ctc_loss=0.06795, att_loss=0.2328, loss=0.1998, over 3275146.32 frames. utt_duration=1261 frames, utt_pad_proportion=0.04942, over 10398.04 utterances.], batch size: 50, lr: 4.47e-03, grad_scale: 8.0 2023-03-09 04:26:41,665 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1343, 5.3422, 5.3442, 5.2612, 5.3853, 5.3907, 5.0478, 4.8832], device='cuda:0'), covar=tensor([0.0916, 0.0508, 0.0271, 0.0473, 0.0274, 0.0278, 0.0385, 0.0286], device='cuda:0'), in_proj_covar=tensor([0.0526, 0.0366, 0.0357, 0.0367, 0.0433, 0.0437, 0.0365, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0003, 0.0004], device='cuda:0') 2023-03-09 04:26:57,187 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-09 04:27:33,014 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94223.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:27:39,577 INFO [train2.py:809] (0/4) Epoch 24, batch 2600, loss[ctc_loss=0.07286, att_loss=0.2431, loss=0.2091, over 17034.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.008358, over 53.00 utterances.], tot_loss[ctc_loss=0.06808, att_loss=0.233, loss=0.2, over 3281101.27 frames. utt_duration=1272 frames, utt_pad_proportion=0.04606, over 10333.59 utterances.], batch size: 53, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:28:14,364 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.213e+02 1.761e+02 2.111e+02 2.578e+02 4.836e+02, threshold=4.223e+02, percent-clipped=1.0 2023-03-09 04:29:01,212 INFO [train2.py:809] (0/4) Epoch 24, batch 2650, loss[ctc_loss=0.05734, att_loss=0.2228, loss=0.1897, over 16379.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.008211, over 44.00 utterances.], tot_loss[ctc_loss=0.06793, att_loss=0.2326, loss=0.1996, over 3270671.95 frames. utt_duration=1250 frames, utt_pad_proportion=0.05397, over 10475.86 utterances.], batch size: 44, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:29:14,061 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1648, 3.7418, 3.1466, 3.4033, 3.9754, 3.6591, 3.0301, 4.2763], device='cuda:0'), covar=tensor([0.0972, 0.0518, 0.1015, 0.0729, 0.0719, 0.0726, 0.0864, 0.0502], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0222, 0.0226, 0.0205, 0.0282, 0.0246, 0.0201, 0.0292], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 04:29:14,098 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94284.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:30:23,808 INFO [train2.py:809] (0/4) Epoch 24, batch 2700, loss[ctc_loss=0.065, att_loss=0.2379, loss=0.2033, over 16871.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.006719, over 49.00 utterances.], tot_loss[ctc_loss=0.06741, att_loss=0.2317, loss=0.1988, over 3267450.50 frames. utt_duration=1258 frames, utt_pad_proportion=0.05267, over 10404.22 utterances.], batch size: 49, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:30:28,862 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0069, 5.2509, 5.2282, 5.1870, 5.2282, 5.2373, 4.8916, 4.7052], device='cuda:0'), covar=tensor([0.0953, 0.0533, 0.0304, 0.0475, 0.0297, 0.0325, 0.0401, 0.0344], device='cuda:0'), in_proj_covar=tensor([0.0529, 0.0369, 0.0359, 0.0369, 0.0435, 0.0441, 0.0367, 0.0401], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 04:30:37,633 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2034, 3.8021, 3.2227, 3.4905, 4.0241, 3.6848, 3.1370, 4.3468], device='cuda:0'), covar=tensor([0.1016, 0.0562, 0.1180, 0.0770, 0.0738, 0.0846, 0.0936, 0.0486], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0223, 0.0227, 0.0206, 0.0284, 0.0248, 0.0202, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 04:30:40,542 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94337.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:30:57,579 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.293e+02 1.818e+02 2.150e+02 2.630e+02 4.839e+02, threshold=4.300e+02, percent-clipped=3.0 2023-03-09 04:30:58,034 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=94348.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:31:45,415 INFO [train2.py:809] (0/4) Epoch 24, batch 2750, loss[ctc_loss=0.06354, att_loss=0.2276, loss=0.1948, over 15899.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008428, over 39.00 utterances.], tot_loss[ctc_loss=0.0682, att_loss=0.2318, loss=0.1991, over 3263661.45 frames. utt_duration=1244 frames, utt_pad_proportion=0.05801, over 10508.63 utterances.], batch size: 39, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:32:37,494 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=94409.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 04:33:05,986 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94426.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:33:07,896 INFO [train2.py:809] (0/4) Epoch 24, batch 2800, loss[ctc_loss=0.07151, att_loss=0.2443, loss=0.2097, over 17245.00 frames. utt_duration=1096 frames, utt_pad_proportion=0.04002, over 63.00 utterances.], tot_loss[ctc_loss=0.06857, att_loss=0.2324, loss=0.1996, over 3271735.76 frames. utt_duration=1265 frames, utt_pad_proportion=0.04992, over 10354.92 utterances.], batch size: 63, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:33:40,419 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.163e+02 1.912e+02 2.313e+02 2.793e+02 7.224e+02, threshold=4.627e+02, percent-clipped=4.0 2023-03-09 04:34:24,215 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94474.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:34:28,880 INFO [train2.py:809] (0/4) Epoch 24, batch 2850, loss[ctc_loss=0.05543, att_loss=0.2118, loss=0.1805, over 15769.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.00862, over 38.00 utterances.], tot_loss[ctc_loss=0.06889, att_loss=0.2325, loss=0.1998, over 3271595.62 frames. utt_duration=1258 frames, utt_pad_proportion=0.05208, over 10410.94 utterances.], batch size: 38, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:34:41,361 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-09 04:35:52,346 INFO [train2.py:809] (0/4) Epoch 24, batch 2900, loss[ctc_loss=0.06009, att_loss=0.2123, loss=0.1819, over 14028.00 frames. utt_duration=1812 frames, utt_pad_proportion=0.05934, over 31.00 utterances.], tot_loss[ctc_loss=0.06881, att_loss=0.2324, loss=0.1997, over 3266350.56 frames. utt_duration=1250 frames, utt_pad_proportion=0.05526, over 10466.69 utterances.], batch size: 31, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:36:26,076 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.225e+02 1.878e+02 2.321e+02 2.758e+02 4.740e+02, threshold=4.642e+02, percent-clipped=1.0 2023-03-09 04:37:14,352 INFO [train2.py:809] (0/4) Epoch 24, batch 2950, loss[ctc_loss=0.06265, att_loss=0.2084, loss=0.1793, over 15340.00 frames. utt_duration=1755 frames, utt_pad_proportion=0.006921, over 35.00 utterances.], tot_loss[ctc_loss=0.06844, att_loss=0.2324, loss=0.1996, over 3264602.22 frames. utt_duration=1248 frames, utt_pad_proportion=0.05571, over 10473.39 utterances.], batch size: 35, lr: 4.46e-03, grad_scale: 8.0 2023-03-09 04:37:17,807 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94579.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:37:51,920 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3939, 4.3750, 4.3262, 4.3957, 4.9396, 4.4707, 4.3236, 2.4488], device='cuda:0'), covar=tensor([0.0246, 0.0345, 0.0398, 0.0297, 0.0708, 0.0228, 0.0376, 0.1846], device='cuda:0'), in_proj_covar=tensor([0.0175, 0.0202, 0.0201, 0.0215, 0.0374, 0.0171, 0.0190, 0.0216], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:38:03,427 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0815, 5.3035, 5.2700, 5.1786, 5.3492, 5.3376, 4.9758, 4.8359], device='cuda:0'), covar=tensor([0.0936, 0.0520, 0.0290, 0.0530, 0.0258, 0.0293, 0.0386, 0.0295], device='cuda:0'), in_proj_covar=tensor([0.0528, 0.0369, 0.0358, 0.0370, 0.0433, 0.0442, 0.0369, 0.0399], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 04:38:34,055 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.13 vs. limit=2.0 2023-03-09 04:38:36,145 INFO [train2.py:809] (0/4) Epoch 24, batch 3000, loss[ctc_loss=0.08752, att_loss=0.2539, loss=0.2206, over 17362.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.0361, over 63.00 utterances.], tot_loss[ctc_loss=0.0687, att_loss=0.2327, loss=0.1999, over 3270743.37 frames. utt_duration=1254 frames, utt_pad_proportion=0.05218, over 10441.41 utterances.], batch size: 63, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:38:36,148 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 04:38:48,226 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6322, 2.7599, 2.7727, 2.5664, 2.8014, 2.5757, 2.8542, 2.0544], device='cuda:0'), covar=tensor([0.1097, 0.1656, 0.1857, 0.3429, 0.1228, 0.2298, 0.1458, 0.4307], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0197, 0.0211, 0.0263, 0.0173, 0.0271, 0.0198, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:38:50,630 INFO [train2.py:843] (0/4) Epoch 24, validation: ctc_loss=0.04165, att_loss=0.2345, loss=0.196, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 04:38:50,630 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 04:39:06,727 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94637.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:39:23,559 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 2.097e+02 2.380e+02 3.047e+02 5.871e+02, threshold=4.759e+02, percent-clipped=2.0 2023-03-09 04:40:11,574 INFO [train2.py:809] (0/4) Epoch 24, batch 3050, loss[ctc_loss=0.09028, att_loss=0.25, loss=0.218, over 17273.00 frames. utt_duration=876.2 frames, utt_pad_proportion=0.08252, over 79.00 utterances.], tot_loss[ctc_loss=0.06982, att_loss=0.2333, loss=0.2006, over 3259969.71 frames. utt_duration=1217 frames, utt_pad_proportion=0.06618, over 10730.03 utterances.], batch size: 79, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:40:22,786 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2870, 4.6553, 4.6056, 4.7018, 4.7335, 4.4475, 3.3361, 4.6022], device='cuda:0'), covar=tensor([0.0149, 0.0135, 0.0151, 0.0113, 0.0129, 0.0133, 0.0737, 0.0253], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0091, 0.0114, 0.0071, 0.0078, 0.0089, 0.0105, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 04:40:24,108 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94685.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:40:54,712 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=94704.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:41:32,546 INFO [train2.py:809] (0/4) Epoch 24, batch 3100, loss[ctc_loss=0.07811, att_loss=0.256, loss=0.2204, over 17096.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01473, over 56.00 utterances.], tot_loss[ctc_loss=0.0697, att_loss=0.2338, loss=0.2009, over 3263681.31 frames. utt_duration=1226 frames, utt_pad_proportion=0.06179, over 10662.94 utterances.], batch size: 56, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:42:05,407 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.429e+02 1.910e+02 2.247e+02 2.795e+02 4.738e+02, threshold=4.495e+02, percent-clipped=0.0 2023-03-09 04:42:08,252 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-09 04:42:54,029 INFO [train2.py:809] (0/4) Epoch 24, batch 3150, loss[ctc_loss=0.07241, att_loss=0.2105, loss=0.1829, over 15655.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.008307, over 37.00 utterances.], tot_loss[ctc_loss=0.06986, att_loss=0.2334, loss=0.2007, over 3258509.49 frames. utt_duration=1226 frames, utt_pad_proportion=0.06418, over 10641.35 utterances.], batch size: 37, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:43:01,857 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.16 vs. limit=5.0 2023-03-09 04:43:23,038 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0389, 4.9187, 4.9294, 5.0134, 5.3936, 4.9924, 4.8688, 2.5697], device='cuda:0'), covar=tensor([0.0123, 0.0183, 0.0186, 0.0154, 0.0569, 0.0131, 0.0183, 0.1680], device='cuda:0'), in_proj_covar=tensor([0.0176, 0.0202, 0.0201, 0.0216, 0.0374, 0.0172, 0.0190, 0.0217], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:44:15,928 INFO [train2.py:809] (0/4) Epoch 24, batch 3200, loss[ctc_loss=0.061, att_loss=0.2114, loss=0.1813, over 16012.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007833, over 40.00 utterances.], tot_loss[ctc_loss=0.06937, att_loss=0.2335, loss=0.2007, over 3265866.56 frames. utt_duration=1229 frames, utt_pad_proportion=0.06226, over 10640.61 utterances.], batch size: 40, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:44:44,507 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3172, 3.8089, 3.2657, 3.5365, 4.0007, 3.7100, 3.0628, 4.2951], device='cuda:0'), covar=tensor([0.0877, 0.0501, 0.1005, 0.0662, 0.0692, 0.0708, 0.0876, 0.0508], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0223, 0.0226, 0.0205, 0.0283, 0.0245, 0.0203, 0.0292], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 04:44:48,825 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.290e+02 1.841e+02 2.191e+02 2.766e+02 6.449e+02, threshold=4.381e+02, percent-clipped=4.0 2023-03-09 04:45:14,010 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.70 vs. limit=2.0 2023-03-09 04:45:36,260 INFO [train2.py:809] (0/4) Epoch 24, batch 3250, loss[ctc_loss=0.09206, att_loss=0.2535, loss=0.2212, over 17040.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009521, over 53.00 utterances.], tot_loss[ctc_loss=0.06946, att_loss=0.2334, loss=0.2006, over 3258506.45 frames. utt_duration=1210 frames, utt_pad_proportion=0.06908, over 10781.54 utterances.], batch size: 53, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:45:37,332 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.24 vs. limit=5.0 2023-03-09 04:45:39,695 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=94879.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:46:56,541 INFO [train2.py:809] (0/4) Epoch 24, batch 3300, loss[ctc_loss=0.08518, att_loss=0.2401, loss=0.2091, over 15972.00 frames. utt_duration=1599 frames, utt_pad_proportion=0.008266, over 40.00 utterances.], tot_loss[ctc_loss=0.06928, att_loss=0.2335, loss=0.2006, over 3259982.87 frames. utt_duration=1209 frames, utt_pad_proportion=0.06915, over 10795.18 utterances.], batch size: 40, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:46:56,641 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=94927.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:47:29,073 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.188e+02 1.857e+02 2.212e+02 2.661e+02 5.199e+02, threshold=4.424e+02, percent-clipped=2.0 2023-03-09 04:48:16,304 INFO [train2.py:809] (0/4) Epoch 24, batch 3350, loss[ctc_loss=0.08325, att_loss=0.2486, loss=0.2155, over 16883.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007649, over 49.00 utterances.], tot_loss[ctc_loss=0.06857, att_loss=0.2324, loss=0.1996, over 3258483.73 frames. utt_duration=1239 frames, utt_pad_proportion=0.06243, over 10533.45 utterances.], batch size: 49, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:48:59,814 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95004.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:49:36,971 INFO [train2.py:809] (0/4) Epoch 24, batch 3400, loss[ctc_loss=0.07568, att_loss=0.2343, loss=0.2026, over 16910.00 frames. utt_duration=684.8 frames, utt_pad_proportion=0.1407, over 99.00 utterances.], tot_loss[ctc_loss=0.06921, att_loss=0.2323, loss=0.1997, over 3256681.31 frames. utt_duration=1229 frames, utt_pad_proportion=0.0656, over 10614.29 utterances.], batch size: 99, lr: 4.45e-03, grad_scale: 8.0 2023-03-09 04:50:09,912 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.325e+02 1.881e+02 2.292e+02 2.864e+02 6.221e+02, threshold=4.584e+02, percent-clipped=2.0 2023-03-09 04:50:16,228 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95052.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:50:49,310 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.25 vs. limit=5.0 2023-03-09 04:50:56,262 INFO [train2.py:809] (0/4) Epoch 24, batch 3450, loss[ctc_loss=0.1263, att_loss=0.2623, loss=0.2351, over 14464.00 frames. utt_duration=397.8 frames, utt_pad_proportion=0.3057, over 146.00 utterances.], tot_loss[ctc_loss=0.06975, att_loss=0.2327, loss=0.2001, over 3261555.54 frames. utt_duration=1212 frames, utt_pad_proportion=0.06812, over 10774.28 utterances.], batch size: 146, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 04:51:14,242 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9798, 3.7572, 3.6702, 3.2005, 3.7427, 3.7740, 3.7335, 2.7885], device='cuda:0'), covar=tensor([0.1063, 0.1024, 0.1782, 0.2937, 0.0926, 0.2168, 0.0991, 0.3178], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0197, 0.0213, 0.0264, 0.0173, 0.0271, 0.0198, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:51:38,280 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3241, 5.2776, 4.9888, 2.8985, 2.4820, 3.4419, 2.8149, 4.1464], device='cuda:0'), covar=tensor([0.0699, 0.0349, 0.0305, 0.3939, 0.4817, 0.1987, 0.3111, 0.1571], device='cuda:0'), in_proj_covar=tensor([0.0362, 0.0287, 0.0274, 0.0249, 0.0342, 0.0337, 0.0261, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 04:51:41,233 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95105.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 04:52:16,793 INFO [train2.py:809] (0/4) Epoch 24, batch 3500, loss[ctc_loss=0.05487, att_loss=0.2263, loss=0.192, over 15953.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006533, over 41.00 utterances.], tot_loss[ctc_loss=0.06984, att_loss=0.2322, loss=0.1997, over 3249593.87 frames. utt_duration=1244 frames, utt_pad_proportion=0.06278, over 10465.76 utterances.], batch size: 41, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 04:52:22,511 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-03-09 04:52:49,701 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.240e+01 1.898e+02 2.336e+02 2.890e+02 6.917e+02, threshold=4.673e+02, percent-clipped=5.0 2023-03-09 04:52:53,490 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.63 vs. limit=2.0 2023-03-09 04:53:18,794 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95166.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 04:53:35,720 INFO [train2.py:809] (0/4) Epoch 24, batch 3550, loss[ctc_loss=0.06457, att_loss=0.2332, loss=0.1995, over 16472.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.007046, over 46.00 utterances.], tot_loss[ctc_loss=0.07047, att_loss=0.2327, loss=0.2002, over 3249838.57 frames. utt_duration=1241 frames, utt_pad_proportion=0.06352, over 10485.20 utterances.], batch size: 46, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 04:54:02,978 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.94 vs. limit=2.0 2023-03-09 04:54:52,844 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-09 04:54:56,143 INFO [train2.py:809] (0/4) Epoch 24, batch 3600, loss[ctc_loss=0.04539, att_loss=0.1929, loss=0.1634, over 15638.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.008724, over 37.00 utterances.], tot_loss[ctc_loss=0.07108, att_loss=0.2329, loss=0.2006, over 3249333.66 frames. utt_duration=1228 frames, utt_pad_proportion=0.06667, over 10600.38 utterances.], batch size: 37, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 04:55:29,598 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 1.952e+02 2.410e+02 2.837e+02 7.042e+02, threshold=4.821e+02, percent-clipped=2.0 2023-03-09 04:55:50,109 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95260.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:56:02,678 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95268.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:56:16,590 INFO [train2.py:809] (0/4) Epoch 24, batch 3650, loss[ctc_loss=0.05935, att_loss=0.2203, loss=0.1881, over 16130.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005498, over 42.00 utterances.], tot_loss[ctc_loss=0.07017, att_loss=0.2331, loss=0.2005, over 3261754.58 frames. utt_duration=1246 frames, utt_pad_proportion=0.05941, over 10487.64 utterances.], batch size: 42, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 04:56:28,201 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6100, 2.4765, 5.0326, 4.0504, 3.1891, 4.3693, 4.8906, 4.7026], device='cuda:0'), covar=tensor([0.0255, 0.1726, 0.0193, 0.0842, 0.1610, 0.0233, 0.0143, 0.0287], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0246, 0.0207, 0.0323, 0.0269, 0.0227, 0.0195, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:57:09,438 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0895, 6.2585, 5.6680, 5.9787, 5.9315, 5.3652, 5.7735, 5.4361], device='cuda:0'), covar=tensor([0.1267, 0.0981, 0.1020, 0.0931, 0.0923, 0.1744, 0.2358, 0.2610], device='cuda:0'), in_proj_covar=tensor([0.0541, 0.0629, 0.0476, 0.0474, 0.0443, 0.0479, 0.0630, 0.0541], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 04:57:28,635 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95321.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:57:37,488 INFO [train2.py:809] (0/4) Epoch 24, batch 3700, loss[ctc_loss=0.1129, att_loss=0.2579, loss=0.2289, over 14367.00 frames. utt_duration=395.1 frames, utt_pad_proportion=0.3104, over 146.00 utterances.], tot_loss[ctc_loss=0.07048, att_loss=0.2334, loss=0.2008, over 3261452.12 frames. utt_duration=1229 frames, utt_pad_proportion=0.06311, over 10631.40 utterances.], batch size: 146, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 04:57:40,849 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95329.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:57:56,701 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95339.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:58:11,157 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.183e+02 1.823e+02 2.114e+02 2.643e+02 5.105e+02, threshold=4.228e+02, percent-clipped=2.0 2023-03-09 04:58:28,779 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8929, 3.6789, 3.6844, 3.2653, 3.7633, 3.7375, 3.7298, 2.7499], device='cuda:0'), covar=tensor([0.1279, 0.1232, 0.1812, 0.2673, 0.0798, 0.2155, 0.0858, 0.3443], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0198, 0.0212, 0.0264, 0.0173, 0.0274, 0.0198, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:58:41,255 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1038, 4.2711, 4.3723, 4.6378, 2.6863, 4.4518, 2.7801, 2.2752], device='cuda:0'), covar=tensor([0.0438, 0.0332, 0.0812, 0.0252, 0.1775, 0.0252, 0.1532, 0.1584], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0176, 0.0263, 0.0168, 0.0224, 0.0160, 0.0232, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 04:58:47,553 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.52 vs. limit=2.0 2023-03-09 04:58:58,643 INFO [train2.py:809] (0/4) Epoch 24, batch 3750, loss[ctc_loss=0.05628, att_loss=0.2293, loss=0.1947, over 17017.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008064, over 51.00 utterances.], tot_loss[ctc_loss=0.07066, att_loss=0.2336, loss=0.201, over 3266828.21 frames. utt_duration=1219 frames, utt_pad_proportion=0.06464, over 10730.06 utterances.], batch size: 51, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 04:59:35,867 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95400.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 04:59:41,669 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9537, 5.1638, 5.4756, 5.3144, 5.3856, 5.9015, 5.2315, 6.0239], device='cuda:0'), covar=tensor([0.0762, 0.0824, 0.0835, 0.1526, 0.2007, 0.0919, 0.0766, 0.0670], device='cuda:0'), in_proj_covar=tensor([0.0896, 0.0520, 0.0626, 0.0673, 0.0904, 0.0649, 0.0509, 0.0633], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:00:19,264 INFO [train2.py:809] (0/4) Epoch 24, batch 3800, loss[ctc_loss=0.04918, att_loss=0.2289, loss=0.1929, over 16625.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.004575, over 47.00 utterances.], tot_loss[ctc_loss=0.07012, att_loss=0.2328, loss=0.2002, over 3259130.74 frames. utt_duration=1225 frames, utt_pad_proportion=0.0641, over 10653.87 utterances.], batch size: 47, lr: 4.44e-03, grad_scale: 8.0 2023-03-09 05:00:41,253 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.5846, 4.9865, 5.2188, 5.0007, 5.0982, 5.5172, 4.9800, 5.5871], device='cuda:0'), covar=tensor([0.0771, 0.0662, 0.0790, 0.1437, 0.1913, 0.0975, 0.0980, 0.0743], device='cuda:0'), in_proj_covar=tensor([0.0900, 0.0523, 0.0630, 0.0676, 0.0908, 0.0652, 0.0511, 0.0637], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:00:52,294 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.438e+02 1.978e+02 2.351e+02 3.015e+02 8.965e+02, threshold=4.702e+02, percent-clipped=6.0 2023-03-09 05:00:57,066 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4866, 3.1603, 3.5841, 4.6261, 4.1582, 3.9881, 2.9941, 2.4113], device='cuda:0'), covar=tensor([0.0781, 0.1783, 0.0833, 0.0499, 0.0906, 0.0565, 0.1649, 0.2119], device='cuda:0'), in_proj_covar=tensor([0.0183, 0.0220, 0.0187, 0.0221, 0.0229, 0.0184, 0.0206, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:01:00,315 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9717, 3.9422, 3.9810, 4.3870, 2.7259, 4.3148, 2.7318, 1.8474], device='cuda:0'), covar=tensor([0.0488, 0.0320, 0.0759, 0.0251, 0.1561, 0.0237, 0.1397, 0.1619], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0175, 0.0263, 0.0168, 0.0224, 0.0161, 0.0231, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:01:13,098 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95461.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 05:01:38,903 INFO [train2.py:809] (0/4) Epoch 24, batch 3850, loss[ctc_loss=0.1002, att_loss=0.2652, loss=0.2322, over 17334.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.0101, over 55.00 utterances.], tot_loss[ctc_loss=0.07054, att_loss=0.2337, loss=0.2011, over 3263303.68 frames. utt_duration=1203 frames, utt_pad_proportion=0.06958, over 10867.65 utterances.], batch size: 55, lr: 4.43e-03, grad_scale: 8.0 2023-03-09 05:01:57,649 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.43 vs. limit=2.0 2023-03-09 05:02:54,832 INFO [train2.py:809] (0/4) Epoch 24, batch 3900, loss[ctc_loss=0.06363, att_loss=0.2341, loss=0.2, over 16607.00 frames. utt_duration=679.4 frames, utt_pad_proportion=0.1486, over 98.00 utterances.], tot_loss[ctc_loss=0.07087, att_loss=0.2346, loss=0.2019, over 3275182.34 frames. utt_duration=1208 frames, utt_pad_proportion=0.06475, over 10857.83 utterances.], batch size: 98, lr: 4.43e-03, grad_scale: 8.0 2023-03-09 05:03:26,620 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.085e+02 1.865e+02 2.329e+02 2.791e+02 6.016e+02, threshold=4.658e+02, percent-clipped=3.0 2023-03-09 05:04:10,395 INFO [train2.py:809] (0/4) Epoch 24, batch 3950, loss[ctc_loss=0.05647, att_loss=0.2324, loss=0.1972, over 16778.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.00583, over 48.00 utterances.], tot_loss[ctc_loss=0.07022, att_loss=0.2336, loss=0.2009, over 3261992.10 frames. utt_duration=1196 frames, utt_pad_proportion=0.07044, over 10921.25 utterances.], batch size: 48, lr: 4.43e-03, grad_scale: 8.0 2023-03-09 05:05:01,991 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-24.pt 2023-03-09 05:05:25,986 INFO [train2.py:809] (0/4) Epoch 25, batch 0, loss[ctc_loss=0.08161, att_loss=0.2304, loss=0.2006, over 16169.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007604, over 41.00 utterances.], tot_loss[ctc_loss=0.08161, att_loss=0.2304, loss=0.2006, over 16169.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007604, over 41.00 utterances.], batch size: 41, lr: 4.34e-03, grad_scale: 8.0 2023-03-09 05:05:25,988 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 05:05:38,244 INFO [train2.py:843] (0/4) Epoch 25, validation: ctc_loss=0.04004, att_loss=0.2344, loss=0.1955, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 05:05:38,245 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 05:05:46,417 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95616.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:05:48,088 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95617.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:05:58,886 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95624.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:06:36,946 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 1.872e+02 2.311e+02 2.775e+02 5.959e+02, threshold=4.621e+02, percent-clipped=4.0 2023-03-09 05:06:54,887 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6620, 3.2673, 3.8331, 3.4477, 3.7516, 4.7927, 4.6468, 3.5058], device='cuda:0'), covar=tensor([0.0384, 0.1572, 0.1214, 0.1194, 0.1078, 0.0834, 0.0589, 0.1208], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0248, 0.0287, 0.0222, 0.0270, 0.0376, 0.0268, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:06:57,663 INFO [train2.py:809] (0/4) Epoch 25, batch 50, loss[ctc_loss=0.05725, att_loss=0.2009, loss=0.1722, over 13639.00 frames. utt_duration=1820 frames, utt_pad_proportion=0.08021, over 30.00 utterances.], tot_loss[ctc_loss=0.07144, att_loss=0.2321, loss=0.2, over 736942.92 frames. utt_duration=1258 frames, utt_pad_proportion=0.05541, over 2346.12 utterances.], batch size: 30, lr: 4.34e-03, grad_scale: 8.0 2023-03-09 05:07:25,576 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95678.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:07:52,027 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95695.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:08:17,154 INFO [train2.py:809] (0/4) Epoch 25, batch 100, loss[ctc_loss=0.07625, att_loss=0.2408, loss=0.2079, over 16884.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006465, over 49.00 utterances.], tot_loss[ctc_loss=0.07065, att_loss=0.2343, loss=0.2016, over 1301786.17 frames. utt_duration=1247 frames, utt_pad_proportion=0.05279, over 4179.22 utterances.], batch size: 49, lr: 4.34e-03, grad_scale: 8.0 2023-03-09 05:09:15,974 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 1.992e+02 2.344e+02 2.996e+02 8.490e+02, threshold=4.688e+02, percent-clipped=4.0 2023-03-09 05:09:36,914 INFO [train2.py:809] (0/4) Epoch 25, batch 150, loss[ctc_loss=0.06991, att_loss=0.2197, loss=0.1898, over 11357.00 frames. utt_duration=1819 frames, utt_pad_proportion=0.05069, over 25.00 utterances.], tot_loss[ctc_loss=0.06926, att_loss=0.2324, loss=0.1998, over 1725460.37 frames. utt_duration=1272 frames, utt_pad_proportion=0.05363, over 5433.82 utterances.], batch size: 25, lr: 4.34e-03, grad_scale: 16.0 2023-03-09 05:09:37,251 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95761.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 05:10:02,052 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9245, 5.2111, 5.0950, 5.0991, 5.2019, 5.1817, 4.8704, 4.6898], device='cuda:0'), covar=tensor([0.1061, 0.0488, 0.0313, 0.0521, 0.0262, 0.0327, 0.0436, 0.0353], device='cuda:0'), in_proj_covar=tensor([0.0529, 0.0371, 0.0359, 0.0371, 0.0432, 0.0438, 0.0367, 0.0400], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:10:55,122 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95809.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 05:10:58,461 INFO [train2.py:809] (0/4) Epoch 25, batch 200, loss[ctc_loss=0.09877, att_loss=0.2699, loss=0.2357, over 17358.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03463, over 63.00 utterances.], tot_loss[ctc_loss=0.06937, att_loss=0.2328, loss=0.2001, over 2071596.71 frames. utt_duration=1267 frames, utt_pad_proportion=0.05065, over 6548.88 utterances.], batch size: 63, lr: 4.34e-03, grad_scale: 16.0 2023-03-09 05:11:00,453 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95812.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 05:11:55,496 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.81 vs. limit=2.0 2023-03-09 05:11:56,643 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95847.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:11:57,806 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.266e+02 1.872e+02 2.363e+02 3.018e+02 4.508e+02, threshold=4.726e+02, percent-clipped=0.0 2023-03-09 05:12:18,646 INFO [train2.py:809] (0/4) Epoch 25, batch 250, loss[ctc_loss=0.07096, att_loss=0.2443, loss=0.2097, over 16725.00 frames. utt_duration=677.3 frames, utt_pad_proportion=0.1459, over 99.00 utterances.], tot_loss[ctc_loss=0.06932, att_loss=0.2329, loss=0.2001, over 2336673.70 frames. utt_duration=1245 frames, utt_pad_proportion=0.0549, over 7516.73 utterances.], batch size: 99, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:12:38,658 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95873.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 05:13:34,960 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=95908.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 05:13:39,386 INFO [train2.py:809] (0/4) Epoch 25, batch 300, loss[ctc_loss=0.06587, att_loss=0.2458, loss=0.2098, over 17297.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02311, over 59.00 utterances.], tot_loss[ctc_loss=0.06896, att_loss=0.2328, loss=0.2, over 2543120.38 frames. utt_duration=1251 frames, utt_pad_proportion=0.05504, over 8142.55 utterances.], batch size: 59, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:13:48,165 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95916.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:14:00,734 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95924.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:14:38,788 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.143e+02 1.855e+02 2.204e+02 2.784e+02 5.954e+02, threshold=4.408e+02, percent-clipped=3.0 2023-03-09 05:14:55,231 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1771, 5.4598, 5.4202, 5.3346, 5.5056, 5.4619, 5.1256, 4.9551], device='cuda:0'), covar=tensor([0.1092, 0.0543, 0.0279, 0.0522, 0.0276, 0.0342, 0.0394, 0.0320], device='cuda:0'), in_proj_covar=tensor([0.0532, 0.0373, 0.0359, 0.0372, 0.0433, 0.0440, 0.0368, 0.0401], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:14:56,265 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.80 vs. limit=5.0 2023-03-09 05:14:59,612 INFO [train2.py:809] (0/4) Epoch 25, batch 350, loss[ctc_loss=0.06591, att_loss=0.2204, loss=0.1895, over 16116.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.005717, over 42.00 utterances.], tot_loss[ctc_loss=0.06877, att_loss=0.2332, loss=0.2003, over 2706104.67 frames. utt_duration=1242 frames, utt_pad_proportion=0.05662, over 8724.89 utterances.], batch size: 42, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:15:04,894 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95964.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:15:08,170 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5269, 2.9561, 3.8034, 2.8897, 3.5468, 4.7257, 4.5511, 3.3742], device='cuda:0'), covar=tensor([0.0471, 0.1873, 0.1197, 0.1575, 0.1155, 0.0770, 0.0591, 0.1240], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0248, 0.0286, 0.0222, 0.0269, 0.0375, 0.0266, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:15:17,277 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=95972.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:15:18,965 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=95973.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:15:30,457 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=95980.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:15:54,744 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=95995.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:16:02,564 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-96000.pt 2023-03-09 05:16:08,722 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-03-09 05:16:17,993 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8920, 5.2056, 5.0537, 5.2042, 5.2692, 4.9275, 3.5710, 5.3121], device='cuda:0'), covar=tensor([0.0119, 0.0118, 0.0155, 0.0080, 0.0127, 0.0125, 0.0758, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0092, 0.0115, 0.0072, 0.0079, 0.0089, 0.0106, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:16:24,355 INFO [train2.py:809] (0/4) Epoch 25, batch 400, loss[ctc_loss=0.07728, att_loss=0.2497, loss=0.2152, over 16623.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005528, over 47.00 utterances.], tot_loss[ctc_loss=0.06796, att_loss=0.2326, loss=0.1997, over 2830084.83 frames. utt_duration=1265 frames, utt_pad_proportion=0.05187, over 8961.02 utterances.], batch size: 47, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:16:33,735 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6779, 5.0100, 5.2426, 4.9801, 5.1561, 5.6300, 5.0430, 5.7141], device='cuda:0'), covar=tensor([0.0713, 0.0793, 0.0882, 0.1447, 0.1732, 0.0908, 0.1000, 0.0716], device='cuda:0'), in_proj_covar=tensor([0.0900, 0.0519, 0.0627, 0.0672, 0.0899, 0.0650, 0.0511, 0.0636], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:17:07,594 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8061, 4.8444, 4.6633, 3.0704, 4.6729, 4.6106, 4.1128, 2.6865], device='cuda:0'), covar=tensor([0.0116, 0.0102, 0.0270, 0.0894, 0.0096, 0.0189, 0.0313, 0.1319], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0105, 0.0107, 0.0113, 0.0088, 0.0116, 0.0101, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:17:12,266 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96041.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:17:15,135 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96043.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:17:22,747 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 1.820e+02 2.178e+02 2.552e+02 4.436e+02, threshold=4.357e+02, percent-clipped=1.0 2023-03-09 05:17:43,676 INFO [train2.py:809] (0/4) Epoch 25, batch 450, loss[ctc_loss=0.0954, att_loss=0.2453, loss=0.2153, over 16331.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006293, over 45.00 utterances.], tot_loss[ctc_loss=0.06811, att_loss=0.2329, loss=0.2, over 2929810.24 frames. utt_duration=1227 frames, utt_pad_proportion=0.05898, over 9564.21 utterances.], batch size: 45, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:18:14,099 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96080.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:19:03,645 INFO [train2.py:809] (0/4) Epoch 25, batch 500, loss[ctc_loss=0.07151, att_loss=0.251, loss=0.2151, over 17030.00 frames. utt_duration=1311 frames, utt_pad_proportion=0.009601, over 52.00 utterances.], tot_loss[ctc_loss=0.06802, att_loss=0.2323, loss=0.1995, over 3003989.04 frames. utt_duration=1241 frames, utt_pad_proportion=0.05559, over 9695.40 utterances.], batch size: 52, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:19:21,066 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0308, 5.3236, 4.9831, 5.3868, 4.7205, 4.9923, 5.5022, 5.2032], device='cuda:0'), covar=tensor([0.0598, 0.0307, 0.0711, 0.0323, 0.0480, 0.0262, 0.0210, 0.0207], device='cuda:0'), in_proj_covar=tensor([0.0397, 0.0330, 0.0372, 0.0361, 0.0331, 0.0244, 0.0313, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 05:19:45,922 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4818, 3.9084, 3.4119, 3.6463, 4.1650, 3.8016, 3.1646, 4.4840], device='cuda:0'), covar=tensor([0.0801, 0.0502, 0.1109, 0.0683, 0.0653, 0.0695, 0.0856, 0.0447], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0226, 0.0228, 0.0208, 0.0288, 0.0246, 0.0204, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 05:19:50,603 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96141.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:20:00,865 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.348e+02 1.791e+02 2.138e+02 2.685e+02 5.642e+02, threshold=4.276e+02, percent-clipped=2.0 2023-03-09 05:20:21,839 INFO [train2.py:809] (0/4) Epoch 25, batch 550, loss[ctc_loss=0.04296, att_loss=0.1976, loss=0.1667, over 15854.00 frames. utt_duration=1628 frames, utt_pad_proportion=0.01117, over 39.00 utterances.], tot_loss[ctc_loss=0.06817, att_loss=0.2317, loss=0.199, over 3054148.51 frames. utt_duration=1251 frames, utt_pad_proportion=0.05782, over 9779.05 utterances.], batch size: 39, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:20:28,326 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96165.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:20:32,703 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96168.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 05:21:27,357 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96203.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 05:21:39,960 INFO [train2.py:809] (0/4) Epoch 25, batch 600, loss[ctc_loss=0.06836, att_loss=0.2206, loss=0.1902, over 15958.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006253, over 41.00 utterances.], tot_loss[ctc_loss=0.06895, att_loss=0.2325, loss=0.1998, over 3101588.54 frames. utt_duration=1222 frames, utt_pad_proportion=0.06387, over 10168.43 utterances.], batch size: 41, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:21:40,472 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5879, 3.7619, 3.9255, 2.3639, 2.2813, 2.8785, 2.4642, 3.4951], device='cuda:0'), covar=tensor([0.0819, 0.0450, 0.0380, 0.4133, 0.4176, 0.2154, 0.2739, 0.1326], device='cuda:0'), in_proj_covar=tensor([0.0363, 0.0290, 0.0276, 0.0250, 0.0342, 0.0336, 0.0260, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 05:22:03,774 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96226.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:22:38,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.174e+02 1.936e+02 2.176e+02 2.702e+02 4.909e+02, threshold=4.353e+02, percent-clipped=2.0 2023-03-09 05:22:59,839 INFO [train2.py:809] (0/4) Epoch 25, batch 650, loss[ctc_loss=0.09037, att_loss=0.2526, loss=0.2201, over 13771.00 frames. utt_duration=381.5 frames, utt_pad_proportion=0.3377, over 145.00 utterances.], tot_loss[ctc_loss=0.06973, att_loss=0.2335, loss=0.2008, over 3140748.35 frames. utt_duration=1202 frames, utt_pad_proportion=0.06832, over 10464.79 utterances.], batch size: 145, lr: 4.33e-03, grad_scale: 16.0 2023-03-09 05:23:19,238 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96273.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:23:41,796 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0886, 6.3089, 5.8579, 6.0291, 6.0402, 5.4557, 5.8048, 5.5007], device='cuda:0'), covar=tensor([0.1390, 0.0861, 0.0977, 0.0777, 0.0785, 0.1470, 0.2331, 0.2060], device='cuda:0'), in_proj_covar=tensor([0.0540, 0.0624, 0.0474, 0.0465, 0.0439, 0.0479, 0.0627, 0.0539], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:24:19,877 INFO [train2.py:809] (0/4) Epoch 25, batch 700, loss[ctc_loss=0.0575, att_loss=0.2156, loss=0.184, over 16015.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.0068, over 40.00 utterances.], tot_loss[ctc_loss=0.06878, att_loss=0.2326, loss=0.1998, over 3169193.81 frames. utt_duration=1230 frames, utt_pad_proportion=0.06026, over 10318.15 utterances.], batch size: 40, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:24:36,200 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96321.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:24:40,919 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96324.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:24:52,081 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-09 05:24:59,803 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96336.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:25:18,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 1.804e+02 2.150e+02 2.706e+02 4.503e+02, threshold=4.300e+02, percent-clipped=0.0 2023-03-09 05:25:23,455 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.22 vs. limit=5.0 2023-03-09 05:25:38,745 INFO [train2.py:809] (0/4) Epoch 25, batch 750, loss[ctc_loss=0.05405, att_loss=0.2254, loss=0.1911, over 16536.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006362, over 45.00 utterances.], tot_loss[ctc_loss=0.06921, att_loss=0.233, loss=0.2003, over 3191592.86 frames. utt_duration=1240 frames, utt_pad_proportion=0.05723, over 10309.36 utterances.], batch size: 45, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:25:56,986 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5711, 3.3996, 3.3829, 3.7513, 2.5395, 3.7275, 2.7569, 2.0890], device='cuda:0'), covar=tensor([0.0528, 0.0455, 0.0901, 0.0372, 0.1718, 0.0320, 0.1332, 0.1550], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0176, 0.0264, 0.0170, 0.0224, 0.0161, 0.0231, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:26:17,478 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96385.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:26:25,342 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96390.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:26:58,840 INFO [train2.py:809] (0/4) Epoch 25, batch 800, loss[ctc_loss=0.0681, att_loss=0.2411, loss=0.2065, over 16779.00 frames. utt_duration=1400 frames, utt_pad_proportion=0.005726, over 48.00 utterances.], tot_loss[ctc_loss=0.0695, att_loss=0.2333, loss=0.2006, over 3215316.38 frames. utt_duration=1249 frames, utt_pad_proportion=0.05304, over 10313.18 utterances.], batch size: 48, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:27:38,801 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96436.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:27:57,436 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 1.925e+02 2.269e+02 2.805e+02 5.385e+02, threshold=4.538e+02, percent-clipped=5.0 2023-03-09 05:28:02,607 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96451.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:28:19,548 INFO [train2.py:809] (0/4) Epoch 25, batch 850, loss[ctc_loss=0.05227, att_loss=0.2165, loss=0.1837, over 16167.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.005683, over 41.00 utterances.], tot_loss[ctc_loss=0.06925, att_loss=0.2332, loss=0.2004, over 3232760.63 frames. utt_duration=1252 frames, utt_pad_proportion=0.05086, over 10338.92 utterances.], batch size: 41, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:28:31,850 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96468.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 05:28:44,962 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7349, 5.2121, 5.2052, 5.1432, 5.2060, 5.2231, 4.8627, 4.6870], device='cuda:0'), covar=tensor([0.1536, 0.0633, 0.0392, 0.0730, 0.0435, 0.0408, 0.0467, 0.0406], device='cuda:0'), in_proj_covar=tensor([0.0537, 0.0376, 0.0364, 0.0377, 0.0437, 0.0444, 0.0372, 0.0407], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:29:28,022 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96503.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:29:41,365 INFO [train2.py:809] (0/4) Epoch 25, batch 900, loss[ctc_loss=0.05311, att_loss=0.2185, loss=0.1854, over 15773.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008338, over 38.00 utterances.], tot_loss[ctc_loss=0.0689, att_loss=0.2326, loss=0.1999, over 3240695.96 frames. utt_duration=1279 frames, utt_pad_proportion=0.04456, over 10143.83 utterances.], batch size: 38, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:29:49,280 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96516.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 05:29:57,194 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96521.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:30:40,144 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.401e+02 1.959e+02 2.352e+02 2.820e+02 9.770e+02, threshold=4.704e+02, percent-clipped=4.0 2023-03-09 05:30:45,064 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96551.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:31:03,282 INFO [train2.py:809] (0/4) Epoch 25, batch 950, loss[ctc_loss=0.07546, att_loss=0.2451, loss=0.2112, over 16871.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007288, over 49.00 utterances.], tot_loss[ctc_loss=0.06905, att_loss=0.2325, loss=0.1998, over 3241960.58 frames. utt_duration=1275 frames, utt_pad_proportion=0.0496, over 10183.28 utterances.], batch size: 49, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:31:52,108 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9385, 6.2023, 5.6514, 5.9150, 5.8684, 5.2898, 5.6177, 5.4005], device='cuda:0'), covar=tensor([0.1215, 0.0830, 0.0807, 0.0774, 0.0813, 0.1487, 0.2184, 0.2181], device='cuda:0'), in_proj_covar=tensor([0.0544, 0.0625, 0.0476, 0.0468, 0.0443, 0.0480, 0.0627, 0.0541], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:32:04,848 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0196, 5.2786, 5.2029, 5.1507, 5.2943, 5.3001, 4.9122, 4.7697], device='cuda:0'), covar=tensor([0.1035, 0.0516, 0.0316, 0.0483, 0.0290, 0.0338, 0.0417, 0.0327], device='cuda:0'), in_proj_covar=tensor([0.0538, 0.0376, 0.0365, 0.0377, 0.0437, 0.0444, 0.0372, 0.0406], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:32:25,899 INFO [train2.py:809] (0/4) Epoch 25, batch 1000, loss[ctc_loss=0.09879, att_loss=0.2484, loss=0.2185, over 13967.00 frames. utt_duration=386.9 frames, utt_pad_proportion=0.3271, over 145.00 utterances.], tot_loss[ctc_loss=0.06969, att_loss=0.2329, loss=0.2002, over 3238541.61 frames. utt_duration=1245 frames, utt_pad_proportion=0.0587, over 10418.71 utterances.], batch size: 145, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:32:38,267 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-09 05:33:05,490 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96636.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:33:23,585 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 1.809e+02 2.155e+02 2.476e+02 5.052e+02, threshold=4.310e+02, percent-clipped=2.0 2023-03-09 05:33:46,050 INFO [train2.py:809] (0/4) Epoch 25, batch 1050, loss[ctc_loss=0.06719, att_loss=0.2208, loss=0.1901, over 15868.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.009913, over 39.00 utterances.], tot_loss[ctc_loss=0.06964, att_loss=0.2334, loss=0.2006, over 3253997.43 frames. utt_duration=1260 frames, utt_pad_proportion=0.05191, over 10341.76 utterances.], batch size: 39, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:34:17,049 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96680.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:34:23,043 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96684.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:35:06,970 INFO [train2.py:809] (0/4) Epoch 25, batch 1100, loss[ctc_loss=0.07041, att_loss=0.2188, loss=0.1891, over 15368.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01143, over 35.00 utterances.], tot_loss[ctc_loss=0.06994, att_loss=0.2336, loss=0.2009, over 3255844.02 frames. utt_duration=1234 frames, utt_pad_proportion=0.05867, over 10562.54 utterances.], batch size: 35, lr: 4.32e-03, grad_scale: 16.0 2023-03-09 05:35:15,968 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.90 vs. limit=2.0 2023-03-09 05:35:47,555 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96736.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:36:03,164 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=96746.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:36:06,425 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 2.003e+02 2.274e+02 2.730e+02 5.677e+02, threshold=4.548e+02, percent-clipped=3.0 2023-03-09 05:36:28,281 INFO [train2.py:809] (0/4) Epoch 25, batch 1150, loss[ctc_loss=0.06016, att_loss=0.2285, loss=0.1949, over 16540.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005706, over 45.00 utterances.], tot_loss[ctc_loss=0.06977, att_loss=0.2336, loss=0.2009, over 3256000.51 frames. utt_duration=1256 frames, utt_pad_proportion=0.05352, over 10383.34 utterances.], batch size: 45, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:36:52,962 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=96776.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:37:05,012 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96784.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:37:48,669 INFO [train2.py:809] (0/4) Epoch 25, batch 1200, loss[ctc_loss=0.05794, att_loss=0.2175, loss=0.1856, over 16541.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005556, over 45.00 utterances.], tot_loss[ctc_loss=0.06996, att_loss=0.2341, loss=0.2013, over 3255164.37 frames. utt_duration=1224 frames, utt_pad_proportion=0.06149, over 10646.85 utterances.], batch size: 45, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:38:05,765 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96821.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:38:30,761 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=96837.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:38:48,487 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.354e+02 2.088e+02 2.358e+02 2.925e+02 6.715e+02, threshold=4.715e+02, percent-clipped=3.0 2023-03-09 05:38:59,887 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.2421, 5.4990, 5.7627, 5.6415, 5.7956, 6.1930, 5.4109, 6.2499], device='cuda:0'), covar=tensor([0.0702, 0.0691, 0.0769, 0.1344, 0.1734, 0.0888, 0.0646, 0.0649], device='cuda:0'), in_proj_covar=tensor([0.0905, 0.0523, 0.0632, 0.0677, 0.0906, 0.0655, 0.0515, 0.0642], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:39:08,867 INFO [train2.py:809] (0/4) Epoch 25, batch 1250, loss[ctc_loss=0.06216, att_loss=0.2235, loss=0.1912, over 16128.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006218, over 42.00 utterances.], tot_loss[ctc_loss=0.06953, att_loss=0.234, loss=0.2011, over 3256353.98 frames. utt_duration=1209 frames, utt_pad_proportion=0.06537, over 10785.93 utterances.], batch size: 42, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:39:22,085 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=96869.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:40:05,599 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7072, 4.9690, 4.5760, 5.0436, 4.4063, 4.6851, 5.0986, 4.8828], device='cuda:0'), covar=tensor([0.0615, 0.0358, 0.0776, 0.0343, 0.0425, 0.0289, 0.0230, 0.0220], device='cuda:0'), in_proj_covar=tensor([0.0399, 0.0331, 0.0372, 0.0362, 0.0332, 0.0243, 0.0314, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 05:40:28,112 INFO [train2.py:809] (0/4) Epoch 25, batch 1300, loss[ctc_loss=0.06948, att_loss=0.2147, loss=0.1857, over 15626.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.00931, over 37.00 utterances.], tot_loss[ctc_loss=0.06882, att_loss=0.2331, loss=0.2002, over 3251593.67 frames. utt_duration=1244 frames, utt_pad_proportion=0.05736, over 10464.90 utterances.], batch size: 37, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:41:27,105 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.103e+02 1.796e+02 2.102e+02 2.487e+02 3.780e+02, threshold=4.203e+02, percent-clipped=0.0 2023-03-09 05:41:47,823 INFO [train2.py:809] (0/4) Epoch 25, batch 1350, loss[ctc_loss=0.05686, att_loss=0.2387, loss=0.2023, over 17393.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.0472, over 69.00 utterances.], tot_loss[ctc_loss=0.0684, att_loss=0.2332, loss=0.2002, over 3264768.75 frames. utt_duration=1214 frames, utt_pad_proportion=0.06064, over 10766.86 utterances.], batch size: 69, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:42:17,989 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=96980.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:42:51,039 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97000.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:43:02,343 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-09 05:43:07,411 INFO [train2.py:809] (0/4) Epoch 25, batch 1400, loss[ctc_loss=0.06486, att_loss=0.23, loss=0.197, over 16127.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.004978, over 42.00 utterances.], tot_loss[ctc_loss=0.06841, att_loss=0.2328, loss=0.1999, over 3260854.89 frames. utt_duration=1201 frames, utt_pad_proportion=0.06393, over 10869.97 utterances.], batch size: 42, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:43:22,717 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 05:43:23,430 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9484, 3.9571, 3.9343, 4.0136, 4.1216, 4.1047, 3.8396, 3.7722], device='cuda:0'), covar=tensor([0.1041, 0.0916, 0.1513, 0.0637, 0.0371, 0.0471, 0.0534, 0.0437], device='cuda:0'), in_proj_covar=tensor([0.0540, 0.0379, 0.0368, 0.0378, 0.0439, 0.0447, 0.0374, 0.0410], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 05:43:33,985 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97028.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:44:02,788 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97046.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:44:05,387 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 1.953e+02 2.309e+02 2.789e+02 6.785e+02, threshold=4.619e+02, percent-clipped=7.0 2023-03-09 05:44:26,317 INFO [train2.py:809] (0/4) Epoch 25, batch 1450, loss[ctc_loss=0.06715, att_loss=0.2167, loss=0.1867, over 15358.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01092, over 35.00 utterances.], tot_loss[ctc_loss=0.06901, att_loss=0.2338, loss=0.2008, over 3267381.44 frames. utt_duration=1207 frames, utt_pad_proportion=0.06236, over 10843.70 utterances.], batch size: 35, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:44:26,688 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97061.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:44:26,787 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9931, 5.1165, 4.9539, 2.3283, 2.1100, 2.9200, 2.3649, 3.9308], device='cuda:0'), covar=tensor([0.0782, 0.0285, 0.0256, 0.5077, 0.5452, 0.2521, 0.4143, 0.1621], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0289, 0.0273, 0.0248, 0.0337, 0.0333, 0.0260, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 05:44:53,345 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4065, 2.6462, 4.9418, 3.7965, 2.8055, 4.2450, 4.6502, 4.6408], device='cuda:0'), covar=tensor([0.0308, 0.1567, 0.0184, 0.0938, 0.1889, 0.0297, 0.0263, 0.0280], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0245, 0.0207, 0.0321, 0.0267, 0.0227, 0.0197, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:45:17,665 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97094.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:45:37,672 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.84 vs. limit=2.0 2023-03-09 05:45:44,472 INFO [train2.py:809] (0/4) Epoch 25, batch 1500, loss[ctc_loss=0.06367, att_loss=0.2069, loss=0.1782, over 15772.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008714, over 38.00 utterances.], tot_loss[ctc_loss=0.06854, att_loss=0.2329, loss=0.2, over 3261239.23 frames. utt_duration=1219 frames, utt_pad_proportion=0.06169, over 10711.94 utterances.], batch size: 38, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:46:09,790 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4285, 2.7516, 4.7965, 3.7728, 2.8055, 4.2337, 4.4456, 4.5179], device='cuda:0'), covar=tensor([0.0259, 0.1473, 0.0226, 0.0878, 0.1768, 0.0263, 0.0236, 0.0273], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0245, 0.0208, 0.0321, 0.0267, 0.0227, 0.0198, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:46:17,400 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97132.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:46:19,093 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9546, 4.2710, 4.2883, 4.4681, 2.6137, 4.3414, 2.8121, 1.6187], device='cuda:0'), covar=tensor([0.0521, 0.0281, 0.0698, 0.0263, 0.1684, 0.0233, 0.1417, 0.1721], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0177, 0.0264, 0.0171, 0.0223, 0.0162, 0.0232, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:46:19,865 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-09 05:46:43,660 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.190e+02 2.011e+02 2.273e+02 2.604e+02 5.523e+02, threshold=4.546e+02, percent-clipped=2.0 2023-03-09 05:46:51,788 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7770, 4.4026, 4.4202, 2.3301, 2.0647, 2.8345, 2.4166, 3.6412], device='cuda:0'), covar=tensor([0.0760, 0.0327, 0.0262, 0.4517, 0.5262, 0.2562, 0.3508, 0.1437], device='cuda:0'), in_proj_covar=tensor([0.0357, 0.0287, 0.0271, 0.0246, 0.0335, 0.0330, 0.0258, 0.0368], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 05:47:03,716 INFO [train2.py:809] (0/4) Epoch 25, batch 1550, loss[ctc_loss=0.05152, att_loss=0.233, loss=0.1967, over 16535.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006051, over 45.00 utterances.], tot_loss[ctc_loss=0.06769, att_loss=0.2326, loss=0.1996, over 3267886.10 frames. utt_duration=1249 frames, utt_pad_proportion=0.05362, over 10482.29 utterances.], batch size: 45, lr: 4.31e-03, grad_scale: 16.0 2023-03-09 05:47:08,220 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.03 vs. limit=5.0 2023-03-09 05:47:11,030 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.73 vs. limit=2.0 2023-03-09 05:48:24,492 INFO [train2.py:809] (0/4) Epoch 25, batch 1600, loss[ctc_loss=0.0689, att_loss=0.2507, loss=0.2144, over 17016.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.007977, over 51.00 utterances.], tot_loss[ctc_loss=0.06842, att_loss=0.2335, loss=0.2005, over 3271352.17 frames. utt_duration=1242 frames, utt_pad_proportion=0.0544, over 10549.10 utterances.], batch size: 51, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:48:55,413 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1239, 4.4981, 4.7173, 4.7772, 3.0506, 4.5965, 2.8484, 1.7015], device='cuda:0'), covar=tensor([0.0450, 0.0246, 0.0521, 0.0201, 0.1375, 0.0211, 0.1383, 0.1692], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0177, 0.0262, 0.0170, 0.0222, 0.0161, 0.0231, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:49:22,640 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 1.951e+02 2.275e+02 2.852e+02 6.769e+02, threshold=4.549e+02, percent-clipped=7.0 2023-03-09 05:49:41,699 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97260.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:49:43,460 INFO [train2.py:809] (0/4) Epoch 25, batch 1650, loss[ctc_loss=0.08604, att_loss=0.2598, loss=0.225, over 17437.00 frames. utt_duration=1012 frames, utt_pad_proportion=0.0458, over 69.00 utterances.], tot_loss[ctc_loss=0.06845, att_loss=0.2331, loss=0.2002, over 3272478.65 frames. utt_duration=1248 frames, utt_pad_proportion=0.053, over 10500.85 utterances.], batch size: 69, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:49:45,395 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3187, 4.0069, 3.5581, 3.7088, 4.2311, 3.8382, 3.4639, 4.5331], device='cuda:0'), covar=tensor([0.0877, 0.0454, 0.0874, 0.0572, 0.0599, 0.0635, 0.0692, 0.0405], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0223, 0.0227, 0.0205, 0.0286, 0.0244, 0.0201, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 05:49:50,066 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9169, 5.1681, 5.3992, 5.2772, 5.4145, 5.8152, 5.1407, 5.9717], device='cuda:0'), covar=tensor([0.0662, 0.0712, 0.0912, 0.1186, 0.1726, 0.1024, 0.0793, 0.0619], device='cuda:0'), in_proj_covar=tensor([0.0904, 0.0523, 0.0634, 0.0679, 0.0907, 0.0656, 0.0514, 0.0643], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 05:51:02,812 INFO [train2.py:809] (0/4) Epoch 25, batch 1700, loss[ctc_loss=0.07046, att_loss=0.2191, loss=0.1893, over 16004.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.008019, over 40.00 utterances.], tot_loss[ctc_loss=0.06771, att_loss=0.2323, loss=0.1994, over 3270993.58 frames. utt_duration=1247 frames, utt_pad_proportion=0.05327, over 10504.79 utterances.], batch size: 40, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:51:08,871 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-03-09 05:51:19,391 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97321.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:51:39,601 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0810, 3.7961, 3.7375, 3.2055, 3.8865, 3.8581, 3.8320, 2.7816], device='cuda:0'), covar=tensor([0.0927, 0.1156, 0.1931, 0.2554, 0.0657, 0.2071, 0.0774, 0.2755], device='cuda:0'), in_proj_covar=tensor([0.0192, 0.0199, 0.0213, 0.0268, 0.0175, 0.0276, 0.0199, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:51:51,957 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97342.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:52:00,348 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97347.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:52:02,098 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 1.753e+02 2.134e+02 2.732e+02 7.784e+02, threshold=4.268e+02, percent-clipped=3.0 2023-03-09 05:52:15,786 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97356.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:52:23,283 INFO [train2.py:809] (0/4) Epoch 25, batch 1750, loss[ctc_loss=0.07399, att_loss=0.2143, loss=0.1863, over 14104.00 frames. utt_duration=1821 frames, utt_pad_proportion=0.05725, over 31.00 utterances.], tot_loss[ctc_loss=0.06704, att_loss=0.2314, loss=0.1985, over 3266415.84 frames. utt_duration=1260 frames, utt_pad_proportion=0.05299, over 10385.12 utterances.], batch size: 31, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:52:44,581 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97374.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:53:31,683 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97403.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 05:53:34,210 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-09 05:53:39,251 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97408.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:53:44,075 INFO [train2.py:809] (0/4) Epoch 25, batch 1800, loss[ctc_loss=0.0604, att_loss=0.2402, loss=0.2043, over 16620.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005827, over 47.00 utterances.], tot_loss[ctc_loss=0.067, att_loss=0.2317, loss=0.1988, over 3270140.63 frames. utt_duration=1243 frames, utt_pad_proportion=0.05536, over 10532.21 utterances.], batch size: 47, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:54:03,213 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7820, 2.4266, 2.6276, 3.3131, 3.0508, 3.2124, 2.5491, 2.3229], device='cuda:0'), covar=tensor([0.0836, 0.1866, 0.0908, 0.0748, 0.1126, 0.0565, 0.1388, 0.1726], device='cuda:0'), in_proj_covar=tensor([0.0184, 0.0220, 0.0187, 0.0221, 0.0231, 0.0185, 0.0206, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 05:54:14,433 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 05:54:16,733 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97432.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:54:21,370 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97435.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:54:43,145 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 1.842e+02 2.140e+02 2.481e+02 6.088e+02, threshold=4.279e+02, percent-clipped=3.0 2023-03-09 05:54:56,591 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.44 vs. limit=2.0 2023-03-09 05:55:03,728 INFO [train2.py:809] (0/4) Epoch 25, batch 1850, loss[ctc_loss=0.07359, att_loss=0.2483, loss=0.2133, over 17559.00 frames. utt_duration=1019 frames, utt_pad_proportion=0.0385, over 69.00 utterances.], tot_loss[ctc_loss=0.0673, att_loss=0.232, loss=0.1991, over 3266189.84 frames. utt_duration=1214 frames, utt_pad_proportion=0.06493, over 10772.59 utterances.], batch size: 69, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:55:33,337 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97480.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:56:15,207 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-03-09 05:56:23,870 INFO [train2.py:809] (0/4) Epoch 25, batch 1900, loss[ctc_loss=0.06697, att_loss=0.2134, loss=0.1841, over 15778.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008086, over 38.00 utterances.], tot_loss[ctc_loss=0.06786, att_loss=0.2327, loss=0.1997, over 3266525.93 frames. utt_duration=1210 frames, utt_pad_proportion=0.06658, over 10814.45 utterances.], batch size: 38, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:57:23,544 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.351e+02 1.849e+02 2.257e+02 2.657e+02 5.129e+02, threshold=4.513e+02, percent-clipped=1.0 2023-03-09 05:57:35,255 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97555.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:57:44,860 INFO [train2.py:809] (0/4) Epoch 25, batch 1950, loss[ctc_loss=0.07621, att_loss=0.2335, loss=0.2021, over 16398.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007888, over 44.00 utterances.], tot_loss[ctc_loss=0.06748, att_loss=0.2323, loss=0.1994, over 3274212.96 frames. utt_duration=1223 frames, utt_pad_proportion=0.05958, over 10720.60 utterances.], batch size: 44, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:58:08,839 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2422, 3.6968, 3.2104, 3.4532, 4.0258, 3.5686, 3.0996, 4.3333], device='cuda:0'), covar=tensor([0.0897, 0.0524, 0.1201, 0.0725, 0.0738, 0.0772, 0.0864, 0.0434], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0226, 0.0230, 0.0208, 0.0290, 0.0248, 0.0205, 0.0297], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 05:58:29,321 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-09 05:58:49,638 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.79 vs. limit=2.0 2023-03-09 05:59:06,992 INFO [train2.py:809] (0/4) Epoch 25, batch 2000, loss[ctc_loss=0.0833, att_loss=0.2493, loss=0.2161, over 17030.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.01095, over 53.00 utterances.], tot_loss[ctc_loss=0.06741, att_loss=0.2326, loss=0.1995, over 3265298.28 frames. utt_duration=1212 frames, utt_pad_proportion=0.06566, over 10792.30 utterances.], batch size: 53, lr: 4.30e-03, grad_scale: 16.0 2023-03-09 05:59:15,327 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 05:59:15,519 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97616.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:00:08,129 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.275e+02 1.752e+02 2.147e+02 2.537e+02 5.924e+02, threshold=4.294e+02, percent-clipped=3.0 2023-03-09 06:00:21,288 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97656.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:00:29,808 INFO [train2.py:809] (0/4) Epoch 25, batch 2050, loss[ctc_loss=0.07891, att_loss=0.244, loss=0.211, over 17246.00 frames. utt_duration=874.9 frames, utt_pad_proportion=0.08294, over 79.00 utterances.], tot_loss[ctc_loss=0.06761, att_loss=0.2333, loss=0.2002, over 3276649.25 frames. utt_duration=1204 frames, utt_pad_proportion=0.06436, over 10895.59 utterances.], batch size: 79, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:00:50,821 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97674.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 06:01:10,337 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7901, 3.3697, 3.8170, 3.3181, 3.6605, 4.8233, 4.6778, 3.6965], device='cuda:0'), covar=tensor([0.0298, 0.1516, 0.1183, 0.1335, 0.1104, 0.0838, 0.0520, 0.1006], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0248, 0.0287, 0.0223, 0.0267, 0.0376, 0.0267, 0.0233], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:01:13,592 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5437, 2.2510, 2.3628, 2.4912, 2.7780, 2.3408, 2.2007, 2.8434], device='cuda:0'), covar=tensor([0.1374, 0.2614, 0.2095, 0.1321, 0.1472, 0.1437, 0.2197, 0.1020], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0134, 0.0130, 0.0123, 0.0139, 0.0121, 0.0144, 0.0118], device='cuda:0'), out_proj_covar=tensor([1.0081e-04, 1.0598e-04, 1.0562e-04, 9.6377e-05, 1.0502e-04, 9.7642e-05, 1.0947e-04, 9.4017e-05], device='cuda:0') 2023-03-09 06:01:22,250 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1940, 5.4544, 5.4921, 5.4093, 5.5240, 5.4411, 5.1511, 4.9459], device='cuda:0'), covar=tensor([0.1027, 0.0518, 0.0242, 0.0429, 0.0245, 0.0305, 0.0391, 0.0292], device='cuda:0'), in_proj_covar=tensor([0.0543, 0.0382, 0.0370, 0.0377, 0.0439, 0.0445, 0.0376, 0.0413], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:01:30,759 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97698.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 06:01:33,947 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9086, 5.0313, 4.5552, 2.5496, 4.7927, 4.7199, 3.9759, 2.1623], device='cuda:0'), covar=tensor([0.0223, 0.0156, 0.0444, 0.1591, 0.0150, 0.0265, 0.0592, 0.2647], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0105, 0.0107, 0.0111, 0.0087, 0.0115, 0.0100, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:01:38,579 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97703.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:01:39,984 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97704.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:01:51,389 INFO [train2.py:809] (0/4) Epoch 25, batch 2100, loss[ctc_loss=0.06167, att_loss=0.2179, loss=0.1867, over 15505.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008514, over 36.00 utterances.], tot_loss[ctc_loss=0.06741, att_loss=0.2331, loss=0.2, over 3282597.24 frames. utt_duration=1210 frames, utt_pad_proportion=0.06089, over 10860.62 utterances.], batch size: 36, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:02:20,936 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97730.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:02:29,459 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97735.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 06:02:50,718 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 1.928e+02 2.285e+02 2.765e+02 5.128e+02, threshold=4.569e+02, percent-clipped=4.0 2023-03-09 06:03:09,857 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3810, 4.4796, 4.1111, 2.5665, 4.2881, 4.2493, 3.7597, 2.3622], device='cuda:0'), covar=tensor([0.0177, 0.0159, 0.0392, 0.1409, 0.0150, 0.0328, 0.0511, 0.2174], device='cuda:0'), in_proj_covar=tensor([0.0076, 0.0104, 0.0106, 0.0111, 0.0087, 0.0115, 0.0100, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:03:11,120 INFO [train2.py:809] (0/4) Epoch 25, batch 2150, loss[ctc_loss=0.07265, att_loss=0.239, loss=0.2057, over 16835.00 frames. utt_duration=681.6 frames, utt_pad_proportion=0.1458, over 99.00 utterances.], tot_loss[ctc_loss=0.06766, att_loss=0.233, loss=0.1999, over 3285727.27 frames. utt_duration=1216 frames, utt_pad_proportion=0.05832, over 10825.86 utterances.], batch size: 99, lr: 4.29e-03, grad_scale: 32.0 2023-03-09 06:03:30,861 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.83 vs. limit=2.0 2023-03-09 06:03:35,211 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-09 06:03:46,525 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6591, 3.0609, 3.7083, 3.0790, 3.5061, 4.7295, 4.4960, 3.4480], device='cuda:0'), covar=tensor([0.0335, 0.1832, 0.1411, 0.1464, 0.1249, 0.0733, 0.0604, 0.1142], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0249, 0.0288, 0.0223, 0.0268, 0.0376, 0.0268, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:04:32,648 INFO [train2.py:809] (0/4) Epoch 25, batch 2200, loss[ctc_loss=0.05634, att_loss=0.2066, loss=0.1765, over 15369.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01118, over 35.00 utterances.], tot_loss[ctc_loss=0.06648, att_loss=0.2316, loss=0.1986, over 3274066.89 frames. utt_duration=1240 frames, utt_pad_proportion=0.05628, over 10577.99 utterances.], batch size: 35, lr: 4.29e-03, grad_scale: 32.0 2023-03-09 06:05:34,922 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 1.901e+02 2.356e+02 2.897e+02 7.922e+02, threshold=4.712e+02, percent-clipped=5.0 2023-03-09 06:05:44,057 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 06:05:53,830 INFO [train2.py:809] (0/4) Epoch 25, batch 2250, loss[ctc_loss=0.06019, att_loss=0.2283, loss=0.1946, over 17354.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03542, over 63.00 utterances.], tot_loss[ctc_loss=0.06726, att_loss=0.2326, loss=0.1995, over 3274766.10 frames. utt_duration=1224 frames, utt_pad_proportion=0.05998, over 10714.50 utterances.], batch size: 63, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:06:37,479 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=97888.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:06:49,103 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-09 06:07:02,676 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0110, 3.7343, 3.6138, 3.1610, 3.6757, 3.7843, 3.7335, 2.7527], device='cuda:0'), covar=tensor([0.1010, 0.1241, 0.1737, 0.3378, 0.2338, 0.1656, 0.0831, 0.3210], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0197, 0.0210, 0.0264, 0.0174, 0.0269, 0.0195, 0.0221], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:07:13,165 INFO [train2.py:809] (0/4) Epoch 25, batch 2300, loss[ctc_loss=0.06133, att_loss=0.2376, loss=0.2024, over 17119.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01491, over 56.00 utterances.], tot_loss[ctc_loss=0.06723, att_loss=0.232, loss=0.199, over 3273036.54 frames. utt_duration=1236 frames, utt_pad_proportion=0.05693, over 10604.68 utterances.], batch size: 56, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:07:13,323 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=97911.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:07:21,267 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97916.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:08:14,120 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.169e+02 1.837e+02 2.120e+02 2.605e+02 6.957e+02, threshold=4.241e+02, percent-clipped=1.0 2023-03-09 06:08:14,585 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=97949.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:08:33,497 INFO [train2.py:809] (0/4) Epoch 25, batch 2350, loss[ctc_loss=0.05775, att_loss=0.2408, loss=0.2042, over 16971.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007159, over 50.00 utterances.], tot_loss[ctc_loss=0.06737, att_loss=0.2322, loss=0.1992, over 3275429.29 frames. utt_duration=1238 frames, utt_pad_proportion=0.05662, over 10599.44 utterances.], batch size: 50, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:08:38,220 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=97964.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:08:59,924 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.16 vs. limit=5.0 2023-03-09 06:09:07,747 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1801, 5.5080, 5.0843, 5.5526, 4.9840, 5.1065, 5.6295, 5.4190], device='cuda:0'), covar=tensor([0.0591, 0.0310, 0.0715, 0.0353, 0.0366, 0.0229, 0.0211, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0399, 0.0332, 0.0374, 0.0366, 0.0333, 0.0244, 0.0316, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 06:09:34,071 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=97998.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:09:37,263 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-98000.pt 2023-03-09 06:09:45,797 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98003.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:09:57,923 INFO [train2.py:809] (0/4) Epoch 25, batch 2400, loss[ctc_loss=0.05894, att_loss=0.2504, loss=0.2121, over 17068.00 frames. utt_duration=1290 frames, utt_pad_proportion=0.008803, over 53.00 utterances.], tot_loss[ctc_loss=0.06711, att_loss=0.2326, loss=0.1995, over 3277204.47 frames. utt_duration=1229 frames, utt_pad_proportion=0.059, over 10675.36 utterances.], batch size: 53, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:10:00,896 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-03-09 06:10:13,236 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3329, 5.1970, 4.9491, 3.0759, 5.0835, 4.9350, 4.6389, 2.8241], device='cuda:0'), covar=tensor([0.0120, 0.0128, 0.0316, 0.1141, 0.0111, 0.0205, 0.0321, 0.1560], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0105, 0.0107, 0.0111, 0.0087, 0.0115, 0.0100, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:10:29,460 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98030.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 06:10:29,537 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98030.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:10:55,006 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98046.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:11:00,022 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.282e+02 1.784e+02 2.263e+02 2.686e+02 5.862e+02, threshold=4.525e+02, percent-clipped=1.0 2023-03-09 06:11:03,249 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98051.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:11:18,645 INFO [train2.py:809] (0/4) Epoch 25, batch 2450, loss[ctc_loss=0.0568, att_loss=0.2247, loss=0.1911, over 16533.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.005976, over 45.00 utterances.], tot_loss[ctc_loss=0.0669, att_loss=0.2325, loss=0.1994, over 3273214.01 frames. utt_duration=1248 frames, utt_pad_proportion=0.05554, over 10503.20 utterances.], batch size: 45, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:11:46,624 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98078.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:12:31,659 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98106.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:12:39,090 INFO [train2.py:809] (0/4) Epoch 25, batch 2500, loss[ctc_loss=0.06252, att_loss=0.211, loss=0.1813, over 15357.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.01156, over 35.00 utterances.], tot_loss[ctc_loss=0.06717, att_loss=0.2323, loss=0.1993, over 3253928.51 frames. utt_duration=1240 frames, utt_pad_proportion=0.06089, over 10508.02 utterances.], batch size: 35, lr: 4.29e-03, grad_scale: 16.0 2023-03-09 06:13:35,882 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5648, 2.1652, 2.2452, 2.4636, 2.5724, 2.1262, 2.2285, 2.7139], device='cuda:0'), covar=tensor([0.1640, 0.2436, 0.2069, 0.1358, 0.2037, 0.1658, 0.1975, 0.1204], device='cuda:0'), in_proj_covar=tensor([0.0131, 0.0135, 0.0131, 0.0124, 0.0140, 0.0121, 0.0142, 0.0119], device='cuda:0'), out_proj_covar=tensor([1.0090e-04, 1.0656e-04, 1.0632e-04, 9.6755e-05, 1.0543e-04, 9.7745e-05, 1.0877e-04, 9.4452e-05], device='cuda:0') 2023-03-09 06:13:40,618 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.428e+02 1.837e+02 2.156e+02 2.644e+02 5.272e+02, threshold=4.313e+02, percent-clipped=2.0 2023-03-09 06:13:59,452 INFO [train2.py:809] (0/4) Epoch 25, batch 2550, loss[ctc_loss=0.04919, att_loss=0.2002, loss=0.17, over 15498.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.008977, over 36.00 utterances.], tot_loss[ctc_loss=0.06732, att_loss=0.2326, loss=0.1995, over 3260346.06 frames. utt_duration=1240 frames, utt_pad_proportion=0.05963, over 10529.73 utterances.], batch size: 36, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:14:10,066 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98167.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:14:41,941 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9521, 5.2233, 4.7507, 5.2749, 4.6729, 4.8763, 5.3223, 5.1517], device='cuda:0'), covar=tensor([0.0556, 0.0302, 0.0831, 0.0323, 0.0420, 0.0274, 0.0268, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0395, 0.0330, 0.0372, 0.0363, 0.0331, 0.0243, 0.0313, 0.0292], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 06:15:20,410 INFO [train2.py:809] (0/4) Epoch 25, batch 2600, loss[ctc_loss=0.07663, att_loss=0.2387, loss=0.2063, over 17351.00 frames. utt_duration=1007 frames, utt_pad_proportion=0.05075, over 69.00 utterances.], tot_loss[ctc_loss=0.06806, att_loss=0.2332, loss=0.2002, over 3258528.54 frames. utt_duration=1229 frames, utt_pad_proportion=0.05892, over 10617.25 utterances.], batch size: 69, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:15:20,709 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98211.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:16:14,189 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98244.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:16:21,476 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 1.793e+02 2.177e+02 2.680e+02 7.649e+02, threshold=4.354e+02, percent-clipped=6.0 2023-03-09 06:16:37,134 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98259.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:16:40,775 INFO [train2.py:809] (0/4) Epoch 25, batch 2650, loss[ctc_loss=0.07928, att_loss=0.2413, loss=0.2089, over 17058.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008766, over 52.00 utterances.], tot_loss[ctc_loss=0.06841, att_loss=0.2333, loss=0.2003, over 3260072.71 frames. utt_duration=1188 frames, utt_pad_proportion=0.06964, over 10992.49 utterances.], batch size: 52, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:18:00,623 INFO [train2.py:809] (0/4) Epoch 25, batch 2700, loss[ctc_loss=0.06844, att_loss=0.2291, loss=0.197, over 16399.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007661, over 44.00 utterances.], tot_loss[ctc_loss=0.06757, att_loss=0.2327, loss=0.1997, over 3266335.04 frames. utt_duration=1224 frames, utt_pad_proportion=0.05987, over 10686.49 utterances.], batch size: 44, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:18:07,842 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.26 vs. limit=2.0 2023-03-09 06:18:31,179 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98330.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 06:19:00,610 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.296e+02 1.887e+02 2.414e+02 2.992e+02 7.902e+02, threshold=4.828e+02, percent-clipped=2.0 2023-03-09 06:19:19,945 INFO [train2.py:809] (0/4) Epoch 25, batch 2750, loss[ctc_loss=0.06889, att_loss=0.2355, loss=0.2022, over 17118.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.0152, over 56.00 utterances.], tot_loss[ctc_loss=0.06792, att_loss=0.2331, loss=0.2, over 3271161.57 frames. utt_duration=1213 frames, utt_pad_proportion=0.0624, over 10797.42 utterances.], batch size: 56, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:19:47,048 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98378.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 06:19:56,173 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8792, 6.1788, 5.6465, 5.8920, 5.8700, 5.3654, 5.5957, 5.3045], device='cuda:0'), covar=tensor([0.1306, 0.0953, 0.1044, 0.0804, 0.0768, 0.1610, 0.2214, 0.2290], device='cuda:0'), in_proj_covar=tensor([0.0543, 0.0631, 0.0478, 0.0466, 0.0435, 0.0482, 0.0628, 0.0539], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:20:37,058 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98410.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:20:38,133 INFO [train2.py:809] (0/4) Epoch 25, batch 2800, loss[ctc_loss=0.08043, att_loss=0.2514, loss=0.2172, over 17039.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.00886, over 53.00 utterances.], tot_loss[ctc_loss=0.06795, att_loss=0.2332, loss=0.2001, over 3265227.40 frames. utt_duration=1205 frames, utt_pad_proportion=0.06698, over 10853.22 utterances.], batch size: 53, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:21:09,921 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6193, 2.7886, 3.6635, 2.8511, 3.5181, 4.7289, 4.5666, 2.9464], device='cuda:0'), covar=tensor([0.0500, 0.2037, 0.1188, 0.1784, 0.1050, 0.0878, 0.0530, 0.1766], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0246, 0.0283, 0.0219, 0.0262, 0.0370, 0.0263, 0.0231], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:21:37,615 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.204e+02 1.960e+02 2.287e+02 2.906e+02 4.125e+02, threshold=4.574e+02, percent-clipped=0.0 2023-03-09 06:21:56,306 INFO [train2.py:809] (0/4) Epoch 25, batch 2850, loss[ctc_loss=0.07167, att_loss=0.245, loss=0.2103, over 17284.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01224, over 55.00 utterances.], tot_loss[ctc_loss=0.06642, att_loss=0.2318, loss=0.1987, over 3267406.19 frames. utt_duration=1237 frames, utt_pad_proportion=0.05872, over 10578.83 utterances.], batch size: 55, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:21:57,967 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98462.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:22:04,224 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0056, 4.4306, 4.9505, 4.6348, 2.8383, 4.3981, 3.2998, 1.7452], device='cuda:0'), covar=tensor([0.0518, 0.0322, 0.0477, 0.0244, 0.1527, 0.0242, 0.1126, 0.1749], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0179, 0.0265, 0.0174, 0.0225, 0.0165, 0.0233, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:22:05,020 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-09 06:22:12,351 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98471.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:23:14,285 INFO [train2.py:809] (0/4) Epoch 25, batch 2900, loss[ctc_loss=0.0752, att_loss=0.2436, loss=0.2099, over 16961.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007991, over 50.00 utterances.], tot_loss[ctc_loss=0.06673, att_loss=0.2319, loss=0.1988, over 3274094.65 frames. utt_duration=1253 frames, utt_pad_proportion=0.05259, over 10460.70 utterances.], batch size: 50, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:23:45,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-09 06:23:52,875 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4053, 2.5279, 4.7903, 3.7584, 3.0545, 4.1419, 4.4399, 4.5234], device='cuda:0'), covar=tensor([0.0249, 0.1585, 0.0175, 0.0914, 0.1622, 0.0288, 0.0228, 0.0262], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0243, 0.0209, 0.0320, 0.0265, 0.0229, 0.0199, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:24:06,278 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98543.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:24:07,630 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98544.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:24:14,934 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.223e+02 1.817e+02 2.098e+02 2.346e+02 4.343e+02, threshold=4.197e+02, percent-clipped=0.0 2023-03-09 06:24:16,105 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-09 06:24:34,075 INFO [train2.py:809] (0/4) Epoch 25, batch 2950, loss[ctc_loss=0.05767, att_loss=0.224, loss=0.1907, over 16336.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.005867, over 45.00 utterances.], tot_loss[ctc_loss=0.06708, att_loss=0.2317, loss=0.1987, over 3270456.80 frames. utt_duration=1273 frames, utt_pad_proportion=0.0493, over 10292.47 utterances.], batch size: 45, lr: 4.28e-03, grad_scale: 16.0 2023-03-09 06:25:23,552 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98592.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:25:43,560 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98604.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:25:51,114 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9121, 6.1833, 5.6023, 5.8849, 5.8653, 5.3038, 5.6737, 5.3149], device='cuda:0'), covar=tensor([0.1344, 0.0820, 0.0914, 0.0786, 0.0804, 0.1618, 0.2164, 0.2518], device='cuda:0'), in_proj_covar=tensor([0.0542, 0.0627, 0.0476, 0.0466, 0.0434, 0.0482, 0.0626, 0.0537], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:25:54,029 INFO [train2.py:809] (0/4) Epoch 25, batch 3000, loss[ctc_loss=0.06625, att_loss=0.2131, loss=0.1837, over 15746.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.008573, over 38.00 utterances.], tot_loss[ctc_loss=0.06771, att_loss=0.2321, loss=0.1993, over 3265221.80 frames. utt_duration=1226 frames, utt_pad_proportion=0.06365, over 10667.89 utterances.], batch size: 38, lr: 4.27e-03, grad_scale: 16.0 2023-03-09 06:25:54,032 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 06:26:08,488 INFO [train2.py:843] (0/4) Epoch 25, validation: ctc_loss=0.04165, att_loss=0.235, loss=0.1963, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 06:26:08,489 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 06:27:09,501 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.233e+02 1.890e+02 2.327e+02 2.970e+02 6.667e+02, threshold=4.653e+02, percent-clipped=5.0 2023-03-09 06:27:28,838 INFO [train2.py:809] (0/4) Epoch 25, batch 3050, loss[ctc_loss=0.07505, att_loss=0.2466, loss=0.2123, over 17313.00 frames. utt_duration=1101 frames, utt_pad_proportion=0.03686, over 63.00 utterances.], tot_loss[ctc_loss=0.06729, att_loss=0.232, loss=0.199, over 3266657.86 frames. utt_duration=1226 frames, utt_pad_proportion=0.0628, over 10674.09 utterances.], batch size: 63, lr: 4.27e-03, grad_scale: 16.0 2023-03-09 06:27:43,792 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0886, 5.0430, 4.9224, 2.1174, 2.0107, 2.6709, 2.1898, 3.9206], device='cuda:0'), covar=tensor([0.0682, 0.0265, 0.0226, 0.4782, 0.5596, 0.2966, 0.4092, 0.1503], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0288, 0.0274, 0.0246, 0.0337, 0.0332, 0.0261, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 06:27:56,269 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0935, 5.5058, 5.6902, 5.4998, 5.6026, 6.0985, 5.3050, 6.1568], device='cuda:0'), covar=tensor([0.0667, 0.0654, 0.0805, 0.1256, 0.1743, 0.0786, 0.0767, 0.0599], device='cuda:0'), in_proj_covar=tensor([0.0912, 0.0522, 0.0637, 0.0680, 0.0910, 0.0662, 0.0517, 0.0636], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:28:48,628 INFO [train2.py:809] (0/4) Epoch 25, batch 3100, loss[ctc_loss=0.05546, att_loss=0.2163, loss=0.1841, over 16172.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006688, over 41.00 utterances.], tot_loss[ctc_loss=0.06756, att_loss=0.2321, loss=0.1992, over 3269375.92 frames. utt_duration=1219 frames, utt_pad_proportion=0.06326, over 10743.08 utterances.], batch size: 41, lr: 4.27e-03, grad_scale: 16.0 2023-03-09 06:29:43,660 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-03-09 06:29:48,739 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.374e+02 1.858e+02 2.243e+02 2.658e+02 4.552e+02, threshold=4.486e+02, percent-clipped=0.0 2023-03-09 06:30:07,764 INFO [train2.py:809] (0/4) Epoch 25, batch 3150, loss[ctc_loss=0.05709, att_loss=0.2204, loss=0.1877, over 15760.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009264, over 38.00 utterances.], tot_loss[ctc_loss=0.06825, att_loss=0.2323, loss=0.1995, over 3263649.70 frames. utt_duration=1217 frames, utt_pad_proportion=0.06377, over 10736.04 utterances.], batch size: 38, lr: 4.27e-03, grad_scale: 16.0 2023-03-09 06:30:09,727 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=98762.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:30:15,574 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98766.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:31:26,657 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=98810.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:31:28,125 INFO [train2.py:809] (0/4) Epoch 25, batch 3200, loss[ctc_loss=0.06051, att_loss=0.2123, loss=0.1819, over 15504.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.00845, over 36.00 utterances.], tot_loss[ctc_loss=0.06846, att_loss=0.2321, loss=0.1994, over 3258123.65 frames. utt_duration=1201 frames, utt_pad_proportion=0.07086, over 10866.77 utterances.], batch size: 36, lr: 4.27e-03, grad_scale: 8.0 2023-03-09 06:32:29,334 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.298e+02 1.857e+02 2.201e+02 2.576e+02 7.126e+02, threshold=4.402e+02, percent-clipped=1.0 2023-03-09 06:32:46,789 INFO [train2.py:809] (0/4) Epoch 25, batch 3250, loss[ctc_loss=0.05782, att_loss=0.2479, loss=0.2099, over 16882.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006656, over 49.00 utterances.], tot_loss[ctc_loss=0.06882, att_loss=0.2327, loss=0.2, over 3264713.01 frames. utt_duration=1202 frames, utt_pad_proportion=0.06846, over 10877.52 utterances.], batch size: 49, lr: 4.27e-03, grad_scale: 8.0 2023-03-09 06:33:33,441 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98890.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:33:41,362 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98895.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:33:47,311 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=98899.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:34:05,907 INFO [train2.py:809] (0/4) Epoch 25, batch 3300, loss[ctc_loss=0.07564, att_loss=0.2476, loss=0.2132, over 16870.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007361, over 49.00 utterances.], tot_loss[ctc_loss=0.06799, att_loss=0.2327, loss=0.1997, over 3271505.39 frames. utt_duration=1240 frames, utt_pad_proportion=0.05815, over 10562.57 utterances.], batch size: 49, lr: 4.27e-03, grad_scale: 8.0 2023-03-09 06:34:50,533 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8013, 3.5341, 3.3993, 3.0443, 3.5224, 3.5624, 3.5982, 2.6678], device='cuda:0'), covar=tensor([0.0923, 0.0887, 0.1559, 0.2880, 0.1192, 0.1367, 0.0860, 0.3101], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0201, 0.0214, 0.0268, 0.0177, 0.0276, 0.0199, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:35:02,963 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7837, 3.4033, 3.3256, 2.9601, 3.3917, 3.5049, 3.5069, 2.4150], device='cuda:0'), covar=tensor([0.1107, 0.1292, 0.2550, 0.3611, 0.1419, 0.1705, 0.0971, 0.3798], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0201, 0.0214, 0.0268, 0.0177, 0.0276, 0.0199, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:35:07,190 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 1.882e+02 2.265e+02 2.775e+02 6.610e+02, threshold=4.530e+02, percent-clipped=3.0 2023-03-09 06:35:09,220 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98951.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:35:17,001 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=98956.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:35:24,890 INFO [train2.py:809] (0/4) Epoch 25, batch 3350, loss[ctc_loss=0.0675, att_loss=0.2316, loss=0.1988, over 16615.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005767, over 47.00 utterances.], tot_loss[ctc_loss=0.06732, att_loss=0.2322, loss=0.1992, over 3272186.32 frames. utt_duration=1245 frames, utt_pad_proportion=0.05736, over 10527.81 utterances.], batch size: 47, lr: 4.27e-03, grad_scale: 8.0 2023-03-09 06:35:41,910 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=98971.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:36:44,960 INFO [train2.py:809] (0/4) Epoch 25, batch 3400, loss[ctc_loss=0.05659, att_loss=0.2075, loss=0.1773, over 15372.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01115, over 35.00 utterances.], tot_loss[ctc_loss=0.06867, att_loss=0.2337, loss=0.2007, over 3276359.22 frames. utt_duration=1198 frames, utt_pad_proportion=0.06765, over 10951.97 utterances.], batch size: 35, lr: 4.27e-03, grad_scale: 8.0 2023-03-09 06:37:07,451 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2053, 5.1694, 5.0437, 3.0701, 4.9751, 4.8012, 4.4243, 3.0960], device='cuda:0'), covar=tensor([0.0100, 0.0112, 0.0230, 0.0934, 0.0101, 0.0191, 0.0303, 0.1140], device='cuda:0'), in_proj_covar=tensor([0.0077, 0.0105, 0.0107, 0.0111, 0.0087, 0.0115, 0.0100, 0.0103], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:37:18,871 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99032.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:37:45,686 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.421e+02 1.932e+02 2.257e+02 2.819e+02 9.991e+02, threshold=4.513e+02, percent-clipped=2.0 2023-03-09 06:37:47,551 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99051.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:38:03,633 INFO [train2.py:809] (0/4) Epoch 25, batch 3450, loss[ctc_loss=0.04457, att_loss=0.2034, loss=0.1716, over 15782.00 frames. utt_duration=1663 frames, utt_pad_proportion=0.007929, over 38.00 utterances.], tot_loss[ctc_loss=0.06811, att_loss=0.233, loss=0.2, over 3271184.17 frames. utt_duration=1194 frames, utt_pad_proportion=0.06986, over 10971.59 utterances.], batch size: 38, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:38:12,066 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99066.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:39:22,270 INFO [train2.py:809] (0/4) Epoch 25, batch 3500, loss[ctc_loss=0.04399, att_loss=0.2, loss=0.1688, over 15382.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.01053, over 35.00 utterances.], tot_loss[ctc_loss=0.06819, att_loss=0.2328, loss=0.1999, over 3265751.65 frames. utt_duration=1193 frames, utt_pad_proportion=0.07132, over 10966.05 utterances.], batch size: 35, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:39:24,099 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99112.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:39:27,482 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99114.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:39:48,584 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5828, 2.2735, 2.1489, 2.4968, 2.7133, 2.5831, 2.1311, 2.9635], device='cuda:0'), covar=tensor([0.1729, 0.2342, 0.1939, 0.1501, 0.1562, 0.1122, 0.2314, 0.1241], device='cuda:0'), in_proj_covar=tensor([0.0132, 0.0135, 0.0131, 0.0124, 0.0140, 0.0121, 0.0142, 0.0120], device='cuda:0'), out_proj_covar=tensor([1.0145e-04, 1.0651e-04, 1.0642e-04, 9.7022e-05, 1.0580e-04, 9.7354e-05, 1.0880e-04, 9.5026e-05], device='cuda:0') 2023-03-09 06:39:55,266 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.14 vs. limit=5.0 2023-03-09 06:40:23,772 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.048e+02 1.663e+02 2.107e+02 2.565e+02 8.689e+02, threshold=4.214e+02, percent-clipped=2.0 2023-03-09 06:40:40,953 INFO [train2.py:809] (0/4) Epoch 25, batch 3550, loss[ctc_loss=0.05337, att_loss=0.2342, loss=0.1981, over 16958.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008108, over 50.00 utterances.], tot_loss[ctc_loss=0.06812, att_loss=0.2326, loss=0.1997, over 3266142.15 frames. utt_duration=1205 frames, utt_pad_proportion=0.06911, over 10856.96 utterances.], batch size: 50, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:41:40,268 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99199.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:41:59,394 INFO [train2.py:809] (0/4) Epoch 25, batch 3600, loss[ctc_loss=0.08254, att_loss=0.2409, loss=0.2092, over 15960.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.005926, over 41.00 utterances.], tot_loss[ctc_loss=0.06807, att_loss=0.2327, loss=0.1998, over 3267375.80 frames. utt_duration=1231 frames, utt_pad_proportion=0.06255, over 10631.62 utterances.], batch size: 41, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:42:28,453 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99229.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:42:55,189 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99246.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:42:57,234 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99247.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:43:01,837 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.167e+02 1.837e+02 2.162e+02 2.807e+02 5.340e+02, threshold=4.325e+02, percent-clipped=4.0 2023-03-09 06:43:03,506 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99251.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:43:11,989 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9779, 5.0147, 4.8367, 2.1754, 1.9470, 2.9194, 2.2401, 3.8122], device='cuda:0'), covar=tensor([0.0788, 0.0296, 0.0298, 0.6072, 0.5707, 0.2631, 0.4298, 0.1740], device='cuda:0'), in_proj_covar=tensor([0.0361, 0.0291, 0.0275, 0.0248, 0.0340, 0.0334, 0.0262, 0.0371], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 06:43:19,232 INFO [train2.py:809] (0/4) Epoch 25, batch 3650, loss[ctc_loss=0.07633, att_loss=0.2384, loss=0.2059, over 17298.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01222, over 55.00 utterances.], tot_loss[ctc_loss=0.06819, att_loss=0.233, loss=0.2001, over 3263349.76 frames. utt_duration=1218 frames, utt_pad_proportion=0.06562, over 10731.28 utterances.], batch size: 55, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:43:30,849 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99268.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:43:33,841 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7595, 5.2141, 5.0417, 5.1868, 5.2805, 4.8021, 3.6360, 5.2960], device='cuda:0'), covar=tensor([0.0118, 0.0106, 0.0114, 0.0079, 0.0097, 0.0132, 0.0638, 0.0163], device='cuda:0'), in_proj_covar=tensor([0.0095, 0.0091, 0.0113, 0.0070, 0.0077, 0.0088, 0.0104, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:43:41,608 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-03-09 06:44:05,939 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99290.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:44:24,440 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1065, 4.3994, 4.6593, 4.4754, 3.0241, 4.3988, 2.8345, 1.6647], device='cuda:0'), covar=tensor([0.0564, 0.0280, 0.0592, 0.0299, 0.1379, 0.0270, 0.1333, 0.1722], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0179, 0.0262, 0.0174, 0.0223, 0.0163, 0.0232, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:44:40,183 INFO [train2.py:809] (0/4) Epoch 25, batch 3700, loss[ctc_loss=0.06955, att_loss=0.2305, loss=0.1983, over 16389.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007485, over 44.00 utterances.], tot_loss[ctc_loss=0.06847, att_loss=0.2333, loss=0.2003, over 3267944.76 frames. utt_duration=1216 frames, utt_pad_proportion=0.06294, over 10762.78 utterances.], batch size: 44, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:45:01,480 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99324.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:45:04,495 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0782, 4.2815, 4.4616, 4.5032, 2.8799, 4.3499, 2.7409, 1.8567], device='cuda:0'), covar=tensor([0.0604, 0.0317, 0.0690, 0.0273, 0.1531, 0.0277, 0.1475, 0.1667], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0179, 0.0263, 0.0174, 0.0224, 0.0164, 0.0233, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:45:05,814 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99327.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:45:09,024 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99329.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:45:35,349 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0916, 2.5411, 2.5542, 3.7042, 3.5275, 3.6821, 2.7350, 2.2453], device='cuda:0'), covar=tensor([0.0701, 0.1915, 0.1192, 0.0634, 0.0903, 0.0401, 0.1438, 0.2027], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0217, 0.0187, 0.0221, 0.0232, 0.0182, 0.0204, 0.0190], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:45:41,895 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.181e+02 1.855e+02 2.238e+02 2.669e+02 6.912e+02, threshold=4.475e+02, percent-clipped=5.0 2023-03-09 06:45:59,439 INFO [train2.py:809] (0/4) Epoch 25, batch 3750, loss[ctc_loss=0.06808, att_loss=0.2271, loss=0.1953, over 15657.00 frames. utt_duration=1694 frames, utt_pad_proportion=0.00807, over 37.00 utterances.], tot_loss[ctc_loss=0.06811, att_loss=0.2333, loss=0.2003, over 3267999.11 frames. utt_duration=1215 frames, utt_pad_proportion=0.06247, over 10772.80 utterances.], batch size: 37, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:46:36,258 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1979, 2.9051, 3.5179, 2.7832, 3.3482, 4.3433, 4.1569, 3.0322], device='cuda:0'), covar=tensor([0.0410, 0.1600, 0.1220, 0.1367, 0.1134, 0.0898, 0.0624, 0.1297], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0247, 0.0288, 0.0222, 0.0268, 0.0375, 0.0269, 0.0233], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:46:37,874 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99385.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:47:13,256 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99407.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:47:19,829 INFO [train2.py:809] (0/4) Epoch 25, batch 3800, loss[ctc_loss=0.09089, att_loss=0.2633, loss=0.2288, over 17142.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01307, over 56.00 utterances.], tot_loss[ctc_loss=0.06769, att_loss=0.2329, loss=0.1998, over 3271947.39 frames. utt_duration=1231 frames, utt_pad_proportion=0.05802, over 10645.16 utterances.], batch size: 56, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:47:34,709 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99420.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:48:02,831 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0173, 5.0557, 4.8727, 2.2899, 1.9650, 2.8508, 2.4029, 3.7899], device='cuda:0'), covar=tensor([0.0756, 0.0293, 0.0271, 0.5512, 0.5861, 0.2748, 0.3966, 0.1742], device='cuda:0'), in_proj_covar=tensor([0.0364, 0.0294, 0.0278, 0.0252, 0.0345, 0.0338, 0.0265, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 06:48:22,108 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.283e+02 1.840e+02 2.193e+02 2.491e+02 3.920e+02, threshold=4.386e+02, percent-clipped=0.0 2023-03-09 06:48:38,758 INFO [train2.py:809] (0/4) Epoch 25, batch 3850, loss[ctc_loss=0.06767, att_loss=0.2148, loss=0.1854, over 15520.00 frames. utt_duration=1726 frames, utt_pad_proportion=0.007555, over 36.00 utterances.], tot_loss[ctc_loss=0.06665, att_loss=0.2316, loss=0.1986, over 3256370.67 frames. utt_duration=1242 frames, utt_pad_proportion=0.05897, over 10498.16 utterances.], batch size: 36, lr: 4.26e-03, grad_scale: 8.0 2023-03-09 06:48:49,787 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99468.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 06:49:09,249 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99481.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:49:44,838 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7177, 5.9888, 5.4871, 5.7218, 5.6532, 5.1967, 5.4157, 5.1731], device='cuda:0'), covar=tensor([0.1353, 0.0908, 0.0979, 0.0855, 0.1012, 0.1584, 0.2291, 0.2407], device='cuda:0'), in_proj_covar=tensor([0.0540, 0.0626, 0.0476, 0.0467, 0.0435, 0.0486, 0.0627, 0.0540], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:49:55,231 INFO [train2.py:809] (0/4) Epoch 25, batch 3900, loss[ctc_loss=0.05391, att_loss=0.2378, loss=0.201, over 17059.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.008534, over 53.00 utterances.], tot_loss[ctc_loss=0.06697, att_loss=0.2319, loss=0.1989, over 3255767.78 frames. utt_duration=1242 frames, utt_pad_proportion=0.05924, over 10494.68 utterances.], batch size: 53, lr: 4.25e-03, grad_scale: 8.0 2023-03-09 06:50:23,049 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99529.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 06:50:48,867 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99546.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:50:54,663 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.126e+02 1.849e+02 2.248e+02 2.718e+02 5.190e+02, threshold=4.496e+02, percent-clipped=3.0 2023-03-09 06:50:56,412 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99551.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:51:11,363 INFO [train2.py:809] (0/4) Epoch 25, batch 3950, loss[ctc_loss=0.1117, att_loss=0.2578, loss=0.2286, over 14500.00 frames. utt_duration=398.9 frames, utt_pad_proportion=0.3051, over 146.00 utterances.], tot_loss[ctc_loss=0.0678, att_loss=0.2326, loss=0.1996, over 3259585.86 frames. utt_duration=1227 frames, utt_pad_proportion=0.06084, over 10638.69 utterances.], batch size: 146, lr: 4.25e-03, grad_scale: 8.0 2023-03-09 06:51:22,240 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99568.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:51:47,848 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99585.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:51:54,113 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99589.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:51:59,707 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8286, 6.0657, 5.4841, 5.7555, 5.7640, 5.1936, 5.4267, 5.2942], device='cuda:0'), covar=tensor([0.1422, 0.0961, 0.0958, 0.0844, 0.0945, 0.1645, 0.2561, 0.2253], device='cuda:0'), in_proj_covar=tensor([0.0543, 0.0629, 0.0478, 0.0469, 0.0436, 0.0486, 0.0628, 0.0540], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 06:52:02,224 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-25.pt 2023-03-09 06:52:25,627 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99594.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:52:27,609 INFO [train2.py:809] (0/4) Epoch 26, batch 0, loss[ctc_loss=0.05771, att_loss=0.2396, loss=0.2032, over 16892.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006362, over 49.00 utterances.], tot_loss[ctc_loss=0.05771, att_loss=0.2396, loss=0.2032, over 16892.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006362, over 49.00 utterances.], batch size: 49, lr: 4.17e-03, grad_scale: 8.0 2023-03-09 06:52:27,611 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 06:52:36,964 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6049, 2.5498, 3.2470, 2.6659, 3.0959, 3.7974, 3.6583, 2.7843], device='cuda:0'), covar=tensor([0.0546, 0.1862, 0.1403, 0.1330, 0.1254, 0.1190, 0.0794, 0.1416], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0247, 0.0287, 0.0221, 0.0265, 0.0375, 0.0269, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:52:40,071 INFO [train2.py:843] (0/4) Epoch 26, validation: ctc_loss=0.04131, att_loss=0.2338, loss=0.1953, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 06:52:40,071 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 06:52:46,818 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99599.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:53:04,707 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8992, 4.8089, 4.7379, 2.4232, 1.8942, 2.9506, 2.1191, 3.8557], device='cuda:0'), covar=tensor([0.0798, 0.0370, 0.0267, 0.3801, 0.5656, 0.2450, 0.4243, 0.1497], device='cuda:0'), in_proj_covar=tensor([0.0360, 0.0292, 0.0276, 0.0249, 0.0341, 0.0335, 0.0261, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 06:53:27,419 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99624.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:53:32,200 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99627.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:53:36,027 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99629.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:53:39,132 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2114, 4.7232, 4.6596, 4.6457, 4.6851, 4.5023, 2.9046, 4.6506], device='cuda:0'), covar=tensor([0.0178, 0.0136, 0.0162, 0.0107, 0.0143, 0.0140, 0.1005, 0.0280], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0092, 0.0115, 0.0071, 0.0078, 0.0090, 0.0106, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 06:54:00,742 INFO [train2.py:809] (0/4) Epoch 26, batch 50, loss[ctc_loss=0.06845, att_loss=0.2391, loss=0.205, over 17341.00 frames. utt_duration=879.6 frames, utt_pad_proportion=0.07802, over 79.00 utterances.], tot_loss[ctc_loss=0.06584, att_loss=0.2315, loss=0.1984, over 740788.10 frames. utt_duration=1246 frames, utt_pad_proportion=0.0487, over 2381.71 utterances.], batch size: 79, lr: 4.17e-03, grad_scale: 8.0 2023-03-09 06:54:08,988 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.180e+02 1.776e+02 2.200e+02 2.639e+02 7.025e+02, threshold=4.399e+02, percent-clipped=1.0 2023-03-09 06:54:09,405 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99650.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:54:48,096 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99675.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:54:56,437 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:54:56,583 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=99680.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:55:19,430 INFO [train2.py:809] (0/4) Epoch 26, batch 100, loss[ctc_loss=0.105, att_loss=0.2667, loss=0.2344, over 17052.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009557, over 53.00 utterances.], tot_loss[ctc_loss=0.06792, att_loss=0.2331, loss=0.2001, over 1305410.05 frames. utt_duration=1314 frames, utt_pad_proportion=0.03212, over 3977.11 utterances.], batch size: 53, lr: 4.17e-03, grad_scale: 8.0 2023-03-09 06:55:38,617 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99707.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:56:09,382 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-09 06:56:12,770 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.79 vs. limit=5.0 2023-03-09 06:56:31,503 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=99741.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:56:37,955 INFO [train2.py:809] (0/4) Epoch 26, batch 150, loss[ctc_loss=0.06833, att_loss=0.2546, loss=0.2173, over 17338.00 frames. utt_duration=1177 frames, utt_pad_proportion=0.02178, over 59.00 utterances.], tot_loss[ctc_loss=0.06743, att_loss=0.2319, loss=0.199, over 1742709.07 frames. utt_duration=1297 frames, utt_pad_proportion=0.04043, over 5378.72 utterances.], batch size: 59, lr: 4.17e-03, grad_scale: 8.0 2023-03-09 06:56:46,324 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 1.892e+02 2.300e+02 2.888e+02 7.505e+02, threshold=4.601e+02, percent-clipped=4.0 2023-03-09 06:56:54,126 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99755.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:57:07,360 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4209, 2.8925, 3.3599, 4.4364, 3.9730, 3.9609, 3.0107, 2.3743], device='cuda:0'), covar=tensor([0.0775, 0.1941, 0.0913, 0.0622, 0.0869, 0.0478, 0.1554, 0.2228], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0220, 0.0189, 0.0224, 0.0234, 0.0186, 0.0206, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:57:27,662 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99776.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 06:57:57,204 INFO [train2.py:809] (0/4) Epoch 26, batch 200, loss[ctc_loss=0.07793, att_loss=0.2524, loss=0.2175, over 17278.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01249, over 55.00 utterances.], tot_loss[ctc_loss=0.06726, att_loss=0.2319, loss=0.199, over 2086305.76 frames. utt_duration=1314 frames, utt_pad_proportion=0.03475, over 6357.30 utterances.], batch size: 55, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 06:58:41,488 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9213, 3.6016, 3.6433, 3.1857, 3.7145, 3.6592, 3.7522, 2.7417], device='cuda:0'), covar=tensor([0.0935, 0.1862, 0.1418, 0.2515, 0.0816, 0.1813, 0.0689, 0.2902], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0200, 0.0213, 0.0267, 0.0177, 0.0275, 0.0196, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 06:58:44,420 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99824.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 06:59:17,381 INFO [train2.py:809] (0/4) Epoch 26, batch 250, loss[ctc_loss=0.05895, att_loss=0.2308, loss=0.1964, over 16688.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.005907, over 46.00 utterances.], tot_loss[ctc_loss=0.06752, att_loss=0.2331, loss=0.2, over 2359986.24 frames. utt_duration=1304 frames, utt_pad_proportion=0.03511, over 7245.90 utterances.], batch size: 46, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 06:59:25,254 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.472e+02 1.972e+02 2.288e+02 2.962e+02 9.624e+02, threshold=4.577e+02, percent-clipped=5.0 2023-03-09 06:59:48,935 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5235, 2.2246, 2.0429, 2.5992, 2.6705, 2.4628, 2.2416, 2.8858], device='cuda:0'), covar=tensor([0.1884, 0.2564, 0.1975, 0.1389, 0.2337, 0.1534, 0.2123, 0.1486], device='cuda:0'), in_proj_covar=tensor([0.0134, 0.0135, 0.0132, 0.0125, 0.0142, 0.0122, 0.0144, 0.0121], device='cuda:0'), out_proj_covar=tensor([1.0259e-04, 1.0682e-04, 1.0698e-04, 9.7999e-05, 1.0706e-04, 9.8241e-05, 1.1032e-04, 9.6017e-05], device='cuda:0') 2023-03-09 07:00:19,823 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99885.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:00:35,345 INFO [train2.py:809] (0/4) Epoch 26, batch 300, loss[ctc_loss=0.06817, att_loss=0.2496, loss=0.2133, over 17295.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01239, over 55.00 utterances.], tot_loss[ctc_loss=0.0678, att_loss=0.2328, loss=0.1998, over 2558366.74 frames. utt_duration=1295 frames, utt_pad_proportion=0.04069, over 7913.46 utterances.], batch size: 55, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:01:21,954 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:01:22,051 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99924.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:01:35,927 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99933.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:01:55,225 INFO [train2.py:809] (0/4) Epoch 26, batch 350, loss[ctc_loss=0.06008, att_loss=0.2349, loss=0.1999, over 16409.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006668, over 44.00 utterances.], tot_loss[ctc_loss=0.06783, att_loss=0.2319, loss=0.1991, over 2713038.06 frames. utt_duration=1287 frames, utt_pad_proportion=0.04328, over 8439.68 utterances.], batch size: 44, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:01:55,366 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=99945.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:02:02,686 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.378e+02 1.873e+02 2.109e+02 2.728e+02 6.807e+02, threshold=4.219e+02, percent-clipped=3.0 2023-03-09 07:02:39,736 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=99972.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:02:52,038 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=99980.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:03:15,360 INFO [train2.py:809] (0/4) Epoch 26, batch 400, loss[ctc_loss=0.06489, att_loss=0.2237, loss=0.1919, over 15629.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009294, over 37.00 utterances.], tot_loss[ctc_loss=0.06813, att_loss=0.2324, loss=0.1996, over 2841951.91 frames. utt_duration=1256 frames, utt_pad_proportion=0.04943, over 9060.06 utterances.], batch size: 37, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:03:23,303 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-100000.pt 2023-03-09 07:03:39,417 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100007.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 07:04:11,912 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100028.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:04:24,012 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100036.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:04:24,203 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7341, 2.3846, 2.3292, 2.5805, 2.8336, 2.8309, 2.4116, 2.9806], device='cuda:0'), covar=tensor([0.1943, 0.2395, 0.1878, 0.1471, 0.1575, 0.1532, 0.2347, 0.1611], device='cuda:0'), in_proj_covar=tensor([0.0133, 0.0134, 0.0130, 0.0124, 0.0142, 0.0121, 0.0144, 0.0121], device='cuda:0'), out_proj_covar=tensor([1.0212e-04, 1.0610e-04, 1.0617e-04, 9.7469e-05, 1.0663e-04, 9.7730e-05, 1.0998e-04, 9.5806e-05], device='cuda:0') 2023-03-09 07:04:38,117 INFO [train2.py:809] (0/4) Epoch 26, batch 450, loss[ctc_loss=0.0839, att_loss=0.251, loss=0.2176, over 13615.00 frames. utt_duration=377.1 frames, utt_pad_proportion=0.3441, over 145.00 utterances.], tot_loss[ctc_loss=0.06753, att_loss=0.2324, loss=0.1994, over 2939016.64 frames. utt_duration=1244 frames, utt_pad_proportion=0.05328, over 9461.10 utterances.], batch size: 145, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:04:46,163 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.318e+02 1.836e+02 2.145e+02 2.710e+02 6.185e+02, threshold=4.291e+02, percent-clipped=5.0 2023-03-09 07:05:14,836 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100068.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 07:05:27,494 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100076.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:05:56,809 INFO [train2.py:809] (0/4) Epoch 26, batch 500, loss[ctc_loss=0.05636, att_loss=0.2259, loss=0.192, over 16268.00 frames. utt_duration=1514 frames, utt_pad_proportion=0.008224, over 43.00 utterances.], tot_loss[ctc_loss=0.06755, att_loss=0.2328, loss=0.1997, over 3017588.32 frames. utt_duration=1244 frames, utt_pad_proportion=0.0528, over 9715.48 utterances.], batch size: 43, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:06:21,981 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.07 vs. limit=5.0 2023-03-09 07:06:43,578 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100124.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:06:43,795 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100124.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:07:16,638 INFO [train2.py:809] (0/4) Epoch 26, batch 550, loss[ctc_loss=0.06726, att_loss=0.2455, loss=0.2099, over 17117.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.0151, over 56.00 utterances.], tot_loss[ctc_loss=0.06682, att_loss=0.2317, loss=0.1987, over 3067370.00 frames. utt_duration=1269 frames, utt_pad_proportion=0.04976, over 9679.78 utterances.], batch size: 56, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:07:24,147 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.301e+02 1.784e+02 2.229e+02 2.699e+02 5.667e+02, threshold=4.458e+02, percent-clipped=3.0 2023-03-09 07:08:00,356 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100172.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:08:19,516 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9218, 2.4195, 2.3964, 2.6362, 2.7315, 2.8521, 2.5250, 3.0554], device='cuda:0'), covar=tensor([0.2086, 0.3590, 0.2506, 0.1968, 0.2799, 0.1691, 0.2826, 0.1811], device='cuda:0'), in_proj_covar=tensor([0.0135, 0.0136, 0.0132, 0.0126, 0.0144, 0.0123, 0.0145, 0.0123], device='cuda:0'), out_proj_covar=tensor([1.0363e-04, 1.0766e-04, 1.0748e-04, 9.9019e-05, 1.0832e-04, 9.9112e-05, 1.1125e-04, 9.7402e-05], device='cuda:0') 2023-03-09 07:08:36,931 INFO [train2.py:809] (0/4) Epoch 26, batch 600, loss[ctc_loss=0.06585, att_loss=0.2456, loss=0.2096, over 16628.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005349, over 47.00 utterances.], tot_loss[ctc_loss=0.06648, att_loss=0.2317, loss=0.1986, over 3113267.26 frames. utt_duration=1248 frames, utt_pad_proportion=0.05407, over 9991.74 utterances.], batch size: 47, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:08:37,065 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9235, 6.1178, 5.5933, 5.8345, 5.8414, 5.2817, 5.6412, 5.3667], device='cuda:0'), covar=tensor([0.1168, 0.0948, 0.1078, 0.0866, 0.0839, 0.1643, 0.2235, 0.2340], device='cuda:0'), in_proj_covar=tensor([0.0542, 0.0626, 0.0480, 0.0469, 0.0436, 0.0486, 0.0626, 0.0540], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 07:09:13,953 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.56 vs. limit=5.0 2023-03-09 07:09:19,657 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8310, 6.0767, 5.5461, 5.8183, 5.7797, 5.2567, 5.5270, 5.2949], device='cuda:0'), covar=tensor([0.1147, 0.0900, 0.1042, 0.0890, 0.0889, 0.1686, 0.2275, 0.2345], device='cuda:0'), in_proj_covar=tensor([0.0540, 0.0626, 0.0480, 0.0469, 0.0436, 0.0484, 0.0625, 0.0540], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 07:09:23,864 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-09 07:09:24,589 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100224.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:09:47,240 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100238.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:09:58,209 INFO [train2.py:809] (0/4) Epoch 26, batch 650, loss[ctc_loss=0.07774, att_loss=0.2476, loss=0.2136, over 17291.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01163, over 55.00 utterances.], tot_loss[ctc_loss=0.06667, att_loss=0.232, loss=0.1989, over 3142503.11 frames. utt_duration=1219 frames, utt_pad_proportion=0.06312, over 10323.14 utterances.], batch size: 55, lr: 4.16e-03, grad_scale: 8.0 2023-03-09 07:09:58,497 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100245.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:10:06,956 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 1.945e+02 2.331e+02 2.893e+02 4.719e+02, threshold=4.662e+02, percent-clipped=1.0 2023-03-09 07:10:12,248 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0210, 2.3110, 2.5340, 2.5219, 2.7942, 2.8463, 2.5443, 3.0704], device='cuda:0'), covar=tensor([0.1584, 0.2654, 0.1815, 0.1326, 0.1627, 0.1224, 0.1888, 0.1199], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0138, 0.0134, 0.0128, 0.0145, 0.0124, 0.0147, 0.0124], device='cuda:0'), out_proj_covar=tensor([1.0446e-04, 1.0896e-04, 1.0882e-04, 1.0000e-04, 1.0934e-04, 1.0013e-04, 1.1240e-04, 9.8185e-05], device='cuda:0') 2023-03-09 07:10:43,050 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100272.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:10:49,770 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0486, 4.9173, 4.9266, 2.0574, 2.0648, 2.7606, 2.2172, 3.9354], device='cuda:0'), covar=tensor([0.0716, 0.0251, 0.0192, 0.5363, 0.5404, 0.2676, 0.3854, 0.1442], device='cuda:0'), in_proj_covar=tensor([0.0362, 0.0295, 0.0278, 0.0250, 0.0343, 0.0335, 0.0262, 0.0372], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 07:11:16,616 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100293.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:11:19,700 INFO [train2.py:809] (0/4) Epoch 26, batch 700, loss[ctc_loss=0.05116, att_loss=0.2216, loss=0.1875, over 15963.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006004, over 41.00 utterances.], tot_loss[ctc_loss=0.06665, att_loss=0.2324, loss=0.1992, over 3178066.93 frames. utt_duration=1236 frames, utt_pad_proportion=0.05738, over 10298.32 utterances.], batch size: 41, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:11:26,815 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100299.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:12:05,530 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-09 07:12:20,612 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-09 07:12:26,332 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100336.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:12:40,623 INFO [train2.py:809] (0/4) Epoch 26, batch 750, loss[ctc_loss=0.05175, att_loss=0.2125, loss=0.1803, over 16123.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006556, over 42.00 utterances.], tot_loss[ctc_loss=0.06641, att_loss=0.2319, loss=0.1988, over 3197669.10 frames. utt_duration=1230 frames, utt_pad_proportion=0.05864, over 10408.86 utterances.], batch size: 42, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:12:48,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.213e+02 1.813e+02 2.158e+02 2.667e+02 1.532e+03, threshold=4.316e+02, percent-clipped=4.0 2023-03-09 07:13:10,583 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100363.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:13:43,862 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100384.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:14:01,727 INFO [train2.py:809] (0/4) Epoch 26, batch 800, loss[ctc_loss=0.07399, att_loss=0.2442, loss=0.2101, over 17144.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01353, over 56.00 utterances.], tot_loss[ctc_loss=0.06691, att_loss=0.2322, loss=0.1992, over 3212778.18 frames. utt_duration=1225 frames, utt_pad_proportion=0.06073, over 10499.52 utterances.], batch size: 56, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:14:13,161 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100402.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:14:14,681 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4782, 4.9358, 4.7316, 4.8554, 5.0089, 4.6532, 3.4463, 4.8843], device='cuda:0'), covar=tensor([0.0124, 0.0107, 0.0140, 0.0083, 0.0095, 0.0120, 0.0658, 0.0182], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0092, 0.0116, 0.0072, 0.0078, 0.0090, 0.0106, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 07:15:05,673 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-03-09 07:15:22,214 INFO [train2.py:809] (0/4) Epoch 26, batch 850, loss[ctc_loss=0.04789, att_loss=0.2047, loss=0.1733, over 15372.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.0105, over 35.00 utterances.], tot_loss[ctc_loss=0.06613, att_loss=0.2321, loss=0.1989, over 3229932.56 frames. utt_duration=1244 frames, utt_pad_proportion=0.0551, over 10394.75 utterances.], batch size: 35, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:15:24,897 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7529, 5.0158, 4.6552, 5.0902, 4.5034, 4.7372, 5.1252, 4.9351], device='cuda:0'), covar=tensor([0.0637, 0.0309, 0.0735, 0.0356, 0.0433, 0.0372, 0.0257, 0.0195], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0336, 0.0376, 0.0369, 0.0336, 0.0246, 0.0321, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 07:15:31,016 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 1.778e+02 2.108e+02 2.517e+02 5.047e+02, threshold=4.216e+02, percent-clipped=2.0 2023-03-09 07:15:52,676 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100463.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 07:16:43,801 INFO [train2.py:809] (0/4) Epoch 26, batch 900, loss[ctc_loss=0.07557, att_loss=0.254, loss=0.2183, over 17402.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03269, over 63.00 utterances.], tot_loss[ctc_loss=0.06615, att_loss=0.2322, loss=0.199, over 3246029.21 frames. utt_duration=1232 frames, utt_pad_proportion=0.05595, over 10547.76 utterances.], batch size: 63, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:18:03,339 INFO [train2.py:809] (0/4) Epoch 26, batch 950, loss[ctc_loss=0.0635, att_loss=0.2332, loss=0.1993, over 17501.00 frames. utt_duration=1016 frames, utt_pad_proportion=0.0417, over 69.00 utterances.], tot_loss[ctc_loss=0.06636, att_loss=0.232, loss=0.1989, over 3248697.16 frames. utt_duration=1260 frames, utt_pad_proportion=0.05076, over 10323.11 utterances.], batch size: 69, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:18:11,130 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.401e+02 1.882e+02 2.194e+02 2.822e+02 5.450e+02, threshold=4.387e+02, percent-clipped=2.0 2023-03-09 07:18:49,480 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5829, 4.8568, 4.5006, 4.9068, 4.3735, 4.5334, 4.9624, 4.7789], device='cuda:0'), covar=tensor([0.0741, 0.0345, 0.0789, 0.0398, 0.0460, 0.0408, 0.0283, 0.0229], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0337, 0.0377, 0.0370, 0.0337, 0.0247, 0.0321, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 07:19:22,764 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100594.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:19:24,219 INFO [train2.py:809] (0/4) Epoch 26, batch 1000, loss[ctc_loss=0.06169, att_loss=0.226, loss=0.1931, over 15998.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008008, over 40.00 utterances.], tot_loss[ctc_loss=0.06655, att_loss=0.2322, loss=0.1991, over 3254239.96 frames. utt_duration=1242 frames, utt_pad_proportion=0.05365, over 10493.15 utterances.], batch size: 40, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:20:41,600 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0192, 5.3578, 4.9511, 5.4175, 4.7579, 5.0316, 5.4878, 5.2532], device='cuda:0'), covar=tensor([0.0654, 0.0309, 0.0812, 0.0338, 0.0444, 0.0256, 0.0237, 0.0206], device='cuda:0'), in_proj_covar=tensor([0.0402, 0.0337, 0.0377, 0.0371, 0.0337, 0.0246, 0.0320, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 07:20:45,240 INFO [train2.py:809] (0/4) Epoch 26, batch 1050, loss[ctc_loss=0.05787, att_loss=0.2132, loss=0.1821, over 15394.00 frames. utt_duration=1761 frames, utt_pad_proportion=0.009738, over 35.00 utterances.], tot_loss[ctc_loss=0.06599, att_loss=0.2313, loss=0.1982, over 3249593.08 frames. utt_duration=1256 frames, utt_pad_proportion=0.05183, over 10362.86 utterances.], batch size: 35, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:20:53,113 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.280e+02 1.800e+02 2.231e+02 2.686e+02 5.529e+02, threshold=4.462e+02, percent-clipped=2.0 2023-03-09 07:21:14,638 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100663.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 07:22:05,996 INFO [train2.py:809] (0/4) Epoch 26, batch 1100, loss[ctc_loss=0.07743, att_loss=0.2468, loss=0.2129, over 16774.00 frames. utt_duration=679.3 frames, utt_pad_proportion=0.1476, over 99.00 utterances.], tot_loss[ctc_loss=0.06626, att_loss=0.2318, loss=0.1987, over 3258203.49 frames. utt_duration=1250 frames, utt_pad_proportion=0.05286, over 10438.62 utterances.], batch size: 99, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:22:07,985 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3320, 2.7568, 4.8602, 3.8783, 3.1620, 4.1263, 4.5081, 4.5958], device='cuda:0'), covar=tensor([0.0278, 0.1374, 0.0197, 0.0855, 0.1465, 0.0288, 0.0202, 0.0247], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0242, 0.0209, 0.0316, 0.0264, 0.0228, 0.0199, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:22:31,629 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100711.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 07:23:01,634 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=100730.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 07:23:24,299 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8541, 4.5792, 4.6714, 2.1086, 2.0317, 2.8396, 2.3206, 3.7246], device='cuda:0'), covar=tensor([0.0753, 0.0306, 0.0241, 0.5537, 0.5319, 0.2508, 0.3543, 0.1547], device='cuda:0'), in_proj_covar=tensor([0.0363, 0.0297, 0.0280, 0.0251, 0.0342, 0.0335, 0.0262, 0.0372], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 07:23:25,348 INFO [train2.py:809] (0/4) Epoch 26, batch 1150, loss[ctc_loss=0.07494, att_loss=0.2494, loss=0.2145, over 17388.00 frames. utt_duration=881.6 frames, utt_pad_proportion=0.07392, over 79.00 utterances.], tot_loss[ctc_loss=0.06705, att_loss=0.2324, loss=0.1994, over 3263841.78 frames. utt_duration=1235 frames, utt_pad_proportion=0.0568, over 10586.50 utterances.], batch size: 79, lr: 4.15e-03, grad_scale: 8.0 2023-03-09 07:23:32,923 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.106e+02 1.823e+02 2.156e+02 2.583e+02 8.005e+02, threshold=4.313e+02, percent-clipped=1.0 2023-03-09 07:23:46,153 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=100758.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 07:24:01,921 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3834, 4.3796, 4.5665, 4.5213, 5.0816, 4.3401, 4.3538, 2.5366], device='cuda:0'), covar=tensor([0.0264, 0.0423, 0.0333, 0.0314, 0.0617, 0.0278, 0.0394, 0.1867], device='cuda:0'), in_proj_covar=tensor([0.0182, 0.0211, 0.0207, 0.0224, 0.0379, 0.0182, 0.0196, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:24:16,129 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0198, 3.8013, 3.2714, 3.3721, 4.0334, 3.6080, 3.0605, 4.2867], device='cuda:0'), covar=tensor([0.0995, 0.0456, 0.0946, 0.0729, 0.0766, 0.0734, 0.0812, 0.0458], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0225, 0.0229, 0.0205, 0.0286, 0.0246, 0.0202, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 07:24:39,242 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=100791.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:24:39,765 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.18 vs. limit=2.0 2023-03-09 07:24:44,981 INFO [train2.py:809] (0/4) Epoch 26, batch 1200, loss[ctc_loss=0.0464, att_loss=0.2223, loss=0.1871, over 16549.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005165, over 45.00 utterances.], tot_loss[ctc_loss=0.06679, att_loss=0.2324, loss=0.1993, over 3274457.71 frames. utt_duration=1234 frames, utt_pad_proportion=0.05456, over 10623.08 utterances.], batch size: 45, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:24:55,062 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0084, 4.9604, 4.8451, 2.2419, 1.9781, 2.9882, 2.4190, 3.9183], device='cuda:0'), covar=tensor([0.0732, 0.0274, 0.0264, 0.5237, 0.5517, 0.2345, 0.3494, 0.1564], device='cuda:0'), in_proj_covar=tensor([0.0360, 0.0295, 0.0277, 0.0249, 0.0339, 0.0332, 0.0260, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 07:26:02,236 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-09 07:26:04,020 INFO [train2.py:809] (0/4) Epoch 26, batch 1250, loss[ctc_loss=0.04027, att_loss=0.2181, loss=0.1825, over 16122.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006541, over 42.00 utterances.], tot_loss[ctc_loss=0.06733, att_loss=0.2322, loss=0.1992, over 3276594.71 frames. utt_duration=1241 frames, utt_pad_proportion=0.05404, over 10574.60 utterances.], batch size: 42, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:26:11,742 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.226e+02 1.779e+02 2.144e+02 2.683e+02 5.396e+02, threshold=4.288e+02, percent-clipped=4.0 2023-03-09 07:26:50,264 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4422, 4.5959, 4.5950, 4.6065, 4.6998, 4.6671, 4.3239, 4.2236], device='cuda:0'), covar=tensor([0.0948, 0.0693, 0.0437, 0.0537, 0.0307, 0.0347, 0.0468, 0.0381], device='cuda:0'), in_proj_covar=tensor([0.0527, 0.0378, 0.0360, 0.0374, 0.0433, 0.0442, 0.0373, 0.0405], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 07:27:22,203 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=100894.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:27:23,463 INFO [train2.py:809] (0/4) Epoch 26, batch 1300, loss[ctc_loss=0.04321, att_loss=0.2237, loss=0.1876, over 16771.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006274, over 48.00 utterances.], tot_loss[ctc_loss=0.06703, att_loss=0.2318, loss=0.1989, over 3267751.91 frames. utt_duration=1241 frames, utt_pad_proportion=0.05587, over 10546.70 utterances.], batch size: 48, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:27:40,283 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4819, 2.5266, 4.9620, 3.9431, 3.0215, 4.1555, 4.6046, 4.6175], device='cuda:0'), covar=tensor([0.0292, 0.1738, 0.0196, 0.0814, 0.1638, 0.0271, 0.0190, 0.0254], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0247, 0.0214, 0.0323, 0.0270, 0.0232, 0.0202, 0.0230], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:28:03,454 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 07:28:07,194 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6771, 5.1061, 4.9282, 5.0726, 5.1766, 4.8677, 3.5432, 5.0767], device='cuda:0'), covar=tensor([0.0118, 0.0114, 0.0152, 0.0092, 0.0094, 0.0121, 0.0712, 0.0196], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0093, 0.0116, 0.0072, 0.0078, 0.0090, 0.0106, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 07:28:38,840 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=100942.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:28:43,600 INFO [train2.py:809] (0/4) Epoch 26, batch 1350, loss[ctc_loss=0.07056, att_loss=0.2528, loss=0.2164, over 17359.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03513, over 63.00 utterances.], tot_loss[ctc_loss=0.06694, att_loss=0.2317, loss=0.1987, over 3263002.53 frames. utt_duration=1246 frames, utt_pad_proportion=0.05734, over 10484.89 utterances.], batch size: 63, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:28:51,469 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 1.935e+02 2.207e+02 2.665e+02 5.260e+02, threshold=4.414e+02, percent-clipped=2.0 2023-03-09 07:29:45,452 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.88 vs. limit=5.0 2023-03-09 07:30:04,559 INFO [train2.py:809] (0/4) Epoch 26, batch 1400, loss[ctc_loss=0.05889, att_loss=0.2403, loss=0.204, over 16949.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.007719, over 50.00 utterances.], tot_loss[ctc_loss=0.06653, att_loss=0.2324, loss=0.1992, over 3271441.69 frames. utt_duration=1234 frames, utt_pad_proportion=0.05824, over 10619.64 utterances.], batch size: 50, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:30:04,880 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1453, 5.1994, 4.9269, 3.0774, 5.0634, 4.8067, 4.4675, 2.9544], device='cuda:0'), covar=tensor([0.0125, 0.0089, 0.0301, 0.1037, 0.0084, 0.0200, 0.0301, 0.1334], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0107, 0.0110, 0.0114, 0.0090, 0.0118, 0.0103, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 07:30:59,610 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.88 vs. limit=2.0 2023-03-09 07:31:24,469 INFO [train2.py:809] (0/4) Epoch 26, batch 1450, loss[ctc_loss=0.07669, att_loss=0.2587, loss=0.2223, over 17043.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009623, over 52.00 utterances.], tot_loss[ctc_loss=0.06618, att_loss=0.2315, loss=0.1985, over 3263530.04 frames. utt_duration=1242 frames, utt_pad_proportion=0.05715, over 10521.40 utterances.], batch size: 52, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:31:32,129 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.264e+02 1.754e+02 2.064e+02 2.618e+02 5.187e+02, threshold=4.127e+02, percent-clipped=2.0 2023-03-09 07:31:45,733 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101058.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:32:30,540 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101086.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 07:32:33,798 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0008, 3.8075, 3.6736, 3.2548, 3.7037, 3.7740, 3.8181, 2.9958], device='cuda:0'), covar=tensor([0.1014, 0.1017, 0.2510, 0.2569, 0.1774, 0.2247, 0.0766, 0.2569], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0199, 0.0214, 0.0264, 0.0176, 0.0275, 0.0195, 0.0226], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:32:44,602 INFO [train2.py:809] (0/4) Epoch 26, batch 1500, loss[ctc_loss=0.07006, att_loss=0.2239, loss=0.1931, over 16140.00 frames. utt_duration=1539 frames, utt_pad_proportion=0.005479, over 42.00 utterances.], tot_loss[ctc_loss=0.06604, att_loss=0.2316, loss=0.1985, over 3264679.21 frames. utt_duration=1251 frames, utt_pad_proportion=0.05497, over 10452.19 utterances.], batch size: 42, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:33:02,271 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101106.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:33:04,134 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101107.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:34:03,294 INFO [train2.py:809] (0/4) Epoch 26, batch 1550, loss[ctc_loss=0.06449, att_loss=0.2458, loss=0.2096, over 16953.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.0084, over 50.00 utterances.], tot_loss[ctc_loss=0.06631, att_loss=0.2318, loss=0.1987, over 3264231.88 frames. utt_duration=1233 frames, utt_pad_proportion=0.06023, over 10601.33 utterances.], batch size: 50, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:34:11,011 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.086e+02 1.859e+02 2.187e+02 2.715e+02 5.591e+02, threshold=4.373e+02, percent-clipped=3.0 2023-03-09 07:34:39,803 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101168.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:35:22,401 INFO [train2.py:809] (0/4) Epoch 26, batch 1600, loss[ctc_loss=0.07343, att_loss=0.2258, loss=0.1953, over 15352.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.01056, over 35.00 utterances.], tot_loss[ctc_loss=0.0667, att_loss=0.2324, loss=0.1992, over 3269019.05 frames. utt_duration=1211 frames, utt_pad_proportion=0.0632, over 10814.53 utterances.], batch size: 35, lr: 4.14e-03, grad_scale: 16.0 2023-03-09 07:35:48,918 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-09 07:36:42,605 INFO [train2.py:809] (0/4) Epoch 26, batch 1650, loss[ctc_loss=0.07401, att_loss=0.2544, loss=0.2183, over 16957.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007208, over 50.00 utterances.], tot_loss[ctc_loss=0.06664, att_loss=0.2327, loss=0.1995, over 3273443.10 frames. utt_duration=1234 frames, utt_pad_proportion=0.05742, over 10625.78 utterances.], batch size: 50, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:36:50,257 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.244e+02 1.931e+02 2.160e+02 2.616e+02 4.577e+02, threshold=4.320e+02, percent-clipped=2.0 2023-03-09 07:37:12,561 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9375, 6.1319, 5.6336, 5.8321, 5.8032, 5.2941, 5.5803, 5.2904], device='cuda:0'), covar=tensor([0.1159, 0.0833, 0.0885, 0.0753, 0.0923, 0.1509, 0.2050, 0.2351], device='cuda:0'), in_proj_covar=tensor([0.0550, 0.0634, 0.0485, 0.0475, 0.0447, 0.0487, 0.0638, 0.0547], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 07:38:02,607 INFO [train2.py:809] (0/4) Epoch 26, batch 1700, loss[ctc_loss=0.08552, att_loss=0.2506, loss=0.2176, over 17447.00 frames. utt_duration=1109 frames, utt_pad_proportion=0.03025, over 63.00 utterances.], tot_loss[ctc_loss=0.06709, att_loss=0.2331, loss=0.1999, over 3273759.49 frames. utt_duration=1233 frames, utt_pad_proportion=0.0587, over 10634.22 utterances.], batch size: 63, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:39:22,756 INFO [train2.py:809] (0/4) Epoch 26, batch 1750, loss[ctc_loss=0.07939, att_loss=0.2308, loss=0.2005, over 15784.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.008086, over 38.00 utterances.], tot_loss[ctc_loss=0.06717, att_loss=0.2325, loss=0.1994, over 3268139.38 frames. utt_duration=1255 frames, utt_pad_proportion=0.05289, over 10430.52 utterances.], batch size: 38, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:39:30,360 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.455e+02 1.899e+02 2.202e+02 2.819e+02 5.865e+02, threshold=4.404e+02, percent-clipped=4.0 2023-03-09 07:39:42,101 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8383, 2.5207, 2.8186, 2.6993, 2.8918, 2.8019, 2.5122, 3.2495], device='cuda:0'), covar=tensor([0.1632, 0.2408, 0.1664, 0.1579, 0.1722, 0.1269, 0.2178, 0.1133], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0139, 0.0133, 0.0128, 0.0146, 0.0125, 0.0147, 0.0124], device='cuda:0'), out_proj_covar=tensor([1.0512e-04, 1.0984e-04, 1.0860e-04, 1.0036e-04, 1.0963e-04, 1.0063e-04, 1.1256e-04, 9.8637e-05], device='cuda:0') 2023-03-09 07:40:08,988 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4643, 2.5107, 4.9563, 3.8768, 2.9848, 4.2142, 4.7443, 4.6054], device='cuda:0'), covar=tensor([0.0313, 0.1588, 0.0216, 0.0937, 0.1708, 0.0296, 0.0191, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0225, 0.0252, 0.0217, 0.0329, 0.0275, 0.0237, 0.0206, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:40:26,891 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101386.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:40:41,617 INFO [train2.py:809] (0/4) Epoch 26, batch 1800, loss[ctc_loss=0.06827, att_loss=0.2167, loss=0.187, over 15880.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009502, over 39.00 utterances.], tot_loss[ctc_loss=0.06676, att_loss=0.232, loss=0.199, over 3268407.79 frames. utt_duration=1245 frames, utt_pad_proportion=0.05473, over 10514.61 utterances.], batch size: 39, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:41:43,343 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101434.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:42:01,376 INFO [train2.py:809] (0/4) Epoch 26, batch 1850, loss[ctc_loss=0.06382, att_loss=0.2276, loss=0.1948, over 15955.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006346, over 41.00 utterances.], tot_loss[ctc_loss=0.06629, att_loss=0.2316, loss=0.1985, over 3270188.83 frames. utt_duration=1269 frames, utt_pad_proportion=0.04836, over 10321.04 utterances.], batch size: 41, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:42:08,708 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.271e+02 1.881e+02 2.093e+02 3.016e+02 9.107e+02, threshold=4.186e+02, percent-clipped=4.0 2023-03-09 07:42:29,645 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=101463.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:43:19,782 INFO [train2.py:809] (0/4) Epoch 26, batch 1900, loss[ctc_loss=0.04526, att_loss=0.2164, loss=0.1822, over 16404.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006835, over 44.00 utterances.], tot_loss[ctc_loss=0.06637, att_loss=0.2315, loss=0.1985, over 3272436.36 frames. utt_duration=1298 frames, utt_pad_proportion=0.04181, over 10099.89 utterances.], batch size: 44, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:44:39,308 INFO [train2.py:809] (0/4) Epoch 26, batch 1950, loss[ctc_loss=0.04168, att_loss=0.2169, loss=0.1819, over 15998.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007745, over 40.00 utterances.], tot_loss[ctc_loss=0.06722, att_loss=0.232, loss=0.199, over 3266536.04 frames. utt_duration=1277 frames, utt_pad_proportion=0.04831, over 10246.60 utterances.], batch size: 40, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:44:47,405 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.265e+02 1.785e+02 2.199e+02 2.810e+02 5.538e+02, threshold=4.398e+02, percent-clipped=5.0 2023-03-09 07:45:57,907 INFO [train2.py:809] (0/4) Epoch 26, batch 2000, loss[ctc_loss=0.07944, att_loss=0.2496, loss=0.2156, over 17414.00 frames. utt_duration=1011 frames, utt_pad_proportion=0.04617, over 69.00 utterances.], tot_loss[ctc_loss=0.06636, att_loss=0.232, loss=0.1988, over 3273497.71 frames. utt_duration=1288 frames, utt_pad_proportion=0.04312, over 10181.02 utterances.], batch size: 69, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:46:03,389 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.15 vs. limit=5.0 2023-03-09 07:47:05,478 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.12 vs. limit=2.0 2023-03-09 07:47:17,701 INFO [train2.py:809] (0/4) Epoch 26, batch 2050, loss[ctc_loss=0.1229, att_loss=0.2706, loss=0.2411, over 14234.00 frames. utt_duration=391.4 frames, utt_pad_proportion=0.3157, over 146.00 utterances.], tot_loss[ctc_loss=0.06632, att_loss=0.232, loss=0.1989, over 3265122.65 frames. utt_duration=1264 frames, utt_pad_proportion=0.05148, over 10346.22 utterances.], batch size: 146, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:47:26,102 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.028e+02 1.755e+02 2.108e+02 2.595e+02 4.367e+02, threshold=4.215e+02, percent-clipped=0.0 2023-03-09 07:47:50,887 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5433, 3.3659, 3.7994, 3.2820, 3.5270, 4.6740, 4.5451, 3.4398], device='cuda:0'), covar=tensor([0.0416, 0.1376, 0.1144, 0.1160, 0.1075, 0.0784, 0.0500, 0.1106], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0248, 0.0286, 0.0221, 0.0267, 0.0376, 0.0270, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 07:47:59,066 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4688, 2.9075, 3.4347, 4.3641, 3.8855, 3.9378, 3.0508, 2.4000], device='cuda:0'), covar=tensor([0.0690, 0.1955, 0.0912, 0.0682, 0.1026, 0.0546, 0.1367, 0.2140], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0220, 0.0188, 0.0223, 0.0234, 0.0189, 0.0204, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:48:19,674 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101684.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:48:37,218 INFO [train2.py:809] (0/4) Epoch 26, batch 2100, loss[ctc_loss=0.07071, att_loss=0.2278, loss=0.1964, over 16111.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006438, over 42.00 utterances.], tot_loss[ctc_loss=0.06635, att_loss=0.2316, loss=0.1985, over 3261757.33 frames. utt_duration=1254 frames, utt_pad_proportion=0.05497, over 10415.59 utterances.], batch size: 42, lr: 4.13e-03, grad_scale: 16.0 2023-03-09 07:48:46,574 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.4735, 5.3991, 5.3489, 3.7073, 5.3176, 5.1253, 4.9832, 3.6805], device='cuda:0'), covar=tensor([0.0082, 0.0085, 0.0175, 0.0733, 0.0074, 0.0137, 0.0211, 0.0882], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0107, 0.0110, 0.0113, 0.0089, 0.0118, 0.0103, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 07:49:26,813 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8409, 5.2375, 5.0848, 5.2822, 5.3677, 5.0388, 3.8824, 5.2324], device='cuda:0'), covar=tensor([0.0107, 0.0090, 0.0119, 0.0061, 0.0062, 0.0100, 0.0583, 0.0134], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0094, 0.0119, 0.0074, 0.0080, 0.0092, 0.0108, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-09 07:49:55,490 INFO [train2.py:809] (0/4) Epoch 26, batch 2150, loss[ctc_loss=0.1117, att_loss=0.2614, loss=0.2315, over 14205.00 frames. utt_duration=393.4 frames, utt_pad_proportion=0.317, over 145.00 utterances.], tot_loss[ctc_loss=0.06586, att_loss=0.2312, loss=0.1981, over 3259966.30 frames. utt_duration=1253 frames, utt_pad_proportion=0.05612, over 10422.65 utterances.], batch size: 145, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 07:49:55,856 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101745.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:50:03,015 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 1.940e+02 2.256e+02 2.810e+02 5.136e+02, threshold=4.511e+02, percent-clipped=2.0 2023-03-09 07:50:23,099 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=101763.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:51:13,849 INFO [train2.py:809] (0/4) Epoch 26, batch 2200, loss[ctc_loss=0.06376, att_loss=0.2388, loss=0.2038, over 17395.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03316, over 63.00 utterances.], tot_loss[ctc_loss=0.06624, att_loss=0.2314, loss=0.1984, over 3250890.40 frames. utt_duration=1237 frames, utt_pad_proportion=0.06186, over 10523.32 utterances.], batch size: 63, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 07:51:33,263 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.56 vs. limit=2.0 2023-03-09 07:51:38,367 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=101811.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:51:54,466 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2655, 4.3693, 4.4943, 4.5357, 5.0184, 4.3796, 4.3779, 2.5431], device='cuda:0'), covar=tensor([0.0327, 0.0406, 0.0350, 0.0331, 0.0762, 0.0289, 0.0373, 0.1819], device='cuda:0'), in_proj_covar=tensor([0.0184, 0.0212, 0.0207, 0.0224, 0.0379, 0.0183, 0.0197, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:52:32,211 INFO [train2.py:809] (0/4) Epoch 26, batch 2250, loss[ctc_loss=0.06617, att_loss=0.2365, loss=0.2024, over 16692.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006171, over 46.00 utterances.], tot_loss[ctc_loss=0.0671, att_loss=0.2319, loss=0.1989, over 3254610.44 frames. utt_duration=1209 frames, utt_pad_proportion=0.068, over 10780.37 utterances.], batch size: 46, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 07:52:39,682 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.215e+02 1.832e+02 2.219e+02 2.925e+02 8.188e+02, threshold=4.439e+02, percent-clipped=2.0 2023-03-09 07:52:44,825 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4736, 2.6982, 4.9519, 3.9530, 3.2827, 4.2918, 4.7525, 4.7274], device='cuda:0'), covar=tensor([0.0272, 0.1480, 0.0192, 0.0856, 0.1452, 0.0247, 0.0181, 0.0228], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0245, 0.0213, 0.0321, 0.0269, 0.0232, 0.0203, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:53:51,086 INFO [train2.py:809] (0/4) Epoch 26, batch 2300, loss[ctc_loss=0.06217, att_loss=0.2311, loss=0.1973, over 17334.00 frames. utt_duration=1102 frames, utt_pad_proportion=0.03575, over 63.00 utterances.], tot_loss[ctc_loss=0.06724, att_loss=0.2324, loss=0.1994, over 3262507.42 frames. utt_duration=1197 frames, utt_pad_proportion=0.0697, over 10919.70 utterances.], batch size: 63, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 07:54:30,193 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101920.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:54:59,204 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7103, 3.1767, 3.9330, 3.2707, 3.6793, 4.7479, 4.6785, 3.5285], device='cuda:0'), covar=tensor([0.0300, 0.1589, 0.0956, 0.1260, 0.0980, 0.0812, 0.0441, 0.1056], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0249, 0.0286, 0.0222, 0.0268, 0.0378, 0.0271, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 07:55:09,344 INFO [train2.py:809] (0/4) Epoch 26, batch 2350, loss[ctc_loss=0.05072, att_loss=0.2101, loss=0.1782, over 11514.00 frames. utt_duration=1844 frames, utt_pad_proportion=0.1885, over 25.00 utterances.], tot_loss[ctc_loss=0.06641, att_loss=0.2317, loss=0.1986, over 3257595.88 frames. utt_duration=1221 frames, utt_pad_proportion=0.0629, over 10685.52 utterances.], batch size: 25, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 07:55:16,748 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 1.850e+02 2.322e+02 2.866e+02 7.956e+02, threshold=4.644e+02, percent-clipped=4.0 2023-03-09 07:55:17,268 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4403, 4.4409, 4.5756, 4.5402, 5.0871, 4.5026, 4.4059, 2.5729], device='cuda:0'), covar=tensor([0.0281, 0.0385, 0.0349, 0.0326, 0.0625, 0.0248, 0.0383, 0.1735], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0213, 0.0207, 0.0225, 0.0380, 0.0184, 0.0198, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 07:55:41,420 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-03-09 07:56:05,568 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=101981.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 07:56:15,050 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=101987.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:56:27,655 INFO [train2.py:809] (0/4) Epoch 26, batch 2400, loss[ctc_loss=0.04456, att_loss=0.2212, loss=0.1859, over 16943.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008824, over 50.00 utterances.], tot_loss[ctc_loss=0.0656, att_loss=0.231, loss=0.198, over 3258307.48 frames. utt_duration=1241 frames, utt_pad_proportion=0.0583, over 10515.22 utterances.], batch size: 50, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 07:56:35,736 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-102000.pt 2023-03-09 07:57:44,259 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102040.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:57:51,789 INFO [train2.py:809] (0/4) Epoch 26, batch 2450, loss[ctc_loss=0.06475, att_loss=0.2093, loss=0.1804, over 15511.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.007416, over 36.00 utterances.], tot_loss[ctc_loss=0.06635, att_loss=0.2315, loss=0.1985, over 3259287.80 frames. utt_duration=1238 frames, utt_pad_proportion=0.0588, over 10539.53 utterances.], batch size: 36, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 07:57:56,717 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102048.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 07:57:59,372 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 1.902e+02 2.246e+02 2.761e+02 8.723e+02, threshold=4.491e+02, percent-clipped=3.0 2023-03-09 07:58:53,616 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.28 vs. limit=5.0 2023-03-09 07:59:11,506 INFO [train2.py:809] (0/4) Epoch 26, batch 2500, loss[ctc_loss=0.08612, att_loss=0.2484, loss=0.2159, over 17129.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01424, over 56.00 utterances.], tot_loss[ctc_loss=0.06611, att_loss=0.2315, loss=0.1984, over 3259773.92 frames. utt_duration=1224 frames, utt_pad_proportion=0.06343, over 10668.59 utterances.], batch size: 56, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 08:00:29,997 INFO [train2.py:809] (0/4) Epoch 26, batch 2550, loss[ctc_loss=0.07257, att_loss=0.2484, loss=0.2132, over 16931.00 frames. utt_duration=692.8 frames, utt_pad_proportion=0.1319, over 98.00 utterances.], tot_loss[ctc_loss=0.06652, att_loss=0.2317, loss=0.1986, over 3262779.71 frames. utt_duration=1211 frames, utt_pad_proportion=0.06565, over 10794.76 utterances.], batch size: 98, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 08:00:37,742 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.009e+02 1.703e+02 2.113e+02 2.564e+02 5.440e+02, threshold=4.227e+02, percent-clipped=2.0 2023-03-09 08:01:48,433 INFO [train2.py:809] (0/4) Epoch 26, batch 2600, loss[ctc_loss=0.04239, att_loss=0.2118, loss=0.1779, over 15966.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.005771, over 41.00 utterances.], tot_loss[ctc_loss=0.0663, att_loss=0.2314, loss=0.1984, over 3264897.41 frames. utt_duration=1227 frames, utt_pad_proportion=0.0619, over 10653.33 utterances.], batch size: 41, lr: 4.12e-03, grad_scale: 16.0 2023-03-09 08:02:08,632 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5707, 3.2595, 3.6112, 4.4496, 3.9952, 3.9655, 3.0542, 2.4217], device='cuda:0'), covar=tensor([0.0674, 0.1674, 0.0829, 0.0569, 0.0928, 0.0551, 0.1478, 0.2173], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0220, 0.0189, 0.0224, 0.0234, 0.0189, 0.0205, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:03:06,803 INFO [train2.py:809] (0/4) Epoch 26, batch 2650, loss[ctc_loss=0.06325, att_loss=0.2252, loss=0.1928, over 16183.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006776, over 41.00 utterances.], tot_loss[ctc_loss=0.06621, att_loss=0.2319, loss=0.1987, over 3276119.91 frames. utt_duration=1240 frames, utt_pad_proportion=0.05511, over 10577.87 utterances.], batch size: 41, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:03:14,512 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 1.711e+02 1.973e+02 2.429e+02 4.776e+02, threshold=3.946e+02, percent-clipped=2.0 2023-03-09 08:03:33,896 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5355, 3.1036, 3.7494, 3.2213, 3.6183, 4.6345, 4.4622, 3.3842], device='cuda:0'), covar=tensor([0.0380, 0.1625, 0.1175, 0.1242, 0.1009, 0.0847, 0.0534, 0.1180], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0250, 0.0289, 0.0223, 0.0269, 0.0380, 0.0272, 0.0236], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:03:55,842 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102276.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 08:04:18,255 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8847, 5.2018, 5.0860, 5.1090, 5.2068, 5.1447, 4.7826, 4.6314], device='cuda:0'), covar=tensor([0.1085, 0.0539, 0.0376, 0.0492, 0.0305, 0.0335, 0.0465, 0.0401], device='cuda:0'), in_proj_covar=tensor([0.0529, 0.0376, 0.0364, 0.0374, 0.0435, 0.0444, 0.0375, 0.0406], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:04:25,662 INFO [train2.py:809] (0/4) Epoch 26, batch 2700, loss[ctc_loss=0.0555, att_loss=0.2094, loss=0.1786, over 15629.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009795, over 37.00 utterances.], tot_loss[ctc_loss=0.06655, att_loss=0.2322, loss=0.1991, over 3274004.18 frames. utt_duration=1236 frames, utt_pad_proportion=0.05712, over 10604.73 utterances.], batch size: 37, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:05:37,441 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102340.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:05:41,910 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=102343.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:05:44,679 INFO [train2.py:809] (0/4) Epoch 26, batch 2750, loss[ctc_loss=0.05917, att_loss=0.2194, loss=0.1874, over 16265.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007498, over 43.00 utterances.], tot_loss[ctc_loss=0.06609, att_loss=0.2324, loss=0.1991, over 3281263.48 frames. utt_duration=1247 frames, utt_pad_proportion=0.05283, over 10539.44 utterances.], batch size: 43, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:05:50,873 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0875, 5.3422, 5.6289, 5.3557, 5.5854, 5.9929, 5.2751, 6.0940], device='cuda:0'), covar=tensor([0.0676, 0.0688, 0.0804, 0.1421, 0.1768, 0.1004, 0.0702, 0.0707], device='cuda:0'), in_proj_covar=tensor([0.0919, 0.0526, 0.0638, 0.0686, 0.0909, 0.0663, 0.0513, 0.0645], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:05:52,187 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.173e+02 1.911e+02 2.234e+02 2.672e+02 6.955e+02, threshold=4.468e+02, percent-clipped=5.0 2023-03-09 08:05:56,882 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9732, 5.2442, 4.8319, 5.3370, 4.6444, 4.9762, 5.4105, 5.1945], device='cuda:0'), covar=tensor([0.0653, 0.0356, 0.0852, 0.0322, 0.0495, 0.0309, 0.0245, 0.0222], device='cuda:0'), in_proj_covar=tensor([0.0399, 0.0336, 0.0376, 0.0368, 0.0336, 0.0245, 0.0315, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 08:05:59,997 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0843, 4.3803, 4.3919, 4.4209, 4.4545, 4.2645, 2.8210, 4.3166], device='cuda:0'), covar=tensor([0.0185, 0.0214, 0.0194, 0.0138, 0.0180, 0.0179, 0.1076, 0.0462], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0095, 0.0119, 0.0074, 0.0080, 0.0092, 0.0108, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-09 08:06:52,258 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102388.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:06:55,583 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7557, 2.3032, 5.1416, 4.1328, 3.2684, 4.5331, 4.9027, 4.8839], device='cuda:0'), covar=tensor([0.0195, 0.1585, 0.0188, 0.0804, 0.1421, 0.0185, 0.0126, 0.0185], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0247, 0.0215, 0.0323, 0.0271, 0.0234, 0.0206, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:07:02,733 INFO [train2.py:809] (0/4) Epoch 26, batch 2800, loss[ctc_loss=0.05331, att_loss=0.2054, loss=0.175, over 15507.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.00837, over 36.00 utterances.], tot_loss[ctc_loss=0.06644, att_loss=0.2326, loss=0.1994, over 3280079.64 frames. utt_duration=1232 frames, utt_pad_proportion=0.05712, over 10665.39 utterances.], batch size: 36, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:07:24,989 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1786, 5.4424, 5.6838, 5.4791, 5.6790, 6.1144, 5.3617, 6.1605], device='cuda:0'), covar=tensor([0.0682, 0.0683, 0.0906, 0.1350, 0.1905, 0.0876, 0.0627, 0.0752], device='cuda:0'), in_proj_covar=tensor([0.0918, 0.0524, 0.0637, 0.0686, 0.0907, 0.0663, 0.0512, 0.0644], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:08:21,716 INFO [train2.py:809] (0/4) Epoch 26, batch 2850, loss[ctc_loss=0.0952, att_loss=0.2211, loss=0.196, over 15616.00 frames. utt_duration=1690 frames, utt_pad_proportion=0.01067, over 37.00 utterances.], tot_loss[ctc_loss=0.06674, att_loss=0.2331, loss=0.1998, over 3274440.10 frames. utt_duration=1193 frames, utt_pad_proportion=0.06926, over 10996.30 utterances.], batch size: 37, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:08:29,303 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.109e+02 1.936e+02 2.281e+02 2.866e+02 5.493e+02, threshold=4.561e+02, percent-clipped=4.0 2023-03-09 08:08:57,572 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.84 vs. limit=2.0 2023-03-09 08:09:19,581 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9138, 3.6233, 3.0510, 3.2645, 3.8185, 3.5030, 2.8557, 4.0365], device='cuda:0'), covar=tensor([0.1024, 0.0480, 0.1020, 0.0797, 0.0757, 0.0773, 0.0923, 0.0430], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0223, 0.0227, 0.0206, 0.0288, 0.0246, 0.0202, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 08:09:40,750 INFO [train2.py:809] (0/4) Epoch 26, batch 2900, loss[ctc_loss=0.06322, att_loss=0.2127, loss=0.1828, over 16280.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006705, over 43.00 utterances.], tot_loss[ctc_loss=0.06744, att_loss=0.2336, loss=0.2004, over 3277514.07 frames. utt_duration=1180 frames, utt_pad_proportion=0.07044, over 11125.29 utterances.], batch size: 43, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:11:00,377 INFO [train2.py:809] (0/4) Epoch 26, batch 2950, loss[ctc_loss=0.06901, att_loss=0.2498, loss=0.2137, over 16953.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008298, over 50.00 utterances.], tot_loss[ctc_loss=0.06711, att_loss=0.2331, loss=0.1999, over 3279889.61 frames. utt_duration=1196 frames, utt_pad_proportion=0.06641, over 10986.04 utterances.], batch size: 50, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:11:08,368 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 1.971e+02 2.322e+02 2.777e+02 5.083e+02, threshold=4.643e+02, percent-clipped=3.0 2023-03-09 08:11:53,983 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102576.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 08:12:22,863 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9541, 4.0141, 4.0418, 4.0808, 4.1357, 4.1218, 3.8393, 3.7872], device='cuda:0'), covar=tensor([0.1100, 0.0848, 0.0836, 0.0595, 0.0387, 0.0428, 0.0517, 0.0403], device='cuda:0'), in_proj_covar=tensor([0.0535, 0.0379, 0.0365, 0.0377, 0.0438, 0.0445, 0.0376, 0.0409], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:12:24,253 INFO [train2.py:809] (0/4) Epoch 26, batch 3000, loss[ctc_loss=0.1118, att_loss=0.2599, loss=0.2303, over 13834.00 frames. utt_duration=380.5 frames, utt_pad_proportion=0.3348, over 146.00 utterances.], tot_loss[ctc_loss=0.06653, att_loss=0.2326, loss=0.1994, over 3274623.91 frames. utt_duration=1212 frames, utt_pad_proportion=0.06283, over 10816.69 utterances.], batch size: 146, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:12:24,256 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 08:12:38,598 INFO [train2.py:843] (0/4) Epoch 26, validation: ctc_loss=0.04046, att_loss=0.2348, loss=0.1959, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 08:12:38,599 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 08:12:38,946 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5765, 3.1787, 3.7376, 3.4054, 3.5210, 4.6355, 4.4827, 3.4773], device='cuda:0'), covar=tensor([0.0386, 0.1596, 0.1315, 0.1119, 0.1051, 0.0897, 0.0567, 0.1061], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0251, 0.0288, 0.0223, 0.0270, 0.0381, 0.0273, 0.0236], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:12:40,688 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3213, 2.6160, 4.7574, 3.8241, 2.9985, 4.1659, 4.3783, 4.4454], device='cuda:0'), covar=tensor([0.0266, 0.1522, 0.0241, 0.0827, 0.1644, 0.0271, 0.0235, 0.0291], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0247, 0.0215, 0.0323, 0.0272, 0.0233, 0.0206, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:13:27,777 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102624.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 08:13:45,105 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0110, 5.0003, 4.6731, 2.9755, 4.7530, 4.7280, 4.2713, 2.5654], device='cuda:0'), covar=tensor([0.0122, 0.0118, 0.0339, 0.0974, 0.0115, 0.0198, 0.0319, 0.1482], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0107, 0.0111, 0.0113, 0.0090, 0.0118, 0.0102, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:14:00,060 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=102643.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:14:03,002 INFO [train2.py:809] (0/4) Epoch 26, batch 3050, loss[ctc_loss=0.04439, att_loss=0.2029, loss=0.1712, over 15509.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.008242, over 36.00 utterances.], tot_loss[ctc_loss=0.06686, att_loss=0.233, loss=0.1998, over 3276024.10 frames. utt_duration=1225 frames, utt_pad_proportion=0.0601, over 10712.68 utterances.], batch size: 36, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:14:10,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.303e+02 1.917e+02 2.316e+02 2.727e+02 9.288e+02, threshold=4.632e+02, percent-clipped=2.0 2023-03-09 08:14:17,639 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1345, 3.7264, 3.0895, 3.3640, 3.9336, 3.5760, 3.0148, 4.2280], device='cuda:0'), covar=tensor([0.0976, 0.0556, 0.1131, 0.0822, 0.0732, 0.0807, 0.0882, 0.0478], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0224, 0.0229, 0.0207, 0.0289, 0.0248, 0.0203, 0.0295], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 08:14:52,306 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6736, 5.9111, 5.3503, 5.5579, 5.5725, 5.0474, 5.3524, 4.9946], device='cuda:0'), covar=tensor([0.1238, 0.0906, 0.1038, 0.0853, 0.0978, 0.1420, 0.2336, 0.2282], device='cuda:0'), in_proj_covar=tensor([0.0550, 0.0633, 0.0488, 0.0479, 0.0448, 0.0486, 0.0634, 0.0544], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:15:10,968 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1875, 3.8247, 3.2103, 3.3685, 4.0133, 3.6551, 3.1320, 4.2695], device='cuda:0'), covar=tensor([0.0931, 0.0522, 0.1087, 0.0836, 0.0766, 0.0713, 0.0823, 0.0583], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0223, 0.0229, 0.0207, 0.0289, 0.0247, 0.0203, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 08:15:18,608 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=102691.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:15:24,934 INFO [train2.py:809] (0/4) Epoch 26, batch 3100, loss[ctc_loss=0.09227, att_loss=0.2627, loss=0.2286, over 17396.00 frames. utt_duration=1010 frames, utt_pad_proportion=0.04541, over 69.00 utterances.], tot_loss[ctc_loss=0.06711, att_loss=0.2332, loss=0.2, over 3276282.42 frames. utt_duration=1230 frames, utt_pad_proportion=0.05886, over 10670.66 utterances.], batch size: 69, lr: 4.11e-03, grad_scale: 16.0 2023-03-09 08:16:08,706 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.82 vs. limit=5.0 2023-03-09 08:16:47,143 INFO [train2.py:809] (0/4) Epoch 26, batch 3150, loss[ctc_loss=0.06023, att_loss=0.2224, loss=0.19, over 16122.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.005929, over 42.00 utterances.], tot_loss[ctc_loss=0.06772, att_loss=0.2342, loss=0.2009, over 3280211.66 frames. utt_duration=1206 frames, utt_pad_proportion=0.06453, over 10894.92 utterances.], batch size: 42, lr: 4.10e-03, grad_scale: 16.0 2023-03-09 08:16:54,908 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.334e+02 1.851e+02 2.183e+02 2.802e+02 6.228e+02, threshold=4.366e+02, percent-clipped=1.0 2023-03-09 08:18:06,657 INFO [train2.py:809] (0/4) Epoch 26, batch 3200, loss[ctc_loss=0.07491, att_loss=0.249, loss=0.2141, over 17062.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008302, over 52.00 utterances.], tot_loss[ctc_loss=0.06703, att_loss=0.2334, loss=0.2001, over 3283167.45 frames. utt_duration=1219 frames, utt_pad_proportion=0.05864, over 10788.52 utterances.], batch size: 52, lr: 4.10e-03, grad_scale: 32.0 2023-03-09 08:19:25,810 INFO [train2.py:809] (0/4) Epoch 26, batch 3250, loss[ctc_loss=0.05068, att_loss=0.2286, loss=0.193, over 16629.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005005, over 47.00 utterances.], tot_loss[ctc_loss=0.06753, att_loss=0.2336, loss=0.2004, over 3287479.38 frames. utt_duration=1220 frames, utt_pad_proportion=0.05769, over 10791.66 utterances.], batch size: 47, lr: 4.10e-03, grad_scale: 32.0 2023-03-09 08:19:33,383 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 1.810e+02 2.101e+02 2.825e+02 6.306e+02, threshold=4.202e+02, percent-clipped=5.0 2023-03-09 08:19:35,338 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102851.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:20:32,517 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=102887.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:20:44,503 INFO [train2.py:809] (0/4) Epoch 26, batch 3300, loss[ctc_loss=0.08235, att_loss=0.2532, loss=0.219, over 17363.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03588, over 63.00 utterances.], tot_loss[ctc_loss=0.06669, att_loss=0.2332, loss=0.1999, over 3288872.14 frames. utt_duration=1230 frames, utt_pad_proportion=0.05466, over 10710.24 utterances.], batch size: 63, lr: 4.10e-03, grad_scale: 32.0 2023-03-09 08:21:11,612 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102912.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:21:44,547 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.37 vs. limit=2.0 2023-03-09 08:21:48,767 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5028, 2.9754, 3.7288, 3.1933, 3.5471, 4.5778, 4.4533, 3.4471], device='cuda:0'), covar=tensor([0.0375, 0.1752, 0.1232, 0.1249, 0.1100, 0.1038, 0.0570, 0.1094], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0250, 0.0288, 0.0223, 0.0270, 0.0380, 0.0272, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:22:03,746 INFO [train2.py:809] (0/4) Epoch 26, batch 3350, loss[ctc_loss=0.06736, att_loss=0.2371, loss=0.2032, over 16752.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006708, over 48.00 utterances.], tot_loss[ctc_loss=0.06603, att_loss=0.2323, loss=0.1991, over 3275179.33 frames. utt_duration=1242 frames, utt_pad_proportion=0.05632, over 10557.34 utterances.], batch size: 48, lr: 4.10e-03, grad_scale: 32.0 2023-03-09 08:22:08,702 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=102948.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:22:11,376 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.207e+02 1.718e+02 2.165e+02 2.625e+02 3.837e+02, threshold=4.330e+02, percent-clipped=0.0 2023-03-09 08:22:47,136 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 08:23:22,223 INFO [train2.py:809] (0/4) Epoch 26, batch 3400, loss[ctc_loss=0.05636, att_loss=0.2259, loss=0.192, over 16183.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006653, over 41.00 utterances.], tot_loss[ctc_loss=0.06582, att_loss=0.2325, loss=0.1992, over 3287720.90 frames. utt_duration=1246 frames, utt_pad_proportion=0.05106, over 10567.34 utterances.], batch size: 41, lr: 4.10e-03, grad_scale: 16.0 2023-03-09 08:23:57,542 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5608, 2.4630, 4.8498, 3.8736, 3.0239, 4.3194, 4.5194, 4.6991], device='cuda:0'), covar=tensor([0.0191, 0.1478, 0.0187, 0.0755, 0.1505, 0.0203, 0.0180, 0.0216], device='cuda:0'), in_proj_covar=tensor([0.0224, 0.0248, 0.0217, 0.0326, 0.0273, 0.0235, 0.0208, 0.0233], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:24:22,988 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0891, 3.6953, 3.0640, 3.3278, 3.8973, 3.5772, 2.8231, 4.1347], device='cuda:0'), covar=tensor([0.0930, 0.0572, 0.1178, 0.0812, 0.0762, 0.0795, 0.0990, 0.0541], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0222, 0.0229, 0.0205, 0.0287, 0.0246, 0.0203, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 08:24:40,905 INFO [train2.py:809] (0/4) Epoch 26, batch 3450, loss[ctc_loss=0.09505, att_loss=0.2594, loss=0.2265, over 17045.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.009216, over 52.00 utterances.], tot_loss[ctc_loss=0.06604, att_loss=0.2325, loss=0.1992, over 3289768.88 frames. utt_duration=1241 frames, utt_pad_proportion=0.05192, over 10617.40 utterances.], batch size: 52, lr: 4.10e-03, grad_scale: 16.0 2023-03-09 08:24:50,036 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.313e+02 1.811e+02 2.239e+02 2.771e+02 6.342e+02, threshold=4.479e+02, percent-clipped=4.0 2023-03-09 08:25:10,278 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8101, 6.0563, 5.5737, 5.7822, 5.7497, 5.2560, 5.5270, 5.2491], device='cuda:0'), covar=tensor([0.1467, 0.0957, 0.0927, 0.0827, 0.1016, 0.1493, 0.2475, 0.2275], device='cuda:0'), in_proj_covar=tensor([0.0559, 0.0642, 0.0492, 0.0484, 0.0453, 0.0492, 0.0645, 0.0551], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:25:53,246 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9847, 5.3286, 5.2123, 5.2299, 5.2656, 5.2302, 4.9112, 4.7150], device='cuda:0'), covar=tensor([0.1066, 0.0440, 0.0308, 0.0500, 0.0319, 0.0318, 0.0397, 0.0346], device='cuda:0'), in_proj_covar=tensor([0.0531, 0.0375, 0.0366, 0.0375, 0.0438, 0.0444, 0.0374, 0.0409], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:26:00,926 INFO [train2.py:809] (0/4) Epoch 26, batch 3500, loss[ctc_loss=0.0754, att_loss=0.2494, loss=0.2146, over 17105.00 frames. utt_duration=1223 frames, utt_pad_proportion=0.01593, over 56.00 utterances.], tot_loss[ctc_loss=0.06633, att_loss=0.2331, loss=0.1998, over 3284031.85 frames. utt_duration=1220 frames, utt_pad_proportion=0.05637, over 10779.07 utterances.], batch size: 56, lr: 4.10e-03, grad_scale: 16.0 2023-03-09 08:26:05,809 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103098.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:26:59,848 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103132.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:27:15,600 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0900, 4.3531, 4.6374, 4.6776, 2.9954, 4.4159, 3.2435, 2.1583], device='cuda:0'), covar=tensor([0.0563, 0.0359, 0.0600, 0.0238, 0.1574, 0.0248, 0.1240, 0.1666], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0184, 0.0263, 0.0176, 0.0223, 0.0164, 0.0232, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:27:19,710 INFO [train2.py:809] (0/4) Epoch 26, batch 3550, loss[ctc_loss=0.07202, att_loss=0.2125, loss=0.1844, over 14122.00 frames. utt_duration=1824 frames, utt_pad_proportion=0.04544, over 31.00 utterances.], tot_loss[ctc_loss=0.06601, att_loss=0.2322, loss=0.1989, over 3272091.59 frames. utt_duration=1250 frames, utt_pad_proportion=0.05223, over 10483.63 utterances.], batch size: 31, lr: 4.10e-03, grad_scale: 16.0 2023-03-09 08:27:28,876 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.344e+02 1.850e+02 2.221e+02 2.819e+02 5.198e+02, threshold=4.442e+02, percent-clipped=1.0 2023-03-09 08:27:42,177 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103159.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:28:29,930 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8023, 2.6076, 2.6848, 2.5461, 2.7058, 2.7002, 2.4693, 3.1206], device='cuda:0'), covar=tensor([0.1403, 0.2081, 0.1683, 0.1427, 0.1744, 0.1153, 0.2047, 0.1112], device='cuda:0'), in_proj_covar=tensor([0.0136, 0.0140, 0.0135, 0.0130, 0.0146, 0.0125, 0.0148, 0.0124], device='cuda:0'), out_proj_covar=tensor([1.0522e-04, 1.1070e-04, 1.1000e-04, 1.0137e-04, 1.1024e-04, 1.0073e-04, 1.1310e-04, 9.8431e-05], device='cuda:0') 2023-03-09 08:28:36,260 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103193.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:28:38,761 INFO [train2.py:809] (0/4) Epoch 26, batch 3600, loss[ctc_loss=0.07549, att_loss=0.2359, loss=0.2038, over 17036.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009063, over 52.00 utterances.], tot_loss[ctc_loss=0.06543, att_loss=0.2313, loss=0.1982, over 3266071.44 frames. utt_duration=1269 frames, utt_pad_proportion=0.04946, over 10309.48 utterances.], batch size: 52, lr: 4.10e-03, grad_scale: 16.0 2023-03-09 08:28:57,710 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103207.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:29:55,676 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103243.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:29:58,624 INFO [train2.py:809] (0/4) Epoch 26, batch 3650, loss[ctc_loss=0.05136, att_loss=0.2224, loss=0.1882, over 16967.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007684, over 50.00 utterances.], tot_loss[ctc_loss=0.06499, att_loss=0.2312, loss=0.198, over 3267392.14 frames. utt_duration=1279 frames, utt_pad_proportion=0.04716, over 10226.69 utterances.], batch size: 50, lr: 4.09e-03, grad_scale: 16.0 2023-03-09 08:29:58,842 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0489, 5.2865, 5.2242, 5.2157, 5.3105, 5.2779, 4.8673, 4.7543], device='cuda:0'), covar=tensor([0.1008, 0.0539, 0.0321, 0.0508, 0.0284, 0.0325, 0.0450, 0.0356], device='cuda:0'), in_proj_covar=tensor([0.0530, 0.0376, 0.0367, 0.0375, 0.0438, 0.0444, 0.0374, 0.0409], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:30:07,783 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.026e+02 1.817e+02 2.258e+02 2.684e+02 6.066e+02, threshold=4.515e+02, percent-clipped=3.0 2023-03-09 08:30:18,790 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3516, 2.4471, 4.8214, 3.5351, 2.9932, 4.0367, 4.4650, 4.4908], device='cuda:0'), covar=tensor([0.0276, 0.1646, 0.0188, 0.1198, 0.1695, 0.0311, 0.0239, 0.0283], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0247, 0.0216, 0.0324, 0.0272, 0.0235, 0.0207, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:31:17,478 INFO [train2.py:809] (0/4) Epoch 26, batch 3700, loss[ctc_loss=0.05732, att_loss=0.2119, loss=0.181, over 15649.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008027, over 37.00 utterances.], tot_loss[ctc_loss=0.06506, att_loss=0.2308, loss=0.1976, over 3257609.42 frames. utt_duration=1283 frames, utt_pad_proportion=0.04886, over 10166.14 utterances.], batch size: 37, lr: 4.09e-03, grad_scale: 16.0 2023-03-09 08:31:55,555 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5882, 5.0305, 4.9046, 4.9952, 5.1626, 4.6859, 3.6099, 5.0222], device='cuda:0'), covar=tensor([0.0140, 0.0116, 0.0152, 0.0093, 0.0093, 0.0146, 0.0685, 0.0177], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0095, 0.0119, 0.0074, 0.0080, 0.0092, 0.0108, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-09 08:32:30,488 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7262, 5.2454, 5.0481, 5.1346, 5.3082, 4.9084, 3.6482, 5.1855], device='cuda:0'), covar=tensor([0.0134, 0.0102, 0.0133, 0.0102, 0.0099, 0.0122, 0.0694, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0095, 0.0119, 0.0074, 0.0080, 0.0091, 0.0108, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-09 08:32:36,352 INFO [train2.py:809] (0/4) Epoch 26, batch 3750, loss[ctc_loss=0.07241, att_loss=0.2368, loss=0.2039, over 17507.00 frames. utt_duration=888.1 frames, utt_pad_proportion=0.07004, over 79.00 utterances.], tot_loss[ctc_loss=0.06546, att_loss=0.2307, loss=0.1977, over 3251720.11 frames. utt_duration=1253 frames, utt_pad_proportion=0.05866, over 10389.89 utterances.], batch size: 79, lr: 4.09e-03, grad_scale: 16.0 2023-03-09 08:32:45,487 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.324e+02 1.935e+02 2.309e+02 3.024e+02 8.462e+02, threshold=4.618e+02, percent-clipped=2.0 2023-03-09 08:33:10,128 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103366.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 08:33:11,616 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7903, 3.5460, 3.5995, 3.0866, 3.4936, 3.4698, 3.5802, 2.5777], device='cuda:0'), covar=tensor([0.0975, 0.0911, 0.1705, 0.2881, 0.1095, 0.2312, 0.0885, 0.3191], device='cuda:0'), in_proj_covar=tensor([0.0199, 0.0204, 0.0220, 0.0270, 0.0180, 0.0281, 0.0201, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:33:54,197 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-09 08:33:56,328 INFO [train2.py:809] (0/4) Epoch 26, batch 3800, loss[ctc_loss=0.05464, att_loss=0.2264, loss=0.192, over 17007.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.009266, over 51.00 utterances.], tot_loss[ctc_loss=0.06545, att_loss=0.231, loss=0.1979, over 3263875.61 frames. utt_duration=1266 frames, utt_pad_proportion=0.0517, over 10323.65 utterances.], batch size: 51, lr: 4.09e-03, grad_scale: 16.0 2023-03-09 08:34:45,437 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0822, 5.1406, 4.9041, 2.3294, 2.1514, 3.0762, 2.5599, 3.9050], device='cuda:0'), covar=tensor([0.0729, 0.0398, 0.0323, 0.4894, 0.5172, 0.2366, 0.3819, 0.1675], device='cuda:0'), in_proj_covar=tensor([0.0360, 0.0298, 0.0277, 0.0249, 0.0336, 0.0329, 0.0260, 0.0366], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 08:34:48,460 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103427.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 08:34:55,131 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5683, 5.0834, 4.8790, 5.0620, 5.1684, 4.7579, 3.5150, 5.0301], device='cuda:0'), covar=tensor([0.0133, 0.0106, 0.0140, 0.0067, 0.0076, 0.0118, 0.0727, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0094, 0.0118, 0.0074, 0.0080, 0.0091, 0.0108, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:35:10,628 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1477, 3.8348, 3.8915, 3.3579, 3.8384, 3.8950, 3.9114, 2.8872], device='cuda:0'), covar=tensor([0.0905, 0.1032, 0.1270, 0.2916, 0.1534, 0.1637, 0.0857, 0.3057], device='cuda:0'), in_proj_covar=tensor([0.0197, 0.0204, 0.0218, 0.0269, 0.0179, 0.0279, 0.0200, 0.0228], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:35:16,370 INFO [train2.py:809] (0/4) Epoch 26, batch 3850, loss[ctc_loss=0.05316, att_loss=0.2367, loss=0.2, over 16974.00 frames. utt_duration=680.5 frames, utt_pad_proportion=0.1386, over 100.00 utterances.], tot_loss[ctc_loss=0.06507, att_loss=0.2311, loss=0.1979, over 3269158.23 frames. utt_duration=1268 frames, utt_pad_proportion=0.04951, over 10325.20 utterances.], batch size: 100, lr: 4.09e-03, grad_scale: 8.0 2023-03-09 08:35:26,998 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.253e+02 1.752e+02 2.146e+02 2.591e+02 4.293e+02, threshold=4.292e+02, percent-clipped=0.0 2023-03-09 08:35:30,341 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103454.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:36:22,164 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103488.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:36:32,451 INFO [train2.py:809] (0/4) Epoch 26, batch 3900, loss[ctc_loss=0.1244, att_loss=0.2734, loss=0.2436, over 14055.00 frames. utt_duration=386.6 frames, utt_pad_proportion=0.3266, over 146.00 utterances.], tot_loss[ctc_loss=0.06617, att_loss=0.2318, loss=0.1987, over 3257949.53 frames. utt_duration=1218 frames, utt_pad_proportion=0.06548, over 10708.84 utterances.], batch size: 146, lr: 4.09e-03, grad_scale: 8.0 2023-03-09 08:36:50,972 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103507.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:37:47,346 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103543.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:37:50,089 INFO [train2.py:809] (0/4) Epoch 26, batch 3950, loss[ctc_loss=0.06742, att_loss=0.2101, loss=0.1815, over 15350.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.01191, over 35.00 utterances.], tot_loss[ctc_loss=0.06621, att_loss=0.2319, loss=0.1988, over 3256465.97 frames. utt_duration=1201 frames, utt_pad_proportion=0.07098, over 10861.16 utterances.], batch size: 35, lr: 4.09e-03, grad_scale: 8.0 2023-03-09 08:38:00,645 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.092e+02 1.745e+02 2.083e+02 2.654e+02 4.442e+02, threshold=4.166e+02, percent-clipped=1.0 2023-03-09 08:38:05,197 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103555.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:38:41,285 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-26.pt 2023-03-09 08:39:05,360 INFO [train2.py:809] (0/4) Epoch 27, batch 0, loss[ctc_loss=0.06542, att_loss=0.2241, loss=0.1924, over 16012.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007017, over 40.00 utterances.], tot_loss[ctc_loss=0.06542, att_loss=0.2241, loss=0.1924, over 16012.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007017, over 40.00 utterances.], batch size: 40, lr: 4.01e-03, grad_scale: 8.0 2023-03-09 08:39:05,362 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 08:39:17,375 INFO [train2.py:843] (0/4) Epoch 27, validation: ctc_loss=0.04075, att_loss=0.2342, loss=0.1955, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 08:39:17,376 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 08:39:36,970 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103591.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:40:35,289 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1701, 4.3047, 4.3719, 4.3509, 4.8787, 4.2477, 4.3684, 2.4361], device='cuda:0'), covar=tensor([0.0336, 0.0460, 0.0431, 0.0362, 0.0661, 0.0301, 0.0378, 0.1826], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0219, 0.0215, 0.0232, 0.0389, 0.0190, 0.0203, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:40:36,383 INFO [train2.py:809] (0/4) Epoch 27, batch 50, loss[ctc_loss=0.06944, att_loss=0.2523, loss=0.2157, over 17273.00 frames. utt_duration=1098 frames, utt_pad_proportion=0.0392, over 63.00 utterances.], tot_loss[ctc_loss=0.06713, att_loss=0.233, loss=0.1998, over 735954.24 frames. utt_duration=1165 frames, utt_pad_proportion=0.07053, over 2529.20 utterances.], batch size: 63, lr: 4.01e-03, grad_scale: 8.0 2023-03-09 08:41:14,259 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.229e+02 1.910e+02 2.259e+02 2.713e+02 6.930e+02, threshold=4.518e+02, percent-clipped=6.0 2023-03-09 08:41:42,430 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4308, 2.7796, 3.6276, 2.9471, 3.4460, 4.5118, 4.3854, 3.2642], device='cuda:0'), covar=tensor([0.0373, 0.1790, 0.1192, 0.1343, 0.1152, 0.0913, 0.0586, 0.1161], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0251, 0.0289, 0.0222, 0.0270, 0.0381, 0.0273, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:41:56,143 INFO [train2.py:809] (0/4) Epoch 27, batch 100, loss[ctc_loss=0.07439, att_loss=0.2441, loss=0.2102, over 17400.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03371, over 63.00 utterances.], tot_loss[ctc_loss=0.06609, att_loss=0.232, loss=0.1989, over 1297036.79 frames. utt_duration=1196 frames, utt_pad_proportion=0.06828, over 4342.05 utterances.], batch size: 63, lr: 4.01e-03, grad_scale: 8.0 2023-03-09 08:42:39,752 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-03-09 08:43:04,831 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=103722.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 08:43:15,220 INFO [train2.py:809] (0/4) Epoch 27, batch 150, loss[ctc_loss=0.04057, att_loss=0.1962, loss=0.165, over 16014.00 frames. utt_duration=1603 frames, utt_pad_proportion=0.007508, over 40.00 utterances.], tot_loss[ctc_loss=0.06619, att_loss=0.2323, loss=0.1991, over 1735337.73 frames. utt_duration=1272 frames, utt_pad_proportion=0.04947, over 5464.60 utterances.], batch size: 40, lr: 4.01e-03, grad_scale: 8.0 2023-03-09 08:43:33,632 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103740.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:43:52,275 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 1.770e+02 2.085e+02 2.452e+02 6.368e+02, threshold=4.170e+02, percent-clipped=1.0 2023-03-09 08:43:54,971 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-03-09 08:43:56,611 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103754.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:44:13,413 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103765.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:44:34,221 INFO [train2.py:809] (0/4) Epoch 27, batch 200, loss[ctc_loss=0.05576, att_loss=0.2309, loss=0.1959, over 17024.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007671, over 51.00 utterances.], tot_loss[ctc_loss=0.06681, att_loss=0.2333, loss=0.2, over 2084058.37 frames. utt_duration=1253 frames, utt_pad_proportion=0.05056, over 6659.48 utterances.], batch size: 51, lr: 4.01e-03, grad_scale: 8.0 2023-03-09 08:44:49,580 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=103788.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:45:10,661 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103801.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:45:11,912 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103802.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:45:49,610 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103826.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:45:53,752 INFO [train2.py:809] (0/4) Epoch 27, batch 250, loss[ctc_loss=0.07519, att_loss=0.2433, loss=0.2097, over 17393.00 frames. utt_duration=1106 frames, utt_pad_proportion=0.03187, over 63.00 utterances.], tot_loss[ctc_loss=0.06682, att_loss=0.2333, loss=0.2, over 2351647.28 frames. utt_duration=1248 frames, utt_pad_proportion=0.05128, over 7544.00 utterances.], batch size: 63, lr: 4.01e-03, grad_scale: 8.0 2023-03-09 08:46:02,588 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0438, 5.0536, 5.0611, 1.8116, 2.0037, 2.5721, 2.1192, 3.7343], device='cuda:0'), covar=tensor([0.0898, 0.0426, 0.0249, 0.5339, 0.6289, 0.3333, 0.4352, 0.1926], device='cuda:0'), in_proj_covar=tensor([0.0364, 0.0300, 0.0280, 0.0252, 0.0340, 0.0333, 0.0262, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 08:46:05,215 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=103836.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:46:30,829 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.121e+02 1.879e+02 2.239e+02 2.738e+02 5.518e+02, threshold=4.478e+02, percent-clipped=1.0 2023-03-09 08:46:42,396 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0599, 5.3519, 5.2863, 5.3034, 5.3769, 5.3131, 5.0069, 4.8268], device='cuda:0'), covar=tensor([0.0979, 0.0457, 0.0272, 0.0520, 0.0278, 0.0308, 0.0412, 0.0313], device='cuda:0'), in_proj_covar=tensor([0.0531, 0.0378, 0.0368, 0.0375, 0.0439, 0.0442, 0.0376, 0.0407], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:47:08,874 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103876.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:47:13,076 INFO [train2.py:809] (0/4) Epoch 27, batch 300, loss[ctc_loss=0.05484, att_loss=0.2356, loss=0.1995, over 16637.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004617, over 47.00 utterances.], tot_loss[ctc_loss=0.0666, att_loss=0.233, loss=0.1997, over 2559085.09 frames. utt_duration=1252 frames, utt_pad_proportion=0.04983, over 8182.97 utterances.], batch size: 47, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:48:32,099 INFO [train2.py:809] (0/4) Epoch 27, batch 350, loss[ctc_loss=0.07114, att_loss=0.2452, loss=0.2104, over 16625.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005364, over 47.00 utterances.], tot_loss[ctc_loss=0.06661, att_loss=0.2324, loss=0.1992, over 2716028.66 frames. utt_duration=1253 frames, utt_pad_proportion=0.05202, over 8683.29 utterances.], batch size: 47, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:48:45,693 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=103937.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 08:48:45,766 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2515, 5.1825, 5.0973, 2.4650, 2.0729, 3.1950, 2.2890, 4.0496], device='cuda:0'), covar=tensor([0.0669, 0.0387, 0.0235, 0.4728, 0.5396, 0.2225, 0.3976, 0.1582], device='cuda:0'), in_proj_covar=tensor([0.0365, 0.0301, 0.0280, 0.0252, 0.0341, 0.0334, 0.0264, 0.0372], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 08:48:54,454 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6522, 5.9811, 5.5486, 5.6705, 5.6470, 5.0341, 5.3490, 5.1413], device='cuda:0'), covar=tensor([0.1362, 0.0811, 0.0969, 0.0823, 0.0953, 0.1687, 0.2104, 0.2133], device='cuda:0'), in_proj_covar=tensor([0.0553, 0.0632, 0.0486, 0.0477, 0.0449, 0.0482, 0.0637, 0.0546], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:49:08,958 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.349e+02 1.861e+02 2.238e+02 2.752e+02 4.825e+02, threshold=4.477e+02, percent-clipped=1.0 2023-03-09 08:49:51,233 INFO [train2.py:809] (0/4) Epoch 27, batch 400, loss[ctc_loss=0.0443, att_loss=0.2215, loss=0.186, over 16463.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.007588, over 46.00 utterances.], tot_loss[ctc_loss=0.06619, att_loss=0.2317, loss=0.1986, over 2835612.04 frames. utt_duration=1269 frames, utt_pad_proportion=0.04925, over 8948.38 utterances.], batch size: 46, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:50:02,161 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6769, 3.0003, 3.9321, 3.0882, 3.7605, 4.8391, 4.6324, 3.4570], device='cuda:0'), covar=tensor([0.0351, 0.1737, 0.1000, 0.1326, 0.0905, 0.0702, 0.0452, 0.1119], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0250, 0.0288, 0.0221, 0.0270, 0.0379, 0.0272, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:50:08,999 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103990.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:50:19,664 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=103997.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:50:24,872 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-104000.pt 2023-03-09 08:51:03,234 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4148, 2.8909, 3.3020, 4.3658, 3.8488, 3.8840, 3.0075, 2.1825], device='cuda:0'), covar=tensor([0.0724, 0.1852, 0.0892, 0.0585, 0.0931, 0.0501, 0.1509, 0.2276], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0219, 0.0187, 0.0224, 0.0233, 0.0190, 0.0204, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:51:04,722 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104022.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 08:51:15,260 INFO [train2.py:809] (0/4) Epoch 27, batch 450, loss[ctc_loss=0.05812, att_loss=0.2375, loss=0.2016, over 17027.00 frames. utt_duration=682.4 frames, utt_pad_proportion=0.134, over 100.00 utterances.], tot_loss[ctc_loss=0.06613, att_loss=0.2313, loss=0.1983, over 2937190.37 frames. utt_duration=1268 frames, utt_pad_proportion=0.04683, over 9278.35 utterances.], batch size: 100, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:51:48,274 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=104049.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:51:51,406 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104051.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:51:52,513 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 1.790e+02 2.279e+02 2.671e+02 6.527e+02, threshold=4.558e+02, percent-clipped=3.0 2023-03-09 08:52:02,864 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104058.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:52:20,934 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104070.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 08:52:21,674 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 08:52:27,074 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0851, 5.3563, 5.5886, 5.3656, 5.6077, 6.0280, 5.2808, 6.0816], device='cuda:0'), covar=tensor([0.0632, 0.0678, 0.0778, 0.1307, 0.1658, 0.0820, 0.0639, 0.0701], device='cuda:0'), in_proj_covar=tensor([0.0901, 0.0522, 0.0631, 0.0681, 0.0898, 0.0659, 0.0506, 0.0638], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:52:27,138 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0310, 5.3469, 4.9411, 5.4361, 4.7628, 5.0846, 5.4962, 5.2266], device='cuda:0'), covar=tensor([0.0671, 0.0312, 0.0781, 0.0308, 0.0433, 0.0228, 0.0216, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0402, 0.0337, 0.0380, 0.0371, 0.0337, 0.0247, 0.0318, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 08:52:34,616 INFO [train2.py:809] (0/4) Epoch 27, batch 500, loss[ctc_loss=0.06211, att_loss=0.2235, loss=0.1912, over 15934.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.00637, over 41.00 utterances.], tot_loss[ctc_loss=0.06556, att_loss=0.2316, loss=0.1984, over 3013280.99 frames. utt_duration=1273 frames, utt_pad_proportion=0.04572, over 9476.72 utterances.], batch size: 41, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:53:02,099 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104096.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:53:23,993 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=104110.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:53:40,950 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104121.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:53:53,121 INFO [train2.py:809] (0/4) Epoch 27, batch 550, loss[ctc_loss=0.06397, att_loss=0.2399, loss=0.2047, over 16953.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.00769, over 50.00 utterances.], tot_loss[ctc_loss=0.0654, att_loss=0.2314, loss=0.1982, over 3074504.01 frames. utt_duration=1290 frames, utt_pad_proportion=0.04253, over 9541.49 utterances.], batch size: 50, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:54:09,443 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0981, 4.3589, 4.6855, 4.5328, 3.0306, 4.4070, 3.0783, 2.0331], device='cuda:0'), covar=tensor([0.0437, 0.0344, 0.0552, 0.0356, 0.1340, 0.0288, 0.1250, 0.1531], device='cuda:0'), in_proj_covar=tensor([0.0213, 0.0184, 0.0263, 0.0177, 0.0221, 0.0163, 0.0232, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 08:54:29,376 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.176e+02 1.746e+02 2.054e+02 2.540e+02 6.808e+02, threshold=4.109e+02, percent-clipped=2.0 2023-03-09 08:54:58,956 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6689, 3.2001, 3.8869, 3.1712, 3.6600, 4.8363, 4.6141, 3.3848], device='cuda:0'), covar=tensor([0.0356, 0.1581, 0.1131, 0.1297, 0.1156, 0.0614, 0.0588, 0.1149], device='cuda:0'), in_proj_covar=tensor([0.0249, 0.0250, 0.0289, 0.0222, 0.0271, 0.0380, 0.0274, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:55:07,884 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6309, 5.9565, 5.4564, 5.6683, 5.6098, 5.0493, 5.2984, 5.1254], device='cuda:0'), covar=tensor([0.1379, 0.0927, 0.1100, 0.0830, 0.0981, 0.1704, 0.2552, 0.2381], device='cuda:0'), in_proj_covar=tensor([0.0554, 0.0633, 0.0488, 0.0478, 0.0449, 0.0483, 0.0640, 0.0547], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:55:10,796 INFO [train2.py:809] (0/4) Epoch 27, batch 600, loss[ctc_loss=0.06213, att_loss=0.234, loss=0.1996, over 16551.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.005165, over 45.00 utterances.], tot_loss[ctc_loss=0.06526, att_loss=0.2313, loss=0.1981, over 3121745.13 frames. utt_duration=1262 frames, utt_pad_proportion=0.04799, over 9903.81 utterances.], batch size: 45, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:56:29,435 INFO [train2.py:809] (0/4) Epoch 27, batch 650, loss[ctc_loss=0.05785, att_loss=0.2143, loss=0.183, over 16140.00 frames. utt_duration=1539 frames, utt_pad_proportion=0.005233, over 42.00 utterances.], tot_loss[ctc_loss=0.06578, att_loss=0.2314, loss=0.1983, over 3158393.17 frames. utt_duration=1252 frames, utt_pad_proportion=0.04945, over 10100.07 utterances.], batch size: 42, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:56:34,206 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104232.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 08:57:05,923 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.363e+02 1.860e+02 2.240e+02 2.877e+02 8.849e+02, threshold=4.479e+02, percent-clipped=3.0 2023-03-09 08:57:29,952 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7613, 3.9531, 3.8974, 3.9335, 3.9892, 3.8042, 3.0961, 3.8915], device='cuda:0'), covar=tensor([0.0156, 0.0146, 0.0176, 0.0106, 0.0118, 0.0147, 0.0647, 0.0238], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0094, 0.0118, 0.0074, 0.0079, 0.0090, 0.0108, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 08:57:47,592 INFO [train2.py:809] (0/4) Epoch 27, batch 700, loss[ctc_loss=0.06414, att_loss=0.2455, loss=0.2092, over 17040.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009463, over 53.00 utterances.], tot_loss[ctc_loss=0.06606, att_loss=0.2315, loss=0.1984, over 3181043.56 frames. utt_duration=1241 frames, utt_pad_proportion=0.05438, over 10262.89 utterances.], batch size: 53, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:58:36,045 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0173, 5.0316, 4.7762, 2.9761, 4.8215, 4.6380, 4.3274, 2.8838], device='cuda:0'), covar=tensor([0.0122, 0.0106, 0.0273, 0.0996, 0.0120, 0.0210, 0.0311, 0.1254], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0105, 0.0109, 0.0112, 0.0089, 0.0117, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 08:59:06,612 INFO [train2.py:809] (0/4) Epoch 27, batch 750, loss[ctc_loss=0.04702, att_loss=0.2156, loss=0.1819, over 16272.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007554, over 43.00 utterances.], tot_loss[ctc_loss=0.0659, att_loss=0.232, loss=0.1988, over 3205601.91 frames. utt_duration=1251 frames, utt_pad_proportion=0.0527, over 10265.14 utterances.], batch size: 43, lr: 4.00e-03, grad_scale: 8.0 2023-03-09 08:59:34,715 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104346.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:59:40,796 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0391, 5.3558, 4.9848, 5.4366, 4.7826, 5.0540, 5.4807, 5.2783], device='cuda:0'), covar=tensor([0.0633, 0.0289, 0.0759, 0.0325, 0.0454, 0.0260, 0.0208, 0.0200], device='cuda:0'), in_proj_covar=tensor([0.0402, 0.0338, 0.0381, 0.0373, 0.0339, 0.0247, 0.0318, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 08:59:43,522 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 1.900e+02 2.289e+02 2.673e+02 5.162e+02, threshold=4.578e+02, percent-clipped=1.0 2023-03-09 08:59:45,858 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104353.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 08:59:52,077 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9055, 4.9579, 4.5814, 2.6021, 4.7221, 4.6465, 3.9424, 2.4247], device='cuda:0'), covar=tensor([0.0174, 0.0134, 0.0385, 0.1315, 0.0133, 0.0243, 0.0471, 0.1722], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0105, 0.0109, 0.0112, 0.0089, 0.0117, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:00:26,055 INFO [train2.py:809] (0/4) Epoch 27, batch 800, loss[ctc_loss=0.04717, att_loss=0.2053, loss=0.1736, over 15774.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.009007, over 38.00 utterances.], tot_loss[ctc_loss=0.06486, att_loss=0.2311, loss=0.1978, over 3218459.44 frames. utt_duration=1261 frames, utt_pad_proportion=0.05102, over 10217.50 utterances.], batch size: 38, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:00:26,695 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7017, 2.4951, 2.4778, 2.4399, 2.8268, 2.8989, 2.4762, 3.0001], device='cuda:0'), covar=tensor([0.1757, 0.2225, 0.1692, 0.1711, 0.1499, 0.1028, 0.2132, 0.1398], device='cuda:0'), in_proj_covar=tensor([0.0139, 0.0143, 0.0137, 0.0133, 0.0149, 0.0127, 0.0151, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 09:00:54,102 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104396.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:01:08,570 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=104405.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:01:33,275 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104421.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:01:46,497 INFO [train2.py:809] (0/4) Epoch 27, batch 850, loss[ctc_loss=0.06366, att_loss=0.2143, loss=0.1842, over 15761.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009776, over 38.00 utterances.], tot_loss[ctc_loss=0.06463, att_loss=0.2308, loss=0.1976, over 3225665.59 frames. utt_duration=1258 frames, utt_pad_proportion=0.05412, over 10272.36 utterances.], batch size: 38, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:01:56,631 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1767, 5.4591, 5.0661, 5.5173, 4.9390, 5.1359, 5.5948, 5.3836], device='cuda:0'), covar=tensor([0.0552, 0.0288, 0.0772, 0.0273, 0.0371, 0.0248, 0.0190, 0.0167], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0336, 0.0380, 0.0373, 0.0337, 0.0246, 0.0318, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 09:02:06,053 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0344, 5.2964, 5.5415, 5.3325, 5.5671, 6.0152, 5.3431, 6.0803], device='cuda:0'), covar=tensor([0.0779, 0.0757, 0.0892, 0.1472, 0.1838, 0.0848, 0.0694, 0.0676], device='cuda:0'), in_proj_covar=tensor([0.0907, 0.0526, 0.0638, 0.0687, 0.0903, 0.0664, 0.0510, 0.0643], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 09:02:10,582 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104444.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:02:23,113 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.335e+02 1.830e+02 2.139e+02 2.642e+02 5.994e+02, threshold=4.278e+02, percent-clipped=1.0 2023-03-09 09:02:49,615 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104469.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:03:05,896 INFO [train2.py:809] (0/4) Epoch 27, batch 900, loss[ctc_loss=0.05382, att_loss=0.207, loss=0.1763, over 15637.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.008645, over 37.00 utterances.], tot_loss[ctc_loss=0.06571, att_loss=0.2316, loss=0.1984, over 3242680.49 frames. utt_duration=1228 frames, utt_pad_proportion=0.05839, over 10571.17 utterances.], batch size: 37, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:04:24,927 INFO [train2.py:809] (0/4) Epoch 27, batch 950, loss[ctc_loss=0.07287, att_loss=0.2376, loss=0.2046, over 17373.00 frames. utt_duration=1104 frames, utt_pad_proportion=0.03458, over 63.00 utterances.], tot_loss[ctc_loss=0.06584, att_loss=0.2318, loss=0.1986, over 3257500.59 frames. utt_duration=1250 frames, utt_pad_proportion=0.05132, over 10440.43 utterances.], batch size: 63, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:04:29,752 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104532.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 09:05:01,190 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.309e+02 1.806e+02 2.235e+02 2.788e+02 5.642e+02, threshold=4.470e+02, percent-clipped=2.0 2023-03-09 09:05:43,839 INFO [train2.py:809] (0/4) Epoch 27, batch 1000, loss[ctc_loss=0.08613, att_loss=0.2483, loss=0.2159, over 17358.00 frames. utt_duration=880.4 frames, utt_pad_proportion=0.07619, over 79.00 utterances.], tot_loss[ctc_loss=0.06651, att_loss=0.2323, loss=0.1991, over 3263188.01 frames. utt_duration=1244 frames, utt_pad_proportion=0.05288, over 10508.89 utterances.], batch size: 79, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:05:45,418 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104580.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:07:03,583 INFO [train2.py:809] (0/4) Epoch 27, batch 1050, loss[ctc_loss=0.07965, att_loss=0.2415, loss=0.2092, over 17292.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02487, over 59.00 utterances.], tot_loss[ctc_loss=0.0662, att_loss=0.2322, loss=0.199, over 3263826.57 frames. utt_duration=1226 frames, utt_pad_proportion=0.0581, over 10665.27 utterances.], batch size: 59, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:07:30,266 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104646.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:07:40,080 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.329e+02 1.873e+02 2.248e+02 2.692e+02 7.712e+02, threshold=4.496e+02, percent-clipped=6.0 2023-03-09 09:07:41,962 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104653.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:08:22,948 INFO [train2.py:809] (0/4) Epoch 27, batch 1100, loss[ctc_loss=0.08546, att_loss=0.2577, loss=0.2232, over 17283.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01227, over 55.00 utterances.], tot_loss[ctc_loss=0.06554, att_loss=0.2316, loss=0.1984, over 3263837.64 frames. utt_duration=1239 frames, utt_pad_proportion=0.05511, over 10549.84 utterances.], batch size: 55, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:08:46,565 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104694.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:08:58,150 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104701.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:08:58,305 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2219, 4.5177, 4.4987, 4.5382, 4.5039, 4.3700, 3.0977, 4.4424], device='cuda:0'), covar=tensor([0.0170, 0.0188, 0.0201, 0.0125, 0.0154, 0.0153, 0.0955, 0.0386], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0092, 0.0117, 0.0073, 0.0078, 0.0089, 0.0107, 0.0110], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 09:08:59,756 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2425, 5.1460, 4.9496, 3.2381, 4.9683, 4.8255, 4.6071, 2.8791], device='cuda:0'), covar=tensor([0.0120, 0.0117, 0.0275, 0.0902, 0.0111, 0.0198, 0.0252, 0.1330], device='cuda:0'), in_proj_covar=tensor([0.0078, 0.0106, 0.0109, 0.0112, 0.0090, 0.0117, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:09:04,420 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=104705.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:09:13,682 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1144, 4.4157, 4.3258, 4.6126, 2.9507, 4.4565, 2.6391, 1.7670], device='cuda:0'), covar=tensor([0.0514, 0.0323, 0.0739, 0.0283, 0.1692, 0.0254, 0.1587, 0.1907], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0187, 0.0268, 0.0179, 0.0225, 0.0167, 0.0236, 0.0208], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:09:42,147 INFO [train2.py:809] (0/4) Epoch 27, batch 1150, loss[ctc_loss=0.07844, att_loss=0.2451, loss=0.2118, over 17278.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01345, over 55.00 utterances.], tot_loss[ctc_loss=0.0658, att_loss=0.2322, loss=0.1989, over 3277348.11 frames. utt_duration=1245 frames, utt_pad_proportion=0.05088, over 10540.03 utterances.], batch size: 55, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:10:19,017 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 1.763e+02 2.147e+02 2.589e+02 3.796e+02, threshold=4.293e+02, percent-clipped=0.0 2023-03-09 09:10:20,705 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=104753.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:11:01,665 INFO [train2.py:809] (0/4) Epoch 27, batch 1200, loss[ctc_loss=0.06714, att_loss=0.2343, loss=0.2009, over 17049.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.00992, over 53.00 utterances.], tot_loss[ctc_loss=0.06542, att_loss=0.2313, loss=0.1982, over 3273999.53 frames. utt_duration=1286 frames, utt_pad_proportion=0.04337, over 10198.18 utterances.], batch size: 53, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:11:08,885 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3421, 5.5865, 5.5077, 5.5370, 5.6059, 5.5933, 5.2369, 5.0541], device='cuda:0'), covar=tensor([0.1020, 0.0493, 0.0273, 0.0390, 0.0290, 0.0278, 0.0408, 0.0317], device='cuda:0'), in_proj_covar=tensor([0.0542, 0.0383, 0.0375, 0.0382, 0.0445, 0.0449, 0.0382, 0.0414], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:11:19,250 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0502, 6.2811, 5.8068, 5.9970, 6.0197, 5.4089, 5.7105, 5.4648], device='cuda:0'), covar=tensor([0.1304, 0.0855, 0.0903, 0.0893, 0.0902, 0.1543, 0.2297, 0.2514], device='cuda:0'), in_proj_covar=tensor([0.0551, 0.0631, 0.0485, 0.0477, 0.0447, 0.0480, 0.0638, 0.0545], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:12:21,905 INFO [train2.py:809] (0/4) Epoch 27, batch 1250, loss[ctc_loss=0.07268, att_loss=0.2428, loss=0.2088, over 16466.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006674, over 46.00 utterances.], tot_loss[ctc_loss=0.06624, att_loss=0.232, loss=0.1988, over 3277083.11 frames. utt_duration=1279 frames, utt_pad_proportion=0.04516, over 10258.96 utterances.], batch size: 46, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:12:58,970 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 1.878e+02 2.209e+02 2.737e+02 5.339e+02, threshold=4.418e+02, percent-clipped=1.0 2023-03-09 09:13:42,093 INFO [train2.py:809] (0/4) Epoch 27, batch 1300, loss[ctc_loss=0.0745, att_loss=0.2441, loss=0.2102, over 16880.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.007459, over 49.00 utterances.], tot_loss[ctc_loss=0.06654, att_loss=0.2327, loss=0.1994, over 3277446.70 frames. utt_duration=1245 frames, utt_pad_proportion=0.05393, over 10543.29 utterances.], batch size: 49, lr: 3.99e-03, grad_scale: 8.0 2023-03-09 09:14:38,573 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8599, 5.1477, 5.0183, 5.0451, 5.1511, 5.1583, 4.7556, 4.6071], device='cuda:0'), covar=tensor([0.1076, 0.0553, 0.0439, 0.0559, 0.0321, 0.0316, 0.0494, 0.0385], device='cuda:0'), in_proj_covar=tensor([0.0549, 0.0387, 0.0380, 0.0387, 0.0450, 0.0455, 0.0387, 0.0419], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:15:01,660 INFO [train2.py:809] (0/4) Epoch 27, batch 1350, loss[ctc_loss=0.06448, att_loss=0.2301, loss=0.197, over 17334.00 frames. utt_duration=879.1 frames, utt_pad_proportion=0.07847, over 79.00 utterances.], tot_loss[ctc_loss=0.06672, att_loss=0.2336, loss=0.2003, over 3288347.59 frames. utt_duration=1228 frames, utt_pad_proportion=0.05489, over 10722.90 utterances.], batch size: 79, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:15:11,472 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6166, 5.0190, 4.8117, 5.0076, 5.0730, 4.7197, 3.2712, 4.9625], device='cuda:0'), covar=tensor([0.0133, 0.0114, 0.0152, 0.0072, 0.0091, 0.0110, 0.0850, 0.0184], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0093, 0.0117, 0.0073, 0.0079, 0.0090, 0.0107, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 09:15:37,626 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.130e+02 1.826e+02 2.285e+02 3.088e+02 1.440e+03, threshold=4.571e+02, percent-clipped=4.0 2023-03-09 09:16:20,668 INFO [train2.py:809] (0/4) Epoch 27, batch 1400, loss[ctc_loss=0.05792, att_loss=0.2003, loss=0.1719, over 15625.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.008465, over 37.00 utterances.], tot_loss[ctc_loss=0.06706, att_loss=0.2333, loss=0.2001, over 3286586.05 frames. utt_duration=1229 frames, utt_pad_proportion=0.05484, over 10711.43 utterances.], batch size: 37, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:16:37,014 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5051, 3.0660, 3.3400, 4.4918, 3.9967, 4.0297, 3.1564, 2.5175], device='cuda:0'), covar=tensor([0.0607, 0.1661, 0.0841, 0.0483, 0.0868, 0.0411, 0.1238, 0.1895], device='cuda:0'), in_proj_covar=tensor([0.0185, 0.0215, 0.0186, 0.0220, 0.0230, 0.0188, 0.0200, 0.0189], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:17:40,902 INFO [train2.py:809] (0/4) Epoch 27, batch 1450, loss[ctc_loss=0.06992, att_loss=0.2484, loss=0.2127, over 17032.00 frames. utt_duration=1338 frames, utt_pad_proportion=0.00768, over 51.00 utterances.], tot_loss[ctc_loss=0.06708, att_loss=0.2334, loss=0.2002, over 3279418.08 frames. utt_duration=1211 frames, utt_pad_proportion=0.06191, over 10848.18 utterances.], batch size: 51, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:18:16,914 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.239e+02 1.845e+02 2.127e+02 2.675e+02 5.625e+02, threshold=4.253e+02, percent-clipped=2.0 2023-03-09 09:19:00,492 INFO [train2.py:809] (0/4) Epoch 27, batch 1500, loss[ctc_loss=0.08426, att_loss=0.2533, loss=0.2195, over 16464.00 frames. utt_duration=1433 frames, utt_pad_proportion=0.006794, over 46.00 utterances.], tot_loss[ctc_loss=0.06637, att_loss=0.2334, loss=0.2, over 3285763.20 frames. utt_duration=1211 frames, utt_pad_proportion=0.06051, over 10865.65 utterances.], batch size: 46, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:19:22,284 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1379, 5.4331, 5.0283, 5.4670, 4.8338, 5.0318, 5.5418, 5.3376], device='cuda:0'), covar=tensor([0.0514, 0.0256, 0.0712, 0.0286, 0.0392, 0.0251, 0.0196, 0.0170], device='cuda:0'), in_proj_covar=tensor([0.0400, 0.0336, 0.0379, 0.0374, 0.0338, 0.0246, 0.0317, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 09:20:18,685 INFO [train2.py:809] (0/4) Epoch 27, batch 1550, loss[ctc_loss=0.0576, att_loss=0.2057, loss=0.1761, over 15494.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.009201, over 36.00 utterances.], tot_loss[ctc_loss=0.06639, att_loss=0.2327, loss=0.1995, over 3279253.35 frames. utt_duration=1235 frames, utt_pad_proportion=0.05626, over 10638.04 utterances.], batch size: 36, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:20:55,284 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.109e+02 1.873e+02 2.333e+02 2.814e+02 5.399e+02, threshold=4.665e+02, percent-clipped=5.0 2023-03-09 09:21:38,743 INFO [train2.py:809] (0/4) Epoch 27, batch 1600, loss[ctc_loss=0.1128, att_loss=0.265, loss=0.2345, over 17313.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02382, over 59.00 utterances.], tot_loss[ctc_loss=0.06708, att_loss=0.2332, loss=0.2, over 3280011.26 frames. utt_duration=1228 frames, utt_pad_proportion=0.05739, over 10696.56 utterances.], batch size: 59, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:22:58,646 INFO [train2.py:809] (0/4) Epoch 27, batch 1650, loss[ctc_loss=0.04761, att_loss=0.2275, loss=0.1916, over 16632.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.004617, over 47.00 utterances.], tot_loss[ctc_loss=0.06745, att_loss=0.2334, loss=0.2002, over 3277574.43 frames. utt_duration=1223 frames, utt_pad_proportion=0.06022, over 10732.40 utterances.], batch size: 47, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:23:34,349 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 2.028e+02 2.397e+02 2.825e+02 1.110e+03, threshold=4.794e+02, percent-clipped=4.0 2023-03-09 09:24:17,719 INFO [train2.py:809] (0/4) Epoch 27, batch 1700, loss[ctc_loss=0.06452, att_loss=0.2175, loss=0.1869, over 15751.00 frames. utt_duration=1659 frames, utt_pad_proportion=0.01045, over 38.00 utterances.], tot_loss[ctc_loss=0.06728, att_loss=0.2326, loss=0.1995, over 3277157.69 frames. utt_duration=1235 frames, utt_pad_proportion=0.05741, over 10628.12 utterances.], batch size: 38, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:25:06,355 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.93 vs. limit=2.0 2023-03-09 09:25:35,430 INFO [train2.py:809] (0/4) Epoch 27, batch 1750, loss[ctc_loss=0.07417, att_loss=0.2416, loss=0.2081, over 16868.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.00764, over 49.00 utterances.], tot_loss[ctc_loss=0.06717, att_loss=0.2322, loss=0.1992, over 3269218.57 frames. utt_duration=1235 frames, utt_pad_proportion=0.06028, over 10604.55 utterances.], batch size: 49, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:25:37,128 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7915, 6.0477, 5.4679, 5.8114, 5.6948, 5.1996, 5.4872, 5.2447], device='cuda:0'), covar=tensor([0.1431, 0.0927, 0.1012, 0.0866, 0.1083, 0.1724, 0.2490, 0.2640], device='cuda:0'), in_proj_covar=tensor([0.0559, 0.0641, 0.0489, 0.0479, 0.0455, 0.0486, 0.0646, 0.0553], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:26:11,489 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.189e+02 1.890e+02 2.121e+02 2.571e+02 3.982e+02, threshold=4.242e+02, percent-clipped=0.0 2023-03-09 09:26:51,035 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.66 vs. limit=5.0 2023-03-09 09:26:54,621 INFO [train2.py:809] (0/4) Epoch 27, batch 1800, loss[ctc_loss=0.05229, att_loss=0.218, loss=0.1849, over 16163.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.0066, over 41.00 utterances.], tot_loss[ctc_loss=0.06671, att_loss=0.232, loss=0.1989, over 3271539.94 frames. utt_duration=1231 frames, utt_pad_proportion=0.06072, over 10645.41 utterances.], batch size: 41, lr: 3.98e-03, grad_scale: 8.0 2023-03-09 09:27:04,272 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4535, 4.4539, 4.5379, 4.6653, 5.1184, 4.4894, 4.4530, 2.6881], device='cuda:0'), covar=tensor([0.0282, 0.0394, 0.0381, 0.0288, 0.0622, 0.0271, 0.0407, 0.1573], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0220, 0.0216, 0.0231, 0.0384, 0.0190, 0.0205, 0.0222], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:27:08,758 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105388.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:28:01,939 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9732, 5.2484, 5.5091, 5.3837, 5.5092, 5.9683, 5.2365, 6.0104], device='cuda:0'), covar=tensor([0.0820, 0.0757, 0.0803, 0.1357, 0.1983, 0.0886, 0.0673, 0.0759], device='cuda:0'), in_proj_covar=tensor([0.0910, 0.0523, 0.0637, 0.0684, 0.0904, 0.0659, 0.0508, 0.0640], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 09:28:15,235 INFO [train2.py:809] (0/4) Epoch 27, batch 1850, loss[ctc_loss=0.07614, att_loss=0.2416, loss=0.2085, over 17032.00 frames. utt_duration=1287 frames, utt_pad_proportion=0.01022, over 53.00 utterances.], tot_loss[ctc_loss=0.06573, att_loss=0.2311, loss=0.198, over 3268442.53 frames. utt_duration=1246 frames, utt_pad_proportion=0.05856, over 10507.85 utterances.], batch size: 53, lr: 3.97e-03, grad_scale: 16.0 2023-03-09 09:28:47,112 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105449.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 09:28:51,301 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.205e+02 1.738e+02 2.150e+02 2.856e+02 4.488e+02, threshold=4.299e+02, percent-clipped=4.0 2023-03-09 09:29:11,407 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8473, 6.1091, 5.4985, 5.8101, 5.7870, 5.2015, 5.5054, 5.2908], device='cuda:0'), covar=tensor([0.1294, 0.0817, 0.0916, 0.0854, 0.0968, 0.1729, 0.2295, 0.2312], device='cuda:0'), in_proj_covar=tensor([0.0561, 0.0643, 0.0491, 0.0481, 0.0455, 0.0487, 0.0647, 0.0552], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:29:34,708 INFO [train2.py:809] (0/4) Epoch 27, batch 1900, loss[ctc_loss=0.08659, att_loss=0.252, loss=0.2189, over 14056.00 frames. utt_duration=384 frames, utt_pad_proportion=0.3275, over 147.00 utterances.], tot_loss[ctc_loss=0.06559, att_loss=0.2312, loss=0.1981, over 3263917.11 frames. utt_duration=1243 frames, utt_pad_proportion=0.05931, over 10516.51 utterances.], batch size: 147, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:29:45,599 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105486.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:30:11,402 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105502.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:30:54,639 INFO [train2.py:809] (0/4) Epoch 27, batch 1950, loss[ctc_loss=0.05367, att_loss=0.2216, loss=0.188, over 16538.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006272, over 45.00 utterances.], tot_loss[ctc_loss=0.0652, att_loss=0.2312, loss=0.198, over 3268632.35 frames. utt_duration=1261 frames, utt_pad_proportion=0.05267, over 10383.35 utterances.], batch size: 45, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:31:23,662 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105547.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:31:32,461 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.251e+02 1.913e+02 2.076e+02 2.767e+02 9.695e+02, threshold=4.152e+02, percent-clipped=4.0 2023-03-09 09:31:49,348 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105563.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:32:04,772 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.84 vs. limit=5.0 2023-03-09 09:32:14,215 INFO [train2.py:809] (0/4) Epoch 27, batch 2000, loss[ctc_loss=0.04379, att_loss=0.2062, loss=0.1737, over 15898.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.007934, over 39.00 utterances.], tot_loss[ctc_loss=0.06489, att_loss=0.2308, loss=0.1976, over 3263958.20 frames. utt_duration=1235 frames, utt_pad_proportion=0.05904, over 10586.94 utterances.], batch size: 39, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:33:15,246 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5446, 4.8779, 4.4818, 4.9309, 4.3726, 4.4901, 4.9802, 4.7852], device='cuda:0'), covar=tensor([0.0687, 0.0341, 0.0796, 0.0363, 0.0440, 0.0404, 0.0248, 0.0231], device='cuda:0'), in_proj_covar=tensor([0.0397, 0.0336, 0.0377, 0.0372, 0.0334, 0.0245, 0.0316, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 09:33:31,807 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9939, 4.9846, 4.9486, 2.1548, 2.0490, 2.9955, 2.6326, 3.8769], device='cuda:0'), covar=tensor([0.0772, 0.0349, 0.0271, 0.5458, 0.5525, 0.2340, 0.3335, 0.1695], device='cuda:0'), in_proj_covar=tensor([0.0359, 0.0299, 0.0277, 0.0251, 0.0338, 0.0330, 0.0262, 0.0369], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 09:33:34,514 INFO [train2.py:809] (0/4) Epoch 27, batch 2050, loss[ctc_loss=0.07468, att_loss=0.2476, loss=0.213, over 17324.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02323, over 59.00 utterances.], tot_loss[ctc_loss=0.06474, att_loss=0.2308, loss=0.1976, over 3268406.89 frames. utt_duration=1247 frames, utt_pad_proportion=0.05357, over 10500.08 utterances.], batch size: 59, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:34:11,423 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 1.673e+02 2.035e+02 2.438e+02 4.266e+02, threshold=4.071e+02, percent-clipped=1.0 2023-03-09 09:34:52,143 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9777, 5.2104, 5.1865, 5.2120, 5.2807, 5.2400, 4.8925, 4.6949], device='cuda:0'), covar=tensor([0.1069, 0.0594, 0.0323, 0.0463, 0.0293, 0.0320, 0.0427, 0.0350], device='cuda:0'), in_proj_covar=tensor([0.0539, 0.0384, 0.0372, 0.0378, 0.0443, 0.0451, 0.0381, 0.0414], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:34:53,526 INFO [train2.py:809] (0/4) Epoch 27, batch 2100, loss[ctc_loss=0.06006, att_loss=0.2433, loss=0.2066, over 17058.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008592, over 52.00 utterances.], tot_loss[ctc_loss=0.06569, att_loss=0.2315, loss=0.1983, over 3270561.23 frames. utt_duration=1238 frames, utt_pad_proportion=0.05533, over 10581.57 utterances.], batch size: 52, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:36:13,362 INFO [train2.py:809] (0/4) Epoch 27, batch 2150, loss[ctc_loss=0.05079, att_loss=0.2224, loss=0.1881, over 16136.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005618, over 42.00 utterances.], tot_loss[ctc_loss=0.06458, att_loss=0.2309, loss=0.1977, over 3267395.58 frames. utt_duration=1272 frames, utt_pad_proportion=0.04819, over 10289.10 utterances.], batch size: 42, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:36:37,188 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105744.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 09:36:51,248 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.182e+02 1.815e+02 2.188e+02 2.690e+02 5.528e+02, threshold=4.376e+02, percent-clipped=1.0 2023-03-09 09:37:32,705 INFO [train2.py:809] (0/4) Epoch 27, batch 2200, loss[ctc_loss=0.07974, att_loss=0.2516, loss=0.2172, over 17286.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01185, over 55.00 utterances.], tot_loss[ctc_loss=0.06415, att_loss=0.2302, loss=0.197, over 3270454.05 frames. utt_duration=1294 frames, utt_pad_proportion=0.04256, over 10117.65 utterances.], batch size: 55, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:37:53,015 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105792.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:38:04,325 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=105799.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:38:52,244 INFO [train2.py:809] (0/4) Epoch 27, batch 2250, loss[ctc_loss=0.06949, att_loss=0.2448, loss=0.2097, over 17425.00 frames. utt_duration=883.7 frames, utt_pad_proportion=0.07466, over 79.00 utterances.], tot_loss[ctc_loss=0.06425, att_loss=0.2309, loss=0.1976, over 3276242.66 frames. utt_duration=1284 frames, utt_pad_proportion=0.04338, over 10216.85 utterances.], batch size: 79, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:39:13,139 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105842.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:39:21,278 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8833, 3.6054, 3.5388, 3.0513, 3.5753, 3.6682, 3.6650, 2.7657], device='cuda:0'), covar=tensor([0.1060, 0.1484, 0.2304, 0.3508, 0.1780, 0.2076, 0.0895, 0.2837], device='cuda:0'), in_proj_covar=tensor([0.0200, 0.0206, 0.0221, 0.0272, 0.0182, 0.0282, 0.0204, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:39:30,859 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.345e+02 1.815e+02 2.086e+02 2.504e+02 6.374e+02, threshold=4.173e+02, percent-clipped=3.0 2023-03-09 09:39:31,283 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105853.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:39:39,170 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=105858.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:39:43,149 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=105860.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:39:58,566 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5175, 3.1191, 3.4870, 4.4754, 3.9397, 3.9122, 3.0305, 2.3539], device='cuda:0'), covar=tensor([0.0690, 0.1717, 0.0820, 0.0577, 0.0955, 0.0508, 0.1564, 0.2257], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0222, 0.0190, 0.0229, 0.0236, 0.0195, 0.0207, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:40:12,502 INFO [train2.py:809] (0/4) Epoch 27, batch 2300, loss[ctc_loss=0.05005, att_loss=0.211, loss=0.1788, over 14612.00 frames. utt_duration=1828 frames, utt_pad_proportion=0.03327, over 32.00 utterances.], tot_loss[ctc_loss=0.06455, att_loss=0.2307, loss=0.1975, over 3252462.31 frames. utt_duration=1269 frames, utt_pad_proportion=0.05273, over 10267.17 utterances.], batch size: 32, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:41:08,207 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3176, 5.5742, 5.1985, 5.6043, 5.0721, 5.1286, 5.6694, 5.4768], device='cuda:0'), covar=tensor([0.0525, 0.0289, 0.0691, 0.0306, 0.0374, 0.0244, 0.0200, 0.0206], device='cuda:0'), in_proj_covar=tensor([0.0400, 0.0338, 0.0381, 0.0374, 0.0336, 0.0247, 0.0318, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 09:41:31,985 INFO [train2.py:809] (0/4) Epoch 27, batch 2350, loss[ctc_loss=0.07053, att_loss=0.217, loss=0.1877, over 12741.00 frames. utt_duration=1822 frames, utt_pad_proportion=0.119, over 28.00 utterances.], tot_loss[ctc_loss=0.06507, att_loss=0.2307, loss=0.1976, over 3242666.57 frames. utt_duration=1291 frames, utt_pad_proportion=0.05019, over 10062.47 utterances.], batch size: 28, lr: 3.97e-03, grad_scale: 8.0 2023-03-09 09:42:00,596 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 09:42:09,624 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.259e+02 1.971e+02 2.209e+02 2.522e+02 5.529e+02, threshold=4.418e+02, percent-clipped=2.0 2023-03-09 09:42:51,494 INFO [train2.py:809] (0/4) Epoch 27, batch 2400, loss[ctc_loss=0.07811, att_loss=0.2475, loss=0.2136, over 17060.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.007872, over 52.00 utterances.], tot_loss[ctc_loss=0.06518, att_loss=0.2313, loss=0.1981, over 3251847.82 frames. utt_duration=1269 frames, utt_pad_proportion=0.05287, over 10265.11 utterances.], batch size: 52, lr: 3.96e-03, grad_scale: 8.0 2023-03-09 09:43:25,311 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-106000.pt 2023-03-09 09:44:16,148 INFO [train2.py:809] (0/4) Epoch 27, batch 2450, loss[ctc_loss=0.07214, att_loss=0.2478, loss=0.2126, over 17411.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03166, over 63.00 utterances.], tot_loss[ctc_loss=0.06588, att_loss=0.2319, loss=0.1987, over 3265453.14 frames. utt_duration=1266 frames, utt_pad_proportion=0.05021, over 10326.28 utterances.], batch size: 63, lr: 3.96e-03, grad_scale: 4.0 2023-03-09 09:44:40,207 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106044.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:44:55,919 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.307e+02 1.876e+02 2.325e+02 2.817e+02 5.594e+02, threshold=4.649e+02, percent-clipped=2.0 2023-03-09 09:45:09,317 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2296, 3.9121, 3.4225, 3.5562, 4.1769, 3.8416, 3.0306, 4.4233], device='cuda:0'), covar=tensor([0.0930, 0.0523, 0.0965, 0.0670, 0.0641, 0.0630, 0.0883, 0.0413], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0225, 0.0229, 0.0205, 0.0287, 0.0247, 0.0203, 0.0293], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 09:45:20,618 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9977, 5.2662, 5.5131, 5.2185, 5.5148, 5.9728, 5.2151, 6.0385], device='cuda:0'), covar=tensor([0.0627, 0.0721, 0.0823, 0.1337, 0.1547, 0.0791, 0.0737, 0.0605], device='cuda:0'), in_proj_covar=tensor([0.0910, 0.0523, 0.0639, 0.0682, 0.0907, 0.0660, 0.0507, 0.0640], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 09:45:33,180 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106077.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:45:35,770 INFO [train2.py:809] (0/4) Epoch 27, batch 2500, loss[ctc_loss=0.132, att_loss=0.2662, loss=0.2393, over 14829.00 frames. utt_duration=407.8 frames, utt_pad_proportion=0.2895, over 146.00 utterances.], tot_loss[ctc_loss=0.06598, att_loss=0.2324, loss=0.1991, over 3270925.50 frames. utt_duration=1265 frames, utt_pad_proportion=0.05029, over 10352.19 utterances.], batch size: 146, lr: 3.96e-03, grad_scale: 4.0 2023-03-09 09:45:57,172 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106092.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:46:03,548 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.6868, 5.9575, 5.3961, 5.6899, 5.6346, 5.0906, 5.3312, 5.1886], device='cuda:0'), covar=tensor([0.1323, 0.0876, 0.0894, 0.0856, 0.1050, 0.1612, 0.2477, 0.2320], device='cuda:0'), in_proj_covar=tensor([0.0558, 0.0633, 0.0483, 0.0476, 0.0451, 0.0479, 0.0643, 0.0543], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:46:56,807 INFO [train2.py:809] (0/4) Epoch 27, batch 2550, loss[ctc_loss=0.05748, att_loss=0.2028, loss=0.1737, over 15646.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008719, over 37.00 utterances.], tot_loss[ctc_loss=0.06574, att_loss=0.232, loss=0.1987, over 3267578.82 frames. utt_duration=1227 frames, utt_pad_proportion=0.06023, over 10669.13 utterances.], batch size: 37, lr: 3.96e-03, grad_scale: 4.0 2023-03-09 09:47:11,968 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106138.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:47:16,528 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7046, 2.4351, 2.4866, 2.6210, 2.7818, 2.8812, 2.5889, 3.1386], device='cuda:0'), covar=tensor([0.1706, 0.2559, 0.1876, 0.1493, 0.1722, 0.2039, 0.1952, 0.1394], device='cuda:0'), in_proj_covar=tensor([0.0143, 0.0145, 0.0141, 0.0135, 0.0153, 0.0130, 0.0152, 0.0129], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 09:47:17,985 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106142.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:47:27,603 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106148.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:47:36,809 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.077e+02 1.781e+02 2.163e+02 2.660e+02 6.949e+02, threshold=4.326e+02, percent-clipped=5.0 2023-03-09 09:47:39,196 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106155.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:47:39,994 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 09:47:44,015 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106158.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:48:16,575 INFO [train2.py:809] (0/4) Epoch 27, batch 2600, loss[ctc_loss=0.04779, att_loss=0.2093, loss=0.177, over 15771.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008667, over 38.00 utterances.], tot_loss[ctc_loss=0.06489, att_loss=0.2314, loss=0.1981, over 3273637.96 frames. utt_duration=1253 frames, utt_pad_proportion=0.05217, over 10464.47 utterances.], batch size: 38, lr: 3.96e-03, grad_scale: 4.0 2023-03-09 09:48:33,782 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106190.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:49:01,000 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106206.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:49:37,252 INFO [train2.py:809] (0/4) Epoch 27, batch 2650, loss[ctc_loss=0.05586, att_loss=0.2272, loss=0.1929, over 16283.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006599, over 43.00 utterances.], tot_loss[ctc_loss=0.0648, att_loss=0.2312, loss=0.1979, over 3273796.78 frames. utt_duration=1256 frames, utt_pad_proportion=0.05244, over 10441.75 utterances.], batch size: 43, lr: 3.96e-03, grad_scale: 4.0 2023-03-09 09:49:48,552 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1174, 4.4898, 4.4564, 4.6762, 2.8244, 4.3995, 2.7054, 1.8425], device='cuda:0'), covar=tensor([0.0509, 0.0348, 0.0662, 0.0261, 0.1522, 0.0257, 0.1446, 0.1663], device='cuda:0'), in_proj_covar=tensor([0.0215, 0.0183, 0.0266, 0.0178, 0.0222, 0.0164, 0.0231, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:50:17,426 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.073e+02 1.781e+02 2.178e+02 2.674e+02 4.972e+02, threshold=4.357e+02, percent-clipped=1.0 2023-03-09 09:50:34,220 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 09:50:39,581 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106268.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:50:55,947 INFO [train2.py:809] (0/4) Epoch 27, batch 2700, loss[ctc_loss=0.07045, att_loss=0.2413, loss=0.2072, over 17050.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008824, over 52.00 utterances.], tot_loss[ctc_loss=0.06585, att_loss=0.2318, loss=0.1986, over 3257732.55 frames. utt_duration=1208 frames, utt_pad_proportion=0.0691, over 10797.21 utterances.], batch size: 52, lr: 3.96e-03, grad_scale: 4.0 2023-03-09 09:51:05,553 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6000, 3.0861, 3.8200, 2.9364, 3.6943, 4.6751, 4.5184, 3.3505], device='cuda:0'), covar=tensor([0.0367, 0.1588, 0.1078, 0.1420, 0.0985, 0.1013, 0.0552, 0.1099], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0249, 0.0290, 0.0221, 0.0270, 0.0382, 0.0275, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 09:51:19,452 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.75 vs. limit=2.0 2023-03-09 09:52:15,253 INFO [train2.py:809] (0/4) Epoch 27, batch 2750, loss[ctc_loss=0.04429, att_loss=0.2003, loss=0.1691, over 14086.00 frames. utt_duration=1819 frames, utt_pad_proportion=0.03794, over 31.00 utterances.], tot_loss[ctc_loss=0.06627, att_loss=0.2314, loss=0.1983, over 3254521.41 frames. utt_duration=1196 frames, utt_pad_proportion=0.07183, over 10896.41 utterances.], batch size: 31, lr: 3.96e-03, grad_scale: 4.0 2023-03-09 09:52:15,612 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106329.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:52:27,148 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106336.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:52:31,504 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9077, 6.1806, 5.6771, 5.9049, 5.8426, 5.3702, 5.5836, 5.3706], device='cuda:0'), covar=tensor([0.1270, 0.0830, 0.1051, 0.0791, 0.1025, 0.1432, 0.2184, 0.2167], device='cuda:0'), in_proj_covar=tensor([0.0557, 0.0633, 0.0483, 0.0475, 0.0451, 0.0478, 0.0641, 0.0546], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 09:52:55,847 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.096e+02 1.858e+02 2.130e+02 2.638e+02 1.636e+03, threshold=4.261e+02, percent-clipped=5.0 2023-03-09 09:53:27,363 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8136, 3.0377, 3.5535, 4.6077, 4.1870, 4.0214, 3.1613, 2.5860], device='cuda:0'), covar=tensor([0.0563, 0.1877, 0.0870, 0.0509, 0.0902, 0.0472, 0.1332, 0.1969], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0217, 0.0187, 0.0225, 0.0232, 0.0191, 0.0203, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:53:34,594 INFO [train2.py:809] (0/4) Epoch 27, batch 2800, loss[ctc_loss=0.06384, att_loss=0.2414, loss=0.2059, over 17113.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01519, over 56.00 utterances.], tot_loss[ctc_loss=0.06549, att_loss=0.2305, loss=0.1975, over 3258993.07 frames. utt_duration=1238 frames, utt_pad_proportion=0.0609, over 10544.72 utterances.], batch size: 56, lr: 3.96e-03, grad_scale: 8.0 2023-03-09 09:54:04,681 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106397.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:54:09,670 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3144, 2.3717, 4.8620, 3.8375, 3.0401, 4.1514, 4.4915, 4.4866], device='cuda:0'), covar=tensor([0.0316, 0.1682, 0.0191, 0.0875, 0.1668, 0.0275, 0.0279, 0.0289], device='cuda:0'), in_proj_covar=tensor([0.0230, 0.0249, 0.0223, 0.0327, 0.0273, 0.0240, 0.0214, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 09:54:55,104 INFO [train2.py:809] (0/4) Epoch 27, batch 2850, loss[ctc_loss=0.07558, att_loss=0.2377, loss=0.2053, over 16256.00 frames. utt_duration=1513 frames, utt_pad_proportion=0.006897, over 43.00 utterances.], tot_loss[ctc_loss=0.06516, att_loss=0.2307, loss=0.1976, over 3265592.08 frames. utt_duration=1251 frames, utt_pad_proportion=0.05656, over 10457.05 utterances.], batch size: 43, lr: 3.96e-03, grad_scale: 8.0 2023-03-09 09:55:01,772 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106433.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:55:26,844 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106448.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:55:30,650 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.25 vs. limit=2.0 2023-03-09 09:55:36,025 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.102e+02 1.809e+02 2.032e+02 2.602e+02 5.247e+02, threshold=4.065e+02, percent-clipped=2.0 2023-03-09 09:55:36,374 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=106454.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:55:37,978 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106455.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:56:15,397 INFO [train2.py:809] (0/4) Epoch 27, batch 2900, loss[ctc_loss=0.05884, att_loss=0.2321, loss=0.1975, over 16413.00 frames. utt_duration=1494 frames, utt_pad_proportion=0.006966, over 44.00 utterances.], tot_loss[ctc_loss=0.06534, att_loss=0.2312, loss=0.198, over 3266762.42 frames. utt_duration=1233 frames, utt_pad_proportion=0.06022, over 10612.63 utterances.], batch size: 44, lr: 3.96e-03, grad_scale: 8.0 2023-03-09 09:56:42,874 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106496.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:56:53,784 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106503.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:57:13,214 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=106515.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 09:57:34,531 INFO [train2.py:809] (0/4) Epoch 27, batch 2950, loss[ctc_loss=0.07786, att_loss=0.2447, loss=0.2113, over 17127.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01387, over 56.00 utterances.], tot_loss[ctc_loss=0.06521, att_loss=0.2313, loss=0.1981, over 3264887.41 frames. utt_duration=1252 frames, utt_pad_proportion=0.05615, over 10445.79 utterances.], batch size: 56, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 09:58:15,059 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.306e+02 2.110e+02 2.427e+02 2.893e+02 4.636e+02, threshold=4.854e+02, percent-clipped=4.0 2023-03-09 09:58:53,985 INFO [train2.py:809] (0/4) Epoch 27, batch 3000, loss[ctc_loss=0.08131, att_loss=0.2476, loss=0.2143, over 17283.00 frames. utt_duration=1173 frames, utt_pad_proportion=0.02223, over 59.00 utterances.], tot_loss[ctc_loss=0.06521, att_loss=0.2311, loss=0.1979, over 3267646.10 frames. utt_duration=1254 frames, utt_pad_proportion=0.05424, over 10438.39 utterances.], batch size: 59, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 09:58:53,987 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 09:59:07,588 INFO [train2.py:843] (0/4) Epoch 27, validation: ctc_loss=0.04056, att_loss=0.2346, loss=0.1958, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 09:59:07,589 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 09:59:47,312 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0420, 5.1285, 4.9330, 2.3358, 2.0702, 3.1945, 2.4951, 3.9924], device='cuda:0'), covar=tensor([0.0747, 0.0355, 0.0327, 0.5265, 0.5576, 0.2109, 0.3871, 0.1574], device='cuda:0'), in_proj_covar=tensor([0.0364, 0.0300, 0.0280, 0.0253, 0.0341, 0.0332, 0.0263, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 10:00:20,299 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106624.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:00:28,229 INFO [train2.py:809] (0/4) Epoch 27, batch 3050, loss[ctc_loss=0.0652, att_loss=0.2366, loss=0.2023, over 16684.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006543, over 46.00 utterances.], tot_loss[ctc_loss=0.06542, att_loss=0.2311, loss=0.1979, over 3266445.42 frames. utt_duration=1261 frames, utt_pad_proportion=0.05186, over 10373.76 utterances.], batch size: 46, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:00:57,696 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0911, 4.3045, 4.4021, 4.4729, 2.7041, 4.3483, 2.7163, 1.7172], device='cuda:0'), covar=tensor([0.0456, 0.0301, 0.0554, 0.0253, 0.1533, 0.0246, 0.1361, 0.1696], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0186, 0.0267, 0.0180, 0.0225, 0.0167, 0.0234, 0.0207], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:01:08,160 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.208e+02 1.917e+02 2.259e+02 2.641e+02 5.465e+02, threshold=4.518e+02, percent-clipped=3.0 2023-03-09 10:01:46,960 INFO [train2.py:809] (0/4) Epoch 27, batch 3100, loss[ctc_loss=0.04808, att_loss=0.2419, loss=0.2031, over 17032.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007147, over 51.00 utterances.], tot_loss[ctc_loss=0.06531, att_loss=0.2308, loss=0.1977, over 3268234.82 frames. utt_duration=1288 frames, utt_pad_proportion=0.04529, over 10163.47 utterances.], batch size: 51, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:02:08,306 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106692.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:03:06,632 INFO [train2.py:809] (0/4) Epoch 27, batch 3150, loss[ctc_loss=0.04982, att_loss=0.2109, loss=0.1787, over 15957.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006346, over 41.00 utterances.], tot_loss[ctc_loss=0.06593, att_loss=0.231, loss=0.198, over 3270445.93 frames. utt_duration=1288 frames, utt_pad_proportion=0.04498, over 10169.09 utterances.], batch size: 41, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:03:13,643 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106733.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:03:38,394 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4309, 4.6189, 4.6547, 4.7011, 5.1952, 4.5130, 4.6252, 2.8488], device='cuda:0'), covar=tensor([0.0320, 0.0386, 0.0340, 0.0340, 0.0707, 0.0281, 0.0358, 0.1631], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0220, 0.0216, 0.0232, 0.0382, 0.0191, 0.0207, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:03:47,041 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-03-09 10:03:47,285 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.396e+02 1.911e+02 2.328e+02 2.721e+02 4.436e+02, threshold=4.656e+02, percent-clipped=0.0 2023-03-09 10:04:01,390 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-03-09 10:04:12,095 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.73 vs. limit=5.0 2023-03-09 10:04:26,570 INFO [train2.py:809] (0/4) Epoch 27, batch 3200, loss[ctc_loss=0.05832, att_loss=0.211, loss=0.1805, over 15761.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009217, over 38.00 utterances.], tot_loss[ctc_loss=0.0652, att_loss=0.2307, loss=0.1976, over 3272797.64 frames. utt_duration=1302 frames, utt_pad_proportion=0.04066, over 10066.73 utterances.], batch size: 38, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:04:30,348 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106781.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:05:17,404 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=106810.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:05:47,161 INFO [train2.py:809] (0/4) Epoch 27, batch 3250, loss[ctc_loss=0.04775, att_loss=0.2044, loss=0.1731, over 14115.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.04658, over 31.00 utterances.], tot_loss[ctc_loss=0.06542, att_loss=0.2312, loss=0.1981, over 3276299.08 frames. utt_duration=1294 frames, utt_pad_proportion=0.04307, over 10139.14 utterances.], batch size: 31, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:06:27,324 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.302e+02 1.765e+02 2.095e+02 2.388e+02 3.844e+02, threshold=4.190e+02, percent-clipped=0.0 2023-03-09 10:07:07,560 INFO [train2.py:809] (0/4) Epoch 27, batch 3300, loss[ctc_loss=0.06073, att_loss=0.2215, loss=0.1893, over 16288.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006823, over 43.00 utterances.], tot_loss[ctc_loss=0.06454, att_loss=0.2306, loss=0.1974, over 3272105.10 frames. utt_duration=1298 frames, utt_pad_proportion=0.04394, over 10094.31 utterances.], batch size: 43, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:08:19,378 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106924.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:08:27,389 INFO [train2.py:809] (0/4) Epoch 27, batch 3350, loss[ctc_loss=0.06948, att_loss=0.2427, loss=0.208, over 17296.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01232, over 55.00 utterances.], tot_loss[ctc_loss=0.06443, att_loss=0.2307, loss=0.1974, over 3275355.88 frames. utt_duration=1312 frames, utt_pad_proportion=0.03911, over 9998.44 utterances.], batch size: 55, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:08:53,724 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9901, 4.3110, 4.4691, 4.4674, 2.8931, 4.2478, 2.8887, 2.1651], device='cuda:0'), covar=tensor([0.0536, 0.0423, 0.0710, 0.0326, 0.1616, 0.0317, 0.1430, 0.1650], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0186, 0.0268, 0.0181, 0.0224, 0.0167, 0.0234, 0.0207], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:09:07,450 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.083e+02 1.952e+02 2.382e+02 3.093e+02 6.808e+02, threshold=4.763e+02, percent-clipped=6.0 2023-03-09 10:09:11,544 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 10:09:34,299 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6489, 3.1885, 3.1782, 2.8791, 3.1610, 3.1932, 3.2812, 2.4257], device='cuda:0'), covar=tensor([0.1290, 0.1336, 0.2218, 0.3028, 0.1152, 0.1643, 0.0887, 0.3285], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0207, 0.0223, 0.0272, 0.0183, 0.0282, 0.0206, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:09:35,671 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=106972.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:09:44,499 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6532, 3.1878, 3.7025, 4.4759, 3.8422, 3.8252, 2.9490, 2.5646], device='cuda:0'), covar=tensor([0.0623, 0.1606, 0.0709, 0.0573, 0.0969, 0.0570, 0.1589, 0.1969], device='cuda:0'), in_proj_covar=tensor([0.0188, 0.0216, 0.0187, 0.0224, 0.0232, 0.0191, 0.0204, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:09:47,289 INFO [train2.py:809] (0/4) Epoch 27, batch 3400, loss[ctc_loss=0.0804, att_loss=0.2481, loss=0.2146, over 16893.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.006783, over 49.00 utterances.], tot_loss[ctc_loss=0.06524, att_loss=0.2318, loss=0.1985, over 3277542.00 frames. utt_duration=1271 frames, utt_pad_proportion=0.04795, over 10328.66 utterances.], batch size: 49, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:10:01,394 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3628, 2.9774, 3.7097, 3.0805, 3.4877, 4.5150, 4.3853, 3.3002], device='cuda:0'), covar=tensor([0.0407, 0.1699, 0.1211, 0.1193, 0.1145, 0.0993, 0.0599, 0.1150], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0253, 0.0296, 0.0224, 0.0274, 0.0387, 0.0278, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:10:09,114 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=106992.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:10:23,695 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8250, 5.0990, 5.3787, 5.1441, 5.3599, 5.7873, 5.1363, 5.8832], device='cuda:0'), covar=tensor([0.0723, 0.0787, 0.0816, 0.1432, 0.1577, 0.0879, 0.0867, 0.0631], device='cuda:0'), in_proj_covar=tensor([0.0914, 0.0523, 0.0644, 0.0680, 0.0908, 0.0666, 0.0510, 0.0643], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:11:07,806 INFO [train2.py:809] (0/4) Epoch 27, batch 3450, loss[ctc_loss=0.06062, att_loss=0.2044, loss=0.1756, over 14476.00 frames. utt_duration=1812 frames, utt_pad_proportion=0.0405, over 32.00 utterances.], tot_loss[ctc_loss=0.0654, att_loss=0.2313, loss=0.1981, over 3264134.97 frames. utt_duration=1257 frames, utt_pad_proportion=0.05532, over 10397.78 utterances.], batch size: 32, lr: 3.95e-03, grad_scale: 8.0 2023-03-09 10:11:25,821 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107040.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:11:48,076 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.134e+02 1.835e+02 2.250e+02 2.844e+02 5.706e+02, threshold=4.501e+02, percent-clipped=2.0 2023-03-09 10:12:26,814 INFO [train2.py:809] (0/4) Epoch 27, batch 3500, loss[ctc_loss=0.05849, att_loss=0.2319, loss=0.1972, over 17344.00 frames. utt_duration=1177 frames, utt_pad_proportion=0.02214, over 59.00 utterances.], tot_loss[ctc_loss=0.06562, att_loss=0.2313, loss=0.1982, over 3262168.53 frames. utt_duration=1250 frames, utt_pad_proportion=0.05828, over 10454.26 utterances.], batch size: 59, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:12:37,298 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8184, 3.6884, 3.5718, 3.1446, 3.6108, 3.6290, 3.6967, 2.6676], device='cuda:0'), covar=tensor([0.1019, 0.0877, 0.1756, 0.2713, 0.1050, 0.2243, 0.0785, 0.2762], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0207, 0.0223, 0.0272, 0.0183, 0.0284, 0.0206, 0.0230], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:13:15,825 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107110.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:13:45,758 INFO [train2.py:809] (0/4) Epoch 27, batch 3550, loss[ctc_loss=0.08623, att_loss=0.242, loss=0.2109, over 16862.00 frames. utt_duration=1378 frames, utt_pad_proportion=0.007802, over 49.00 utterances.], tot_loss[ctc_loss=0.06557, att_loss=0.231, loss=0.1979, over 3260906.78 frames. utt_duration=1250 frames, utt_pad_proportion=0.05695, over 10449.41 utterances.], batch size: 49, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:13:51,013 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1825, 5.3882, 5.6883, 5.4553, 5.6924, 6.1053, 5.3575, 6.2119], device='cuda:0'), covar=tensor([0.0656, 0.0698, 0.0800, 0.1372, 0.1693, 0.0929, 0.0678, 0.0640], device='cuda:0'), in_proj_covar=tensor([0.0914, 0.0526, 0.0644, 0.0683, 0.0910, 0.0667, 0.0511, 0.0644], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:14:03,580 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107140.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 10:14:22,413 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107152.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:14:24,971 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.809e+01 1.862e+02 2.158e+02 2.587e+02 5.069e+02, threshold=4.316e+02, percent-clipped=3.0 2023-03-09 10:14:31,568 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107158.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:15:05,779 INFO [train2.py:809] (0/4) Epoch 27, batch 3600, loss[ctc_loss=0.06485, att_loss=0.2398, loss=0.2048, over 16454.00 frames. utt_duration=1432 frames, utt_pad_proportion=0.007472, over 46.00 utterances.], tot_loss[ctc_loss=0.06524, att_loss=0.2314, loss=0.1982, over 3266168.68 frames. utt_duration=1245 frames, utt_pad_proportion=0.05597, over 10508.00 utterances.], batch size: 46, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:15:41,160 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107201.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 10:15:59,486 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107213.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 10:16:24,610 INFO [train2.py:809] (0/4) Epoch 27, batch 3650, loss[ctc_loss=0.07444, att_loss=0.249, loss=0.2141, over 17059.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.007872, over 52.00 utterances.], tot_loss[ctc_loss=0.06552, att_loss=0.2315, loss=0.1983, over 3266380.14 frames. utt_duration=1251 frames, utt_pad_proportion=0.05528, over 10456.11 utterances.], batch size: 52, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:17:03,101 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 1.892e+02 2.154e+02 2.728e+02 3.735e+02, threshold=4.308e+02, percent-clipped=0.0 2023-03-09 10:17:32,390 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1306, 4.5121, 4.5570, 4.6400, 2.9169, 4.4168, 2.9670, 2.3330], device='cuda:0'), covar=tensor([0.0485, 0.0295, 0.0575, 0.0243, 0.1450, 0.0284, 0.1238, 0.1366], device='cuda:0'), in_proj_covar=tensor([0.0218, 0.0186, 0.0268, 0.0181, 0.0224, 0.0167, 0.0233, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:17:43,432 INFO [train2.py:809] (0/4) Epoch 27, batch 3700, loss[ctc_loss=0.07434, att_loss=0.2508, loss=0.2155, over 16761.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.00691, over 48.00 utterances.], tot_loss[ctc_loss=0.0656, att_loss=0.2321, loss=0.1988, over 3264129.32 frames. utt_duration=1228 frames, utt_pad_proportion=0.061, over 10644.02 utterances.], batch size: 48, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:19:01,339 INFO [train2.py:809] (0/4) Epoch 27, batch 3750, loss[ctc_loss=0.07229, att_loss=0.2221, loss=0.1922, over 15897.00 frames. utt_duration=1632 frames, utt_pad_proportion=0.008677, over 39.00 utterances.], tot_loss[ctc_loss=0.06579, att_loss=0.2318, loss=0.1986, over 3253242.95 frames. utt_duration=1227 frames, utt_pad_proportion=0.06323, over 10620.16 utterances.], batch size: 39, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:19:40,407 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.252e+02 1.917e+02 2.215e+02 2.702e+02 1.402e+03, threshold=4.430e+02, percent-clipped=2.0 2023-03-09 10:20:19,854 INFO [train2.py:809] (0/4) Epoch 27, batch 3800, loss[ctc_loss=0.0734, att_loss=0.2485, loss=0.2135, over 17286.00 frames. utt_duration=876.8 frames, utt_pad_proportion=0.08094, over 79.00 utterances.], tot_loss[ctc_loss=0.06594, att_loss=0.2323, loss=0.1991, over 3267061.55 frames. utt_duration=1245 frames, utt_pad_proportion=0.05516, over 10513.37 utterances.], batch size: 79, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:21:20,588 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3445, 3.9192, 3.3540, 3.4975, 4.1553, 3.7966, 3.0971, 4.4364], device='cuda:0'), covar=tensor([0.0872, 0.0443, 0.0974, 0.0789, 0.0643, 0.0657, 0.0849, 0.0463], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0225, 0.0229, 0.0206, 0.0288, 0.0246, 0.0202, 0.0294], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 10:21:25,855 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8932, 5.1569, 5.3730, 5.1244, 5.4436, 5.8387, 5.1702, 5.9589], device='cuda:0'), covar=tensor([0.0729, 0.0735, 0.0959, 0.1594, 0.1673, 0.0967, 0.0765, 0.0686], device='cuda:0'), in_proj_covar=tensor([0.0920, 0.0527, 0.0646, 0.0685, 0.0915, 0.0670, 0.0512, 0.0643], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:21:39,084 INFO [train2.py:809] (0/4) Epoch 27, batch 3850, loss[ctc_loss=0.06662, att_loss=0.235, loss=0.2013, over 17011.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.009135, over 51.00 utterances.], tot_loss[ctc_loss=0.06581, att_loss=0.2316, loss=0.1984, over 3270955.93 frames. utt_duration=1260 frames, utt_pad_proportion=0.05127, over 10392.61 utterances.], batch size: 51, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:22:17,271 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.331e+02 1.947e+02 2.251e+02 2.768e+02 8.015e+02, threshold=4.503e+02, percent-clipped=5.0 2023-03-09 10:22:32,624 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0140, 5.3873, 4.9949, 5.3950, 4.8107, 5.0016, 5.5217, 5.2864], device='cuda:0'), covar=tensor([0.0644, 0.0306, 0.0688, 0.0401, 0.0397, 0.0260, 0.0199, 0.0191], device='cuda:0'), in_proj_covar=tensor([0.0401, 0.0343, 0.0379, 0.0383, 0.0339, 0.0248, 0.0321, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 10:22:55,305 INFO [train2.py:809] (0/4) Epoch 27, batch 3900, loss[ctc_loss=0.07034, att_loss=0.2486, loss=0.213, over 17119.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01511, over 56.00 utterances.], tot_loss[ctc_loss=0.06562, att_loss=0.2316, loss=0.1984, over 3271152.30 frames. utt_duration=1236 frames, utt_pad_proportion=0.05697, over 10599.69 utterances.], batch size: 56, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:22:58,824 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1986, 5.1558, 4.9775, 3.2842, 4.9596, 4.8793, 4.5697, 3.1653], device='cuda:0'), covar=tensor([0.0103, 0.0103, 0.0239, 0.0855, 0.0102, 0.0173, 0.0273, 0.1067], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0108, 0.0112, 0.0114, 0.0091, 0.0120, 0.0103, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 10:23:00,388 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5876, 4.9872, 4.7608, 4.8310, 5.0832, 4.7220, 3.5981, 4.9329], device='cuda:0'), covar=tensor([0.0146, 0.0108, 0.0170, 0.0102, 0.0089, 0.0130, 0.0706, 0.0190], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0092, 0.0116, 0.0073, 0.0078, 0.0089, 0.0105, 0.0108], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:23:22,174 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107496.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 10:23:29,972 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0539, 4.4236, 4.4690, 4.5525, 2.9049, 4.3965, 3.1236, 2.1056], device='cuda:0'), covar=tensor([0.0511, 0.0294, 0.0612, 0.0265, 0.1436, 0.0224, 0.1145, 0.1456], device='cuda:0'), in_proj_covar=tensor([0.0217, 0.0186, 0.0267, 0.0180, 0.0222, 0.0166, 0.0233, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:23:40,489 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107508.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 10:23:43,583 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107510.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:23:45,155 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7086, 2.4853, 2.3235, 2.5418, 2.8561, 2.6940, 2.3974, 3.0291], device='cuda:0'), covar=tensor([0.1472, 0.2125, 0.2016, 0.1270, 0.1574, 0.1182, 0.1973, 0.1133], device='cuda:0'), in_proj_covar=tensor([0.0141, 0.0142, 0.0139, 0.0133, 0.0150, 0.0129, 0.0149, 0.0127], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 10:24:12,007 INFO [train2.py:809] (0/4) Epoch 27, batch 3950, loss[ctc_loss=0.0687, att_loss=0.2232, loss=0.1923, over 16181.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006412, over 41.00 utterances.], tot_loss[ctc_loss=0.066, att_loss=0.232, loss=0.1988, over 3269489.67 frames. utt_duration=1234 frames, utt_pad_proportion=0.05752, over 10609.27 utterances.], batch size: 41, lr: 3.94e-03, grad_scale: 8.0 2023-03-09 10:24:49,844 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.368e+02 1.912e+02 2.157e+02 2.759e+02 5.743e+02, threshold=4.314e+02, percent-clipped=3.0 2023-03-09 10:25:01,834 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-27.pt 2023-03-09 10:25:20,651 INFO [train2.py:809] (0/4) Epoch 28, batch 0, loss[ctc_loss=0.05692, att_loss=0.2175, loss=0.1854, over 16122.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006022, over 42.00 utterances.], tot_loss[ctc_loss=0.05692, att_loss=0.2175, loss=0.1854, over 16122.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006022, over 42.00 utterances.], batch size: 42, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:25:20,654 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 10:25:32,829 INFO [train2.py:843] (0/4) Epoch 28, validation: ctc_loss=0.04041, att_loss=0.2344, loss=0.1956, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 10:25:32,830 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 10:25:46,736 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107571.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:26:52,868 INFO [train2.py:809] (0/4) Epoch 28, batch 50, loss[ctc_loss=0.07502, att_loss=0.2405, loss=0.2074, over 17308.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01073, over 55.00 utterances.], tot_loss[ctc_loss=0.06734, att_loss=0.2339, loss=0.2006, over 739165.82 frames. utt_duration=1107 frames, utt_pad_proportion=0.08977, over 2675.44 utterances.], batch size: 55, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:28:00,263 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.063e+02 1.779e+02 2.036e+02 2.554e+02 4.414e+02, threshold=4.071e+02, percent-clipped=2.0 2023-03-09 10:28:08,299 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107659.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:28:12,547 INFO [train2.py:809] (0/4) Epoch 28, batch 100, loss[ctc_loss=0.08243, att_loss=0.2484, loss=0.2152, over 16770.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006496, over 48.00 utterances.], tot_loss[ctc_loss=0.06676, att_loss=0.2339, loss=0.2004, over 1307828.16 frames. utt_duration=1156 frames, utt_pad_proportion=0.07071, over 4531.45 utterances.], batch size: 48, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:28:56,265 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2224, 3.8046, 3.3565, 3.4610, 4.0408, 3.6520, 3.3054, 4.2585], device='cuda:0'), covar=tensor([0.0887, 0.0402, 0.0936, 0.0705, 0.0701, 0.0701, 0.0712, 0.0460], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0226, 0.0231, 0.0208, 0.0290, 0.0249, 0.0204, 0.0296], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 10:29:32,143 INFO [train2.py:809] (0/4) Epoch 28, batch 150, loss[ctc_loss=0.05177, att_loss=0.1921, loss=0.164, over 15361.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01176, over 35.00 utterances.], tot_loss[ctc_loss=0.06652, att_loss=0.2329, loss=0.1996, over 1737711.82 frames. utt_duration=1211 frames, utt_pad_proportion=0.06371, over 5749.18 utterances.], batch size: 35, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:29:45,365 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107720.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:30:11,765 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=107737.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:30:38,888 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 1.937e+02 2.331e+02 2.865e+02 6.453e+02, threshold=4.662e+02, percent-clipped=4.0 2023-03-09 10:30:51,738 INFO [train2.py:809] (0/4) Epoch 28, batch 200, loss[ctc_loss=0.06837, att_loss=0.2541, loss=0.217, over 17050.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.00825, over 52.00 utterances.], tot_loss[ctc_loss=0.06475, att_loss=0.2314, loss=0.1981, over 2077697.49 frames. utt_duration=1248 frames, utt_pad_proportion=0.05354, over 6667.00 utterances.], batch size: 52, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:31:45,230 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107796.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 10:31:48,992 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=107798.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 10:32:05,087 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=107808.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:32:10,868 INFO [train2.py:809] (0/4) Epoch 28, batch 250, loss[ctc_loss=0.05566, att_loss=0.2219, loss=0.1887, over 15944.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007202, over 41.00 utterances.], tot_loss[ctc_loss=0.0658, att_loss=0.2321, loss=0.1988, over 2337784.59 frames. utt_duration=1230 frames, utt_pad_proportion=0.06007, over 7613.58 utterances.], batch size: 41, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:32:53,402 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5252, 2.6984, 4.9871, 4.0170, 3.1705, 4.3684, 4.9051, 4.7004], device='cuda:0'), covar=tensor([0.0321, 0.1466, 0.0282, 0.0844, 0.1548, 0.0248, 0.0195, 0.0308], device='cuda:0'), in_proj_covar=tensor([0.0228, 0.0246, 0.0223, 0.0322, 0.0268, 0.0238, 0.0212, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:33:01,095 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107844.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 10:33:16,917 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.125e+02 1.778e+02 2.118e+02 2.564e+02 4.844e+02, threshold=4.236e+02, percent-clipped=1.0 2023-03-09 10:33:20,620 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=107856.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:33:29,916 INFO [train2.py:809] (0/4) Epoch 28, batch 300, loss[ctc_loss=0.05189, att_loss=0.2154, loss=0.1827, over 16139.00 frames. utt_duration=1539 frames, utt_pad_proportion=0.005402, over 42.00 utterances.], tot_loss[ctc_loss=0.06541, att_loss=0.231, loss=0.1978, over 2538950.85 frames. utt_duration=1262 frames, utt_pad_proportion=0.0545, over 8059.81 utterances.], batch size: 42, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:33:36,221 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=107866.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:33:47,639 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8364, 4.8659, 4.9234, 4.9408, 5.3846, 4.8394, 4.7866, 2.3907], device='cuda:0'), covar=tensor([0.0188, 0.0215, 0.0222, 0.0224, 0.0429, 0.0174, 0.0246, 0.1806], device='cuda:0'), in_proj_covar=tensor([0.0192, 0.0220, 0.0215, 0.0231, 0.0379, 0.0191, 0.0206, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:34:49,116 INFO [train2.py:809] (0/4) Epoch 28, batch 350, loss[ctc_loss=0.06851, att_loss=0.2319, loss=0.1992, over 16973.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007115, over 50.00 utterances.], tot_loss[ctc_loss=0.06561, att_loss=0.2312, loss=0.1981, over 2705313.13 frames. utt_duration=1264 frames, utt_pad_proportion=0.05138, over 8570.60 utterances.], batch size: 50, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:35:55,813 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.165e+02 1.778e+02 2.080e+02 2.502e+02 5.851e+02, threshold=4.160e+02, percent-clipped=2.0 2023-03-09 10:36:08,360 INFO [train2.py:809] (0/4) Epoch 28, batch 400, loss[ctc_loss=0.07112, att_loss=0.2322, loss=0.2, over 16889.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006318, over 49.00 utterances.], tot_loss[ctc_loss=0.06577, att_loss=0.2318, loss=0.1986, over 2833972.92 frames. utt_duration=1261 frames, utt_pad_proportion=0.05132, over 8998.36 utterances.], batch size: 49, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:36:30,759 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3389, 5.2508, 5.0715, 3.1563, 5.0972, 4.9523, 4.6616, 3.1644], device='cuda:0'), covar=tensor([0.0102, 0.0105, 0.0269, 0.0891, 0.0099, 0.0170, 0.0249, 0.1110], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0107, 0.0112, 0.0114, 0.0091, 0.0120, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 10:37:08,488 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-108000.pt 2023-03-09 10:37:33,053 INFO [train2.py:809] (0/4) Epoch 28, batch 450, loss[ctc_loss=0.07924, att_loss=0.2518, loss=0.2173, over 17275.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01346, over 55.00 utterances.], tot_loss[ctc_loss=0.06555, att_loss=0.2318, loss=0.1985, over 2933819.11 frames. utt_duration=1277 frames, utt_pad_proportion=0.04742, over 9202.14 utterances.], batch size: 55, lr: 3.86e-03, grad_scale: 8.0 2023-03-09 10:37:37,574 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108015.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:38:20,528 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2504, 5.2298, 4.7376, 2.7107, 4.9773, 5.0582, 4.3409, 2.6142], device='cuda:0'), covar=tensor([0.0135, 0.0132, 0.0411, 0.1437, 0.0145, 0.0172, 0.0451, 0.2067], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0106, 0.0111, 0.0113, 0.0090, 0.0119, 0.0101, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 10:38:39,470 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.152e+02 1.861e+02 2.240e+02 2.601e+02 4.871e+02, threshold=4.480e+02, percent-clipped=1.0 2023-03-09 10:38:51,887 INFO [train2.py:809] (0/4) Epoch 28, batch 500, loss[ctc_loss=0.08484, att_loss=0.2564, loss=0.2221, over 14856.00 frames. utt_duration=408.6 frames, utt_pad_proportion=0.2894, over 146.00 utterances.], tot_loss[ctc_loss=0.0655, att_loss=0.2315, loss=0.1983, over 2995948.38 frames. utt_duration=1228 frames, utt_pad_proportion=0.06103, over 9770.84 utterances.], batch size: 146, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:38:55,262 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7076, 3.1078, 3.7341, 3.3812, 3.6684, 4.6842, 4.5337, 3.4472], device='cuda:0'), covar=tensor([0.0291, 0.1628, 0.1175, 0.1105, 0.0975, 0.0907, 0.0561, 0.1155], device='cuda:0'), in_proj_covar=tensor([0.0248, 0.0249, 0.0289, 0.0218, 0.0269, 0.0379, 0.0272, 0.0236], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:39:05,315 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8053, 3.5779, 3.5956, 3.1202, 3.6068, 3.5846, 3.6095, 2.7273], device='cuda:0'), covar=tensor([0.0959, 0.0954, 0.1566, 0.2860, 0.0793, 0.1517, 0.0806, 0.3058], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0206, 0.0222, 0.0270, 0.0182, 0.0280, 0.0206, 0.0227], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:39:41,163 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108093.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 10:39:58,046 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9408, 4.9129, 4.8749, 2.2911, 1.9907, 2.8842, 2.2359, 3.7484], device='cuda:0'), covar=tensor([0.0743, 0.0289, 0.0260, 0.4477, 0.5270, 0.2380, 0.3941, 0.1681], device='cuda:0'), in_proj_covar=tensor([0.0365, 0.0302, 0.0282, 0.0254, 0.0341, 0.0335, 0.0265, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 10:40:05,567 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7307, 3.5678, 3.5954, 3.0980, 3.5795, 3.5640, 3.5859, 2.6661], device='cuda:0'), covar=tensor([0.1025, 0.1207, 0.1546, 0.2502, 0.1044, 0.1905, 0.0785, 0.2888], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0206, 0.0222, 0.0270, 0.0182, 0.0280, 0.0206, 0.0227], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:40:11,190 INFO [train2.py:809] (0/4) Epoch 28, batch 550, loss[ctc_loss=0.0716, att_loss=0.2416, loss=0.2076, over 16341.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005487, over 45.00 utterances.], tot_loss[ctc_loss=0.06509, att_loss=0.2315, loss=0.1982, over 3058531.75 frames. utt_duration=1219 frames, utt_pad_proportion=0.06375, over 10048.97 utterances.], batch size: 45, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:40:43,338 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1135, 6.3394, 5.8030, 6.0069, 6.0708, 5.4459, 5.7800, 5.4478], device='cuda:0'), covar=tensor([0.1145, 0.0756, 0.0874, 0.0815, 0.0760, 0.1532, 0.1951, 0.2547], device='cuda:0'), in_proj_covar=tensor([0.0563, 0.0643, 0.0490, 0.0478, 0.0457, 0.0485, 0.0645, 0.0550], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 10:41:18,472 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.142e+02 1.757e+02 2.157e+02 2.611e+02 5.249e+02, threshold=4.315e+02, percent-clipped=2.0 2023-03-09 10:41:30,577 INFO [train2.py:809] (0/4) Epoch 28, batch 600, loss[ctc_loss=0.05401, att_loss=0.2187, loss=0.1858, over 16267.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.0073, over 43.00 utterances.], tot_loss[ctc_loss=0.06504, att_loss=0.231, loss=0.1978, over 3104189.21 frames. utt_duration=1234 frames, utt_pad_proportion=0.05983, over 10075.13 utterances.], batch size: 43, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:41:37,527 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108166.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:42:49,655 INFO [train2.py:809] (0/4) Epoch 28, batch 650, loss[ctc_loss=0.05773, att_loss=0.2397, loss=0.2033, over 16974.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007159, over 50.00 utterances.], tot_loss[ctc_loss=0.06501, att_loss=0.2309, loss=0.1977, over 3136588.62 frames. utt_duration=1208 frames, utt_pad_proportion=0.06713, over 10398.59 utterances.], batch size: 50, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:42:52,650 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108214.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:43:56,527 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.058e+02 1.814e+02 2.237e+02 2.665e+02 5.467e+02, threshold=4.474e+02, percent-clipped=5.0 2023-03-09 10:44:08,563 INFO [train2.py:809] (0/4) Epoch 28, batch 700, loss[ctc_loss=0.05433, att_loss=0.2308, loss=0.1955, over 16626.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.00517, over 47.00 utterances.], tot_loss[ctc_loss=0.06524, att_loss=0.2318, loss=0.1985, over 3177782.19 frames. utt_duration=1223 frames, utt_pad_proportion=0.06062, over 10403.33 utterances.], batch size: 47, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:44:08,960 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.5446, 1.9752, 2.1717, 2.2609, 2.1581, 2.2721, 1.9025, 2.4799], device='cuda:0'), covar=tensor([0.1101, 0.2101, 0.1583, 0.1204, 0.1778, 0.1192, 0.1328, 0.1195], device='cuda:0'), in_proj_covar=tensor([0.0145, 0.0147, 0.0142, 0.0137, 0.0154, 0.0132, 0.0153, 0.0131], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 10:44:46,728 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7445, 3.3105, 3.7580, 3.3769, 3.7037, 4.7634, 4.6236, 3.6774], device='cuda:0'), covar=tensor([0.0369, 0.1606, 0.1262, 0.1231, 0.1094, 0.0894, 0.0577, 0.1030], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0251, 0.0293, 0.0221, 0.0273, 0.0383, 0.0275, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:44:50,541 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.17 vs. limit=5.0 2023-03-09 10:45:27,465 INFO [train2.py:809] (0/4) Epoch 28, batch 750, loss[ctc_loss=0.05367, att_loss=0.2035, loss=0.1735, over 15646.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008845, over 37.00 utterances.], tot_loss[ctc_loss=0.06552, att_loss=0.2316, loss=0.1984, over 3185768.96 frames. utt_duration=1208 frames, utt_pad_proportion=0.06925, over 10561.55 utterances.], batch size: 37, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:45:32,965 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108315.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:45:48,262 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108325.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:46:33,461 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.198e+02 1.877e+02 2.184e+02 2.631e+02 4.478e+02, threshold=4.368e+02, percent-clipped=1.0 2023-03-09 10:46:45,997 INFO [train2.py:809] (0/4) Epoch 28, batch 800, loss[ctc_loss=0.07265, att_loss=0.2371, loss=0.2042, over 16973.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007144, over 50.00 utterances.], tot_loss[ctc_loss=0.06535, att_loss=0.2316, loss=0.1983, over 3206593.95 frames. utt_duration=1234 frames, utt_pad_proportion=0.06201, over 10405.19 utterances.], batch size: 50, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:46:47,659 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108363.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:47:10,570 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3834, 2.6789, 4.8078, 3.7364, 2.9283, 4.0883, 4.5542, 4.5215], device='cuda:0'), covar=tensor([0.0295, 0.1500, 0.0253, 0.0918, 0.1747, 0.0294, 0.0245, 0.0282], device='cuda:0'), in_proj_covar=tensor([0.0233, 0.0250, 0.0227, 0.0329, 0.0273, 0.0242, 0.0217, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:47:23,504 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108386.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:47:34,649 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108393.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:48:04,322 INFO [train2.py:809] (0/4) Epoch 28, batch 850, loss[ctc_loss=0.1075, att_loss=0.2572, loss=0.2273, over 14321.00 frames. utt_duration=393.8 frames, utt_pad_proportion=0.314, over 146.00 utterances.], tot_loss[ctc_loss=0.06564, att_loss=0.2317, loss=0.1985, over 3223888.89 frames. utt_duration=1221 frames, utt_pad_proportion=0.06339, over 10577.01 utterances.], batch size: 146, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:48:27,352 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8914, 6.1415, 5.6062, 5.8157, 5.8287, 5.2783, 5.6653, 5.2919], device='cuda:0'), covar=tensor([0.1195, 0.0909, 0.1131, 0.0818, 0.0955, 0.1655, 0.2114, 0.2560], device='cuda:0'), in_proj_covar=tensor([0.0564, 0.0646, 0.0492, 0.0481, 0.0459, 0.0487, 0.0648, 0.0552], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 10:48:35,083 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108432.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:48:49,680 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=108441.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:49:03,428 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.13 vs. limit=5.0 2023-03-09 10:49:10,175 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.210e+02 1.750e+02 2.196e+02 2.771e+02 5.305e+02, threshold=4.392e+02, percent-clipped=2.0 2023-03-09 10:49:23,196 INFO [train2.py:809] (0/4) Epoch 28, batch 900, loss[ctc_loss=0.05803, att_loss=0.2431, loss=0.2061, over 16951.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007602, over 50.00 utterances.], tot_loss[ctc_loss=0.06505, att_loss=0.2318, loss=0.1985, over 3240170.00 frames. utt_duration=1257 frames, utt_pad_proportion=0.05341, over 10323.21 utterances.], batch size: 50, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:50:12,154 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108493.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 10:50:31,037 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2261, 4.2779, 4.3663, 4.3504, 4.9167, 4.2162, 4.2992, 2.4597], device='cuda:0'), covar=tensor([0.0353, 0.0522, 0.0410, 0.0351, 0.0610, 0.0313, 0.0409, 0.1847], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0221, 0.0216, 0.0233, 0.0379, 0.0192, 0.0207, 0.0220], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:50:42,012 INFO [train2.py:809] (0/4) Epoch 28, batch 950, loss[ctc_loss=0.04791, att_loss=0.2277, loss=0.1918, over 16950.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008327, over 50.00 utterances.], tot_loss[ctc_loss=0.06457, att_loss=0.2312, loss=0.1979, over 3240273.78 frames. utt_duration=1269 frames, utt_pad_proportion=0.05122, over 10223.32 utterances.], batch size: 50, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:51:36,938 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4062, 2.8761, 3.5412, 4.4723, 3.9792, 3.9461, 3.0167, 2.4324], device='cuda:0'), covar=tensor([0.0807, 0.2082, 0.0814, 0.0531, 0.0963, 0.0536, 0.1494, 0.2028], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0221, 0.0188, 0.0228, 0.0236, 0.0194, 0.0206, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:51:47,830 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 1.763e+02 2.072e+02 2.486e+02 7.743e+02, threshold=4.143e+02, percent-clipped=3.0 2023-03-09 10:52:00,758 INFO [train2.py:809] (0/4) Epoch 28, batch 1000, loss[ctc_loss=0.07837, att_loss=0.2279, loss=0.198, over 15967.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.005724, over 41.00 utterances.], tot_loss[ctc_loss=0.06457, att_loss=0.2309, loss=0.1977, over 3247831.13 frames. utt_duration=1281 frames, utt_pad_proportion=0.04832, over 10156.24 utterances.], batch size: 41, lr: 3.85e-03, grad_scale: 16.0 2023-03-09 10:52:08,849 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3177, 3.9247, 3.4400, 3.5885, 4.1474, 3.7824, 3.2927, 4.3497], device='cuda:0'), covar=tensor([0.0899, 0.0527, 0.1085, 0.0739, 0.0732, 0.0709, 0.0825, 0.0506], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0228, 0.0233, 0.0210, 0.0293, 0.0252, 0.0207, 0.0299], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 10:52:24,724 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108577.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:53:13,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 10:53:20,714 INFO [train2.py:809] (0/4) Epoch 28, batch 1050, loss[ctc_loss=0.05562, att_loss=0.2214, loss=0.1882, over 16040.00 frames. utt_duration=1605 frames, utt_pad_proportion=0.004991, over 40.00 utterances.], tot_loss[ctc_loss=0.06458, att_loss=0.2311, loss=0.1978, over 3250127.80 frames. utt_duration=1264 frames, utt_pad_proportion=0.05336, over 10300.80 utterances.], batch size: 40, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 10:53:50,544 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.46 vs. limit=2.0 2023-03-09 10:54:02,723 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108638.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:54:27,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 1.905e+02 2.250e+02 2.754e+02 7.007e+02, threshold=4.500e+02, percent-clipped=3.0 2023-03-09 10:54:40,458 INFO [train2.py:809] (0/4) Epoch 28, batch 1100, loss[ctc_loss=0.06881, att_loss=0.2489, loss=0.2129, over 16469.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006538, over 46.00 utterances.], tot_loss[ctc_loss=0.06504, att_loss=0.2316, loss=0.1983, over 3244701.30 frames. utt_duration=1218 frames, utt_pad_proportion=0.06662, over 10670.76 utterances.], batch size: 46, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 10:55:10,654 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108681.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:55:16,895 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4826, 4.4274, 4.6081, 4.6301, 5.1800, 4.4584, 4.4711, 2.5645], device='cuda:0'), covar=tensor([0.0295, 0.0545, 0.0370, 0.0351, 0.0625, 0.0280, 0.0413, 0.1777], device='cuda:0'), in_proj_covar=tensor([0.0192, 0.0220, 0.0215, 0.0231, 0.0377, 0.0191, 0.0206, 0.0219], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:56:00,161 INFO [train2.py:809] (0/4) Epoch 28, batch 1150, loss[ctc_loss=0.03458, att_loss=0.2034, loss=0.1696, over 15501.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.008674, over 36.00 utterances.], tot_loss[ctc_loss=0.06503, att_loss=0.232, loss=0.1986, over 3257339.59 frames. utt_duration=1216 frames, utt_pad_proportion=0.06424, over 10727.09 utterances.], batch size: 36, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 10:56:08,399 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7240, 2.5036, 2.6745, 2.5520, 2.9595, 2.8865, 2.5181, 3.2260], device='cuda:0'), covar=tensor([0.1605, 0.2148, 0.1615, 0.1489, 0.1401, 0.1014, 0.1993, 0.1101], device='cuda:0'), in_proj_covar=tensor([0.0146, 0.0147, 0.0143, 0.0138, 0.0155, 0.0132, 0.0154, 0.0132], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 10:57:07,175 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 1.925e+02 2.338e+02 2.824e+02 7.096e+02, threshold=4.677e+02, percent-clipped=6.0 2023-03-09 10:57:10,326 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4557, 3.0042, 3.6409, 3.0168, 3.5400, 4.5838, 4.4505, 3.3546], device='cuda:0'), covar=tensor([0.0414, 0.1727, 0.1226, 0.1324, 0.1095, 0.0820, 0.0494, 0.1180], device='cuda:0'), in_proj_covar=tensor([0.0246, 0.0248, 0.0287, 0.0217, 0.0268, 0.0377, 0.0270, 0.0234], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 10:57:19,309 INFO [train2.py:809] (0/4) Epoch 28, batch 1200, loss[ctc_loss=0.0721, att_loss=0.2504, loss=0.2148, over 17450.00 frames. utt_duration=1013 frames, utt_pad_proportion=0.04442, over 69.00 utterances.], tot_loss[ctc_loss=0.06476, att_loss=0.2314, loss=0.1981, over 3259508.88 frames. utt_duration=1228 frames, utt_pad_proportion=0.06079, over 10629.32 utterances.], batch size: 69, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 10:58:00,720 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108788.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 10:58:05,460 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=108791.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:58:39,210 INFO [train2.py:809] (0/4) Epoch 28, batch 1250, loss[ctc_loss=0.08515, att_loss=0.2479, loss=0.2153, over 17051.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009789, over 53.00 utterances.], tot_loss[ctc_loss=0.06455, att_loss=0.231, loss=0.1977, over 3262819.13 frames. utt_duration=1231 frames, utt_pad_proportion=0.06012, over 10613.32 utterances.], batch size: 53, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 10:59:17,516 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0387, 6.2455, 5.7310, 5.9702, 5.9366, 5.4050, 5.6972, 5.4587], device='cuda:0'), covar=tensor([0.1133, 0.0853, 0.0899, 0.0758, 0.0868, 0.1529, 0.2158, 0.2307], device='cuda:0'), in_proj_covar=tensor([0.0557, 0.0643, 0.0488, 0.0475, 0.0454, 0.0482, 0.0644, 0.0548], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 10:59:28,353 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1800, 3.7944, 3.7387, 3.2725, 3.7513, 3.8379, 3.8274, 2.8399], device='cuda:0'), covar=tensor([0.0985, 0.1340, 0.1802, 0.2849, 0.1048, 0.1913, 0.0779, 0.2968], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0209, 0.0226, 0.0275, 0.0185, 0.0286, 0.0208, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 10:59:42,560 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=108852.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 10:59:45,115 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.146e+02 1.829e+02 2.128e+02 2.551e+02 3.725e+02, threshold=4.255e+02, percent-clipped=0.0 2023-03-09 10:59:58,167 INFO [train2.py:809] (0/4) Epoch 28, batch 1300, loss[ctc_loss=0.05931, att_loss=0.2185, loss=0.1867, over 15989.00 frames. utt_duration=1600 frames, utt_pad_proportion=0.008473, over 40.00 utterances.], tot_loss[ctc_loss=0.06417, att_loss=0.2306, loss=0.1973, over 3269357.45 frames. utt_duration=1267 frames, utt_pad_proportion=0.05078, over 10333.30 utterances.], batch size: 40, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 11:00:34,512 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6824, 5.0386, 4.8636, 5.0271, 5.1701, 4.7764, 3.5103, 4.9927], device='cuda:0'), covar=tensor([0.0131, 0.0112, 0.0160, 0.0085, 0.0091, 0.0114, 0.0699, 0.0174], device='cuda:0'), in_proj_covar=tensor([0.0096, 0.0092, 0.0116, 0.0073, 0.0079, 0.0089, 0.0105, 0.0109], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 11:01:13,670 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1083, 4.5368, 4.6043, 4.7022, 2.9644, 4.5548, 2.8037, 2.0117], device='cuda:0'), covar=tensor([0.0477, 0.0284, 0.0582, 0.0348, 0.1371, 0.0226, 0.1395, 0.1544], device='cuda:0'), in_proj_covar=tensor([0.0219, 0.0187, 0.0268, 0.0183, 0.0224, 0.0169, 0.0236, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:01:17,643 INFO [train2.py:809] (0/4) Epoch 28, batch 1350, loss[ctc_loss=0.06741, att_loss=0.231, loss=0.1983, over 15961.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006113, over 41.00 utterances.], tot_loss[ctc_loss=0.06477, att_loss=0.231, loss=0.1978, over 3274552.04 frames. utt_duration=1259 frames, utt_pad_proportion=0.05035, over 10419.91 utterances.], batch size: 41, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 11:01:23,448 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.38 vs. limit=2.0 2023-03-09 11:01:50,630 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=108933.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:02:24,360 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.034e+02 1.787e+02 2.090e+02 2.740e+02 6.253e+02, threshold=4.180e+02, percent-clipped=4.0 2023-03-09 11:02:36,712 INFO [train2.py:809] (0/4) Epoch 28, batch 1400, loss[ctc_loss=0.05606, att_loss=0.2272, loss=0.193, over 16322.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006536, over 45.00 utterances.], tot_loss[ctc_loss=0.06458, att_loss=0.2308, loss=0.1975, over 3276074.88 frames. utt_duration=1270 frames, utt_pad_proportion=0.04758, over 10331.15 utterances.], batch size: 45, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 11:03:07,504 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=108981.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:03:26,012 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1155, 4.5177, 4.7606, 4.6445, 2.7141, 4.5344, 3.0186, 1.6602], device='cuda:0'), covar=tensor([0.0509, 0.0285, 0.0570, 0.0262, 0.1556, 0.0248, 0.1262, 0.1770], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0189, 0.0269, 0.0184, 0.0226, 0.0171, 0.0237, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:03:56,707 INFO [train2.py:809] (0/4) Epoch 28, batch 1450, loss[ctc_loss=0.05629, att_loss=0.2258, loss=0.1919, over 15946.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.007275, over 41.00 utterances.], tot_loss[ctc_loss=0.065, att_loss=0.2315, loss=0.1982, over 3267103.48 frames. utt_duration=1203 frames, utt_pad_proportion=0.06753, over 10879.41 utterances.], batch size: 41, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 11:04:24,404 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109029.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:04:36,919 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109037.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 11:04:38,416 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7762, 3.4956, 3.5233, 3.0883, 3.5229, 3.5330, 3.5458, 2.6518], device='cuda:0'), covar=tensor([0.1012, 0.1452, 0.1373, 0.2380, 0.0710, 0.1394, 0.0760, 0.2828], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0209, 0.0225, 0.0275, 0.0185, 0.0286, 0.0207, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:05:03,520 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.105e+02 1.756e+02 2.072e+02 2.557e+02 4.820e+02, threshold=4.143e+02, percent-clipped=2.0 2023-03-09 11:05:06,504 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-09 11:05:16,007 INFO [train2.py:809] (0/4) Epoch 28, batch 1500, loss[ctc_loss=0.06857, att_loss=0.2311, loss=0.1986, over 17009.00 frames. utt_duration=1335 frames, utt_pad_proportion=0.008545, over 51.00 utterances.], tot_loss[ctc_loss=0.06501, att_loss=0.2316, loss=0.1983, over 3273088.29 frames. utt_duration=1219 frames, utt_pad_proportion=0.06158, over 10750.09 utterances.], batch size: 51, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 11:05:29,907 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9711, 5.2702, 5.5286, 5.3870, 5.4752, 5.9231, 5.2522, 5.9929], device='cuda:0'), covar=tensor([0.0821, 0.0779, 0.0884, 0.1514, 0.1997, 0.1027, 0.0757, 0.0796], device='cuda:0'), in_proj_covar=tensor([0.0924, 0.0531, 0.0651, 0.0688, 0.0923, 0.0670, 0.0517, 0.0652], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 11:05:57,470 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109088.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:06:13,801 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109098.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 11:06:34,719 INFO [train2.py:809] (0/4) Epoch 28, batch 1550, loss[ctc_loss=0.06435, att_loss=0.2271, loss=0.1945, over 16113.00 frames. utt_duration=1536 frames, utt_pad_proportion=0.006926, over 42.00 utterances.], tot_loss[ctc_loss=0.06455, att_loss=0.2312, loss=0.1979, over 3265973.43 frames. utt_duration=1221 frames, utt_pad_proportion=0.0637, over 10716.43 utterances.], batch size: 42, lr: 3.84e-03, grad_scale: 16.0 2023-03-09 11:07:05,336 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8302, 4.3914, 4.4667, 2.3353, 2.2366, 3.0192, 2.4778, 3.6326], device='cuda:0'), covar=tensor([0.0711, 0.0368, 0.0277, 0.4190, 0.4492, 0.2098, 0.3341, 0.1546], device='cuda:0'), in_proj_covar=tensor([0.0364, 0.0303, 0.0281, 0.0255, 0.0342, 0.0334, 0.0266, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 11:07:12,425 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109136.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:07:30,574 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109147.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:07:41,450 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.161e+02 1.952e+02 2.217e+02 2.637e+02 4.191e+02, threshold=4.435e+02, percent-clipped=1.0 2023-03-09 11:07:53,851 INFO [train2.py:809] (0/4) Epoch 28, batch 1600, loss[ctc_loss=0.06165, att_loss=0.2525, loss=0.2143, over 17118.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.01497, over 56.00 utterances.], tot_loss[ctc_loss=0.06436, att_loss=0.2309, loss=0.1976, over 3268116.65 frames. utt_duration=1236 frames, utt_pad_proportion=0.05887, over 10585.63 utterances.], batch size: 56, lr: 3.83e-03, grad_scale: 16.0 2023-03-09 11:08:44,703 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109194.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:09:12,890 INFO [train2.py:809] (0/4) Epoch 28, batch 1650, loss[ctc_loss=0.06314, att_loss=0.2324, loss=0.1986, over 16324.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006642, over 45.00 utterances.], tot_loss[ctc_loss=0.06459, att_loss=0.2311, loss=0.1978, over 3274675.10 frames. utt_duration=1234 frames, utt_pad_proportion=0.0585, over 10627.84 utterances.], batch size: 45, lr: 3.83e-03, grad_scale: 16.0 2023-03-09 11:09:16,174 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0513, 5.2869, 5.2434, 5.2964, 5.3706, 5.3023, 4.9518, 4.7900], device='cuda:0'), covar=tensor([0.1041, 0.0562, 0.0311, 0.0470, 0.0282, 0.0324, 0.0442, 0.0329], device='cuda:0'), in_proj_covar=tensor([0.0539, 0.0384, 0.0377, 0.0380, 0.0444, 0.0450, 0.0380, 0.0417], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 11:09:34,155 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8721, 3.6023, 3.5680, 3.1068, 3.6015, 3.6706, 3.6561, 2.6475], device='cuda:0'), covar=tensor([0.1048, 0.1143, 0.1493, 0.2729, 0.1095, 0.2090, 0.0813, 0.3299], device='cuda:0'), in_proj_covar=tensor([0.0204, 0.0208, 0.0225, 0.0274, 0.0185, 0.0286, 0.0208, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:09:47,302 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109233.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:09:47,321 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3333, 2.4516, 2.8045, 4.2087, 3.8763, 3.9380, 2.9282, 2.3634], device='cuda:0'), covar=tensor([0.0743, 0.2249, 0.1304, 0.0627, 0.0710, 0.0426, 0.1370, 0.2033], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0219, 0.0188, 0.0226, 0.0234, 0.0192, 0.0205, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:10:22,239 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.344e+02 1.779e+02 2.256e+02 2.612e+02 6.095e+02, threshold=4.513e+02, percent-clipped=4.0 2023-03-09 11:10:22,660 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109255.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:10:33,278 INFO [train2.py:809] (0/4) Epoch 28, batch 1700, loss[ctc_loss=0.1036, att_loss=0.2586, loss=0.2276, over 17044.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008512, over 52.00 utterances.], tot_loss[ctc_loss=0.06538, att_loss=0.2319, loss=0.1986, over 3270980.15 frames. utt_duration=1218 frames, utt_pad_proportion=0.06275, over 10755.56 utterances.], batch size: 52, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:11:03,686 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109281.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:11:23,975 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109294.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:11:52,251 INFO [train2.py:809] (0/4) Epoch 28, batch 1750, loss[ctc_loss=0.07973, att_loss=0.242, loss=0.2095, over 17286.00 frames. utt_duration=876.9 frames, utt_pad_proportion=0.07983, over 79.00 utterances.], tot_loss[ctc_loss=0.06543, att_loss=0.2313, loss=0.1981, over 3268340.67 frames. utt_duration=1225 frames, utt_pad_proportion=0.061, over 10687.69 utterances.], batch size: 79, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:12:02,558 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109318.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:12:23,640 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.17 vs. limit=5.0 2023-03-09 11:13:00,660 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.203e+02 1.738e+02 2.027e+02 2.395e+02 5.989e+02, threshold=4.055e+02, percent-clipped=1.0 2023-03-09 11:13:01,124 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109355.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:13:11,724 INFO [train2.py:809] (0/4) Epoch 28, batch 1800, loss[ctc_loss=0.1022, att_loss=0.2583, loss=0.2271, over 16970.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007115, over 50.00 utterances.], tot_loss[ctc_loss=0.0648, att_loss=0.231, loss=0.1978, over 3270662.40 frames. utt_duration=1214 frames, utt_pad_proportion=0.06251, over 10794.22 utterances.], batch size: 50, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:13:39,562 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109379.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 11:14:01,187 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109393.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 11:14:21,176 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.33 vs. limit=2.0 2023-03-09 11:14:25,277 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0127, 3.9707, 3.7968, 2.6699, 3.7966, 3.7908, 3.4483, 2.6666], device='cuda:0'), covar=tensor([0.0120, 0.0158, 0.0271, 0.0992, 0.0143, 0.0390, 0.0380, 0.1282], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0107, 0.0113, 0.0113, 0.0090, 0.0119, 0.0103, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 11:14:31,015 INFO [train2.py:809] (0/4) Epoch 28, batch 1850, loss[ctc_loss=0.0829, att_loss=0.2402, loss=0.2087, over 16894.00 frames. utt_duration=1381 frames, utt_pad_proportion=0.006112, over 49.00 utterances.], tot_loss[ctc_loss=0.06526, att_loss=0.2317, loss=0.1984, over 3270079.18 frames. utt_duration=1223 frames, utt_pad_proportion=0.06052, over 10706.62 utterances.], batch size: 49, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:15:26,821 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109447.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:15:39,076 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.065e+02 1.724e+02 2.158e+02 2.874e+02 5.021e+02, threshold=4.317e+02, percent-clipped=1.0 2023-03-09 11:15:50,148 INFO [train2.py:809] (0/4) Epoch 28, batch 1900, loss[ctc_loss=0.06135, att_loss=0.2407, loss=0.2049, over 16624.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005394, over 47.00 utterances.], tot_loss[ctc_loss=0.06577, att_loss=0.2319, loss=0.1987, over 3264586.82 frames. utt_duration=1211 frames, utt_pad_proportion=0.06455, over 10798.02 utterances.], batch size: 47, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:16:43,229 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-09 11:16:43,560 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109495.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:17:09,534 INFO [train2.py:809] (0/4) Epoch 28, batch 1950, loss[ctc_loss=0.05941, att_loss=0.2352, loss=0.2001, over 16761.00 frames. utt_duration=678.8 frames, utt_pad_proportion=0.1483, over 99.00 utterances.], tot_loss[ctc_loss=0.06586, att_loss=0.2322, loss=0.1989, over 3268249.19 frames. utt_duration=1192 frames, utt_pad_proportion=0.0685, over 10982.92 utterances.], batch size: 99, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:18:10,500 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109550.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:18:17,570 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.372e+02 1.926e+02 2.337e+02 2.750e+02 5.694e+02, threshold=4.673e+02, percent-clipped=4.0 2023-03-09 11:18:28,471 INFO [train2.py:809] (0/4) Epoch 28, batch 2000, loss[ctc_loss=0.06836, att_loss=0.2241, loss=0.1929, over 17327.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02229, over 59.00 utterances.], tot_loss[ctc_loss=0.06563, att_loss=0.2317, loss=0.1985, over 3266935.68 frames. utt_duration=1203 frames, utt_pad_proportion=0.06611, over 10875.36 utterances.], batch size: 59, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:18:55,932 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109579.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:19:00,578 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2878, 2.8390, 3.1809, 4.4144, 3.9307, 3.9191, 2.8156, 2.3560], device='cuda:0'), covar=tensor([0.0915, 0.2010, 0.0975, 0.0542, 0.0847, 0.0488, 0.1699, 0.2233], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0221, 0.0188, 0.0226, 0.0236, 0.0194, 0.0208, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:19:47,002 INFO [train2.py:809] (0/4) Epoch 28, batch 2050, loss[ctc_loss=0.0823, att_loss=0.2514, loss=0.2176, over 17118.00 frames. utt_duration=1224 frames, utt_pad_proportion=0.0152, over 56.00 utterances.], tot_loss[ctc_loss=0.0661, att_loss=0.2319, loss=0.1988, over 3268072.59 frames. utt_duration=1176 frames, utt_pad_proportion=0.07206, over 11129.60 utterances.], batch size: 56, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:20:33,211 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109640.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:20:48,892 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109650.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:20:56,143 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 1.837e+02 2.171e+02 2.517e+02 4.514e+02, threshold=4.342e+02, percent-clipped=0.0 2023-03-09 11:21:02,517 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=109659.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:21:06,770 INFO [train2.py:809] (0/4) Epoch 28, batch 2100, loss[ctc_loss=0.06052, att_loss=0.2306, loss=0.1966, over 16944.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008576, over 50.00 utterances.], tot_loss[ctc_loss=0.06589, att_loss=0.2319, loss=0.1987, over 3271378.80 frames. utt_duration=1189 frames, utt_pad_proportion=0.06862, over 11015.19 utterances.], batch size: 50, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:21:26,407 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109674.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 11:21:57,847 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109693.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 11:22:26,683 INFO [train2.py:809] (0/4) Epoch 28, batch 2150, loss[ctc_loss=0.06328, att_loss=0.2347, loss=0.2004, over 16750.00 frames. utt_duration=1397 frames, utt_pad_proportion=0.007635, over 48.00 utterances.], tot_loss[ctc_loss=0.06631, att_loss=0.2319, loss=0.1988, over 3264918.86 frames. utt_duration=1201 frames, utt_pad_proportion=0.06822, over 10887.86 utterances.], batch size: 48, lr: 3.83e-03, grad_scale: 8.0 2023-03-09 11:22:40,141 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=109720.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:23:13,181 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109741.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 11:23:34,194 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.122e+02 1.864e+02 2.379e+02 2.851e+02 5.187e+02, threshold=4.758e+02, percent-clipped=2.0 2023-03-09 11:23:45,111 INFO [train2.py:809] (0/4) Epoch 28, batch 2200, loss[ctc_loss=0.06029, att_loss=0.2274, loss=0.194, over 17138.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01346, over 56.00 utterances.], tot_loss[ctc_loss=0.06635, att_loss=0.2322, loss=0.199, over 3272234.22 frames. utt_duration=1214 frames, utt_pad_proportion=0.06142, over 10797.75 utterances.], batch size: 56, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:24:31,785 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0541, 4.4414, 4.5625, 4.6789, 2.8832, 4.4154, 2.9648, 1.8410], device='cuda:0'), covar=tensor([0.0519, 0.0282, 0.0571, 0.0234, 0.1441, 0.0237, 0.1392, 0.1645], device='cuda:0'), in_proj_covar=tensor([0.0220, 0.0190, 0.0268, 0.0183, 0.0225, 0.0171, 0.0235, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:25:03,654 INFO [train2.py:809] (0/4) Epoch 28, batch 2250, loss[ctc_loss=0.07058, att_loss=0.2219, loss=0.1916, over 15512.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.00813, over 36.00 utterances.], tot_loss[ctc_loss=0.06708, att_loss=0.2327, loss=0.1996, over 3271597.95 frames. utt_duration=1213 frames, utt_pad_proportion=0.06169, over 10804.28 utterances.], batch size: 36, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:26:03,229 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109850.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:26:10,586 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.320e+02 1.849e+02 2.212e+02 2.625e+02 5.541e+02, threshold=4.423e+02, percent-clipped=2.0 2023-03-09 11:26:21,265 INFO [train2.py:809] (0/4) Epoch 28, batch 2300, loss[ctc_loss=0.06723, att_loss=0.225, loss=0.1934, over 16411.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006336, over 44.00 utterances.], tot_loss[ctc_loss=0.06734, att_loss=0.2329, loss=0.1998, over 3272597.40 frames. utt_duration=1217 frames, utt_pad_proportion=0.0607, over 10765.56 utterances.], batch size: 44, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:27:17,541 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109898.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:27:28,036 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.1909, 6.3175, 5.7974, 6.0078, 6.0930, 5.5531, 5.9262, 5.5152], device='cuda:0'), covar=tensor([0.0921, 0.0701, 0.1001, 0.0751, 0.0691, 0.1308, 0.1648, 0.2191], device='cuda:0'), in_proj_covar=tensor([0.0562, 0.0642, 0.0493, 0.0480, 0.0455, 0.0484, 0.0647, 0.0547], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 11:27:37,413 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9090, 3.6373, 3.6140, 3.1000, 3.6698, 3.6637, 3.7228, 2.6923], device='cuda:0'), covar=tensor([0.0956, 0.1037, 0.1691, 0.2713, 0.0922, 0.1914, 0.0657, 0.2845], device='cuda:0'), in_proj_covar=tensor([0.0205, 0.0209, 0.0226, 0.0276, 0.0186, 0.0288, 0.0209, 0.0233], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 11:27:38,454 INFO [train2.py:809] (0/4) Epoch 28, batch 2350, loss[ctc_loss=0.07158, att_loss=0.248, loss=0.2127, over 16774.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006081, over 48.00 utterances.], tot_loss[ctc_loss=0.06747, att_loss=0.2327, loss=0.1997, over 3274408.49 frames. utt_duration=1225 frames, utt_pad_proportion=0.05745, over 10708.96 utterances.], batch size: 48, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:28:16,092 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=109935.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:28:39,542 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109950.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:28:46,856 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.356e+02 1.839e+02 2.224e+02 2.680e+02 5.661e+02, threshold=4.448e+02, percent-clipped=2.0 2023-03-09 11:28:57,734 INFO [train2.py:809] (0/4) Epoch 28, batch 2400, loss[ctc_loss=0.04497, att_loss=0.1998, loss=0.1688, over 15760.00 frames. utt_duration=1660 frames, utt_pad_proportion=0.009327, over 38.00 utterances.], tot_loss[ctc_loss=0.06623, att_loss=0.2319, loss=0.1988, over 3277258.29 frames. utt_duration=1233 frames, utt_pad_proportion=0.05498, over 10642.74 utterances.], batch size: 38, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:29:17,539 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=109974.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:29:55,766 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=109998.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:29:59,108 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-110000.pt 2023-03-09 11:30:22,866 INFO [train2.py:809] (0/4) Epoch 28, batch 2450, loss[ctc_loss=0.06056, att_loss=0.2269, loss=0.1936, over 15934.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007606, over 41.00 utterances.], tot_loss[ctc_loss=0.06565, att_loss=0.2314, loss=0.1983, over 3273574.49 frames. utt_duration=1236 frames, utt_pad_proportion=0.05581, over 10605.15 utterances.], batch size: 41, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:30:28,512 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110015.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:30:31,889 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110017.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:30:39,403 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110022.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:31:14,794 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110044.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:31:31,406 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.139e+02 1.872e+02 2.152e+02 2.645e+02 5.223e+02, threshold=4.304e+02, percent-clipped=1.0 2023-03-09 11:31:41,819 INFO [train2.py:809] (0/4) Epoch 28, batch 2500, loss[ctc_loss=0.07896, att_loss=0.2289, loss=0.1989, over 15956.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006933, over 41.00 utterances.], tot_loss[ctc_loss=0.06588, att_loss=0.232, loss=0.1988, over 3271835.16 frames. utt_duration=1233 frames, utt_pad_proportion=0.05698, over 10623.76 utterances.], batch size: 41, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:31:53,648 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3199, 4.6914, 3.9299, 4.7936, 4.2474, 4.3554, 4.6731, 4.6022], device='cuda:0'), covar=tensor([0.0682, 0.0384, 0.1069, 0.0402, 0.0405, 0.0599, 0.0380, 0.0255], device='cuda:0'), in_proj_covar=tensor([0.0407, 0.0344, 0.0383, 0.0382, 0.0340, 0.0248, 0.0325, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 11:32:07,575 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110078.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:32:50,547 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110105.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:32:58,061 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110110.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:33:01,505 INFO [train2.py:809] (0/4) Epoch 28, batch 2550, loss[ctc_loss=0.0536, att_loss=0.2302, loss=0.1949, over 17350.00 frames. utt_duration=879.9 frames, utt_pad_proportion=0.07959, over 79.00 utterances.], tot_loss[ctc_loss=0.06461, att_loss=0.2313, loss=0.198, over 3272554.29 frames. utt_duration=1247 frames, utt_pad_proportion=0.0522, over 10507.76 utterances.], batch size: 79, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:33:07,172 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110115.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:34:08,240 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.183e+02 1.726e+02 2.160e+02 2.628e+02 5.720e+02, threshold=4.319e+02, percent-clipped=3.0 2023-03-09 11:34:19,904 INFO [train2.py:809] (0/4) Epoch 28, batch 2600, loss[ctc_loss=0.05173, att_loss=0.226, loss=0.1911, over 16622.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005274, over 47.00 utterances.], tot_loss[ctc_loss=0.06533, att_loss=0.2315, loss=0.1983, over 3273389.24 frames. utt_duration=1230 frames, utt_pad_proportion=0.05714, over 10657.69 utterances.], batch size: 47, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:34:34,591 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110171.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:34:42,138 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110176.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:35:21,696 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2530, 3.7834, 3.2959, 3.5204, 4.0694, 3.7418, 3.0744, 4.3734], device='cuda:0'), covar=tensor([0.0893, 0.0553, 0.1048, 0.0680, 0.0721, 0.0743, 0.0892, 0.0410], device='cuda:0'), in_proj_covar=tensor([0.0210, 0.0230, 0.0234, 0.0211, 0.0292, 0.0252, 0.0207, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 11:35:38,092 INFO [train2.py:809] (0/4) Epoch 28, batch 2650, loss[ctc_loss=0.05074, att_loss=0.2222, loss=0.1879, over 16278.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007356, over 43.00 utterances.], tot_loss[ctc_loss=0.06483, att_loss=0.231, loss=0.1977, over 3271722.49 frames. utt_duration=1236 frames, utt_pad_proportion=0.05698, over 10597.08 utterances.], batch size: 43, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:36:16,247 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110235.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:36:43,967 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3983, 4.8108, 4.6954, 4.8010, 4.9436, 4.6060, 3.4416, 4.8416], device='cuda:0'), covar=tensor([0.0149, 0.0125, 0.0143, 0.0078, 0.0083, 0.0112, 0.0677, 0.0192], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0093, 0.0118, 0.0073, 0.0080, 0.0091, 0.0106, 0.0111], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 11:36:46,734 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.140e+02 1.822e+02 2.163e+02 2.650e+02 7.141e+02, threshold=4.325e+02, percent-clipped=4.0 2023-03-09 11:36:57,426 INFO [train2.py:809] (0/4) Epoch 28, batch 2700, loss[ctc_loss=0.05668, att_loss=0.2307, loss=0.1959, over 16624.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005409, over 47.00 utterances.], tot_loss[ctc_loss=0.06529, att_loss=0.2319, loss=0.1986, over 3277255.63 frames. utt_duration=1234 frames, utt_pad_proportion=0.05666, over 10639.09 utterances.], batch size: 47, lr: 3.82e-03, grad_scale: 8.0 2023-03-09 11:37:32,521 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:37:32,796 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110283.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:38:03,417 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110303.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:38:16,815 INFO [train2.py:809] (0/4) Epoch 28, batch 2750, loss[ctc_loss=0.0746, att_loss=0.2428, loss=0.2092, over 16620.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005143, over 47.00 utterances.], tot_loss[ctc_loss=0.0656, att_loss=0.2322, loss=0.1988, over 3276680.54 frames. utt_duration=1222 frames, utt_pad_proportion=0.05993, over 10736.29 utterances.], batch size: 47, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:38:22,365 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110315.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:38:27,625 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110318.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:39:08,946 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110344.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:39:13,449 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1559, 5.4805, 5.0483, 5.5106, 4.8793, 5.1041, 5.5759, 5.3523], device='cuda:0'), covar=tensor([0.0579, 0.0257, 0.0734, 0.0350, 0.0385, 0.0224, 0.0222, 0.0199], device='cuda:0'), in_proj_covar=tensor([0.0406, 0.0344, 0.0382, 0.0382, 0.0341, 0.0247, 0.0324, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 11:39:25,560 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.237e+02 1.842e+02 2.112e+02 2.497e+02 5.270e+02, threshold=4.224e+02, percent-clipped=1.0 2023-03-09 11:39:36,856 INFO [train2.py:809] (0/4) Epoch 28, batch 2800, loss[ctc_loss=0.08252, att_loss=0.2565, loss=0.2217, over 17053.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.00976, over 53.00 utterances.], tot_loss[ctc_loss=0.06467, att_loss=0.2316, loss=0.1982, over 3276354.24 frames. utt_duration=1249 frames, utt_pad_proportion=0.05448, over 10505.25 utterances.], batch size: 53, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:39:38,483 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110363.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:39:40,824 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110364.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:39:54,204 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110373.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:40:04,269 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110379.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:40:37,359 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110400.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:40:55,888 INFO [train2.py:809] (0/4) Epoch 28, batch 2850, loss[ctc_loss=0.09774, att_loss=0.2524, loss=0.2215, over 17043.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009347, over 53.00 utterances.], tot_loss[ctc_loss=0.0648, att_loss=0.2317, loss=0.1983, over 3282273.48 frames. utt_duration=1275 frames, utt_pad_proportion=0.047, over 10309.47 utterances.], batch size: 53, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:42:02,951 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.137e+02 1.889e+02 2.295e+02 2.895e+02 8.128e+02, threshold=4.591e+02, percent-clipped=8.0 2023-03-09 11:42:14,319 INFO [train2.py:809] (0/4) Epoch 28, batch 2900, loss[ctc_loss=0.05149, att_loss=0.2034, loss=0.173, over 15494.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.008993, over 36.00 utterances.], tot_loss[ctc_loss=0.06535, att_loss=0.2316, loss=0.1983, over 3275843.25 frames. utt_duration=1264 frames, utt_pad_proportion=0.05199, over 10383.00 utterances.], batch size: 36, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:42:21,208 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110466.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:42:28,782 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110471.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:43:23,907 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-03-09 11:43:32,997 INFO [train2.py:809] (0/4) Epoch 28, batch 2950, loss[ctc_loss=0.04606, att_loss=0.2097, loss=0.177, over 16024.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006641, over 40.00 utterances.], tot_loss[ctc_loss=0.06481, att_loss=0.2312, loss=0.1979, over 3274319.13 frames. utt_duration=1260 frames, utt_pad_proportion=0.05334, over 10409.39 utterances.], batch size: 40, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:44:39,990 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.946e+01 1.771e+02 2.102e+02 2.638e+02 6.862e+02, threshold=4.204e+02, percent-clipped=2.0 2023-03-09 11:44:52,458 INFO [train2.py:809] (0/4) Epoch 28, batch 3000, loss[ctc_loss=0.08059, att_loss=0.2529, loss=0.2184, over 17311.00 frames. utt_duration=878.2 frames, utt_pad_proportion=0.07853, over 79.00 utterances.], tot_loss[ctc_loss=0.06462, att_loss=0.2312, loss=0.1979, over 3277769.51 frames. utt_duration=1268 frames, utt_pad_proportion=0.05013, over 10353.33 utterances.], batch size: 79, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:44:52,461 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 11:45:06,795 INFO [train2.py:843] (0/4) Epoch 28, validation: ctc_loss=0.04082, att_loss=0.2346, loss=0.1958, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 11:45:06,795 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 11:45:11,551 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1270, 2.6636, 2.8861, 4.2492, 3.7158, 3.8475, 2.7496, 2.2167], device='cuda:0'), covar=tensor([0.0907, 0.2047, 0.1075, 0.0515, 0.0920, 0.0547, 0.1687, 0.2225], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0222, 0.0188, 0.0230, 0.0236, 0.0195, 0.0209, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-03-09 11:45:33,697 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-09 11:46:25,714 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110611.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:46:26,958 INFO [train2.py:809] (0/4) Epoch 28, batch 3050, loss[ctc_loss=0.0619, att_loss=0.2255, loss=0.1928, over 16181.00 frames. utt_duration=1580 frames, utt_pad_proportion=0.006715, over 41.00 utterances.], tot_loss[ctc_loss=0.06428, att_loss=0.2305, loss=0.1973, over 3269797.10 frames. utt_duration=1273 frames, utt_pad_proportion=0.05151, over 10288.22 utterances.], batch size: 41, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:47:10,715 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110639.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:47:35,117 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.322e+02 1.799e+02 2.141e+02 2.649e+02 5.384e+02, threshold=4.281e+02, percent-clipped=1.0 2023-03-09 11:47:42,878 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110659.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:47:47,510 INFO [train2.py:809] (0/4) Epoch 28, batch 3100, loss[ctc_loss=0.09399, att_loss=0.2562, loss=0.2238, over 17420.00 frames. utt_duration=1108 frames, utt_pad_proportion=0.03252, over 63.00 utterances.], tot_loss[ctc_loss=0.0646, att_loss=0.2307, loss=0.1975, over 3273319.01 frames. utt_duration=1260 frames, utt_pad_proportion=0.05338, over 10401.01 utterances.], batch size: 63, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:48:03,600 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110672.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:48:05,185 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110673.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:48:06,645 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110674.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:48:48,120 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110700.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:49:07,615 INFO [train2.py:809] (0/4) Epoch 28, batch 3150, loss[ctc_loss=0.06757, att_loss=0.2425, loss=0.2075, over 17293.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01232, over 55.00 utterances.], tot_loss[ctc_loss=0.06418, att_loss=0.2314, loss=0.1979, over 3282297.65 frames. utt_duration=1263 frames, utt_pad_proportion=0.0499, over 10411.23 utterances.], batch size: 55, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:49:21,660 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110721.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:49:46,534 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110736.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:50:04,806 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110748.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:50:15,626 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.358e+02 1.871e+02 2.189e+02 2.637e+02 5.370e+02, threshold=4.378e+02, percent-clipped=2.0 2023-03-09 11:50:27,799 INFO [train2.py:809] (0/4) Epoch 28, batch 3200, loss[ctc_loss=0.05178, att_loss=0.2058, loss=0.175, over 15638.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009241, over 37.00 utterances.], tot_loss[ctc_loss=0.06392, att_loss=0.2314, loss=0.1979, over 3285206.04 frames. utt_duration=1244 frames, utt_pad_proportion=0.05296, over 10574.99 utterances.], batch size: 37, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:50:34,360 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110766.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:50:42,133 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110771.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:50:45,116 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3558, 5.0332, 5.0494, 5.1416, 5.0280, 5.1024, 4.9056, 4.6497], device='cuda:0'), covar=tensor([0.1786, 0.0893, 0.0473, 0.0568, 0.0684, 0.0493, 0.0454, 0.0429], device='cuda:0'), in_proj_covar=tensor([0.0531, 0.0375, 0.0371, 0.0375, 0.0436, 0.0442, 0.0375, 0.0409], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0004, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 11:51:01,872 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.32 vs. limit=5.0 2023-03-09 11:51:23,905 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110797.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 11:51:48,161 INFO [train2.py:809] (0/4) Epoch 28, batch 3250, loss[ctc_loss=0.05225, att_loss=0.234, loss=0.1976, over 16617.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006096, over 47.00 utterances.], tot_loss[ctc_loss=0.06358, att_loss=0.2317, loss=0.1981, over 3292888.87 frames. utt_duration=1268 frames, utt_pad_proportion=0.0453, over 10401.35 utterances.], batch size: 47, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:51:51,345 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110814.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:51:59,315 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110819.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:52:34,674 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.8711, 4.9512, 4.7752, 2.1517, 1.9863, 2.9866, 2.3340, 3.8291], device='cuda:0'), covar=tensor([0.0789, 0.0325, 0.0296, 0.4781, 0.5425, 0.2332, 0.4101, 0.1573], device='cuda:0'), in_proj_covar=tensor([0.0364, 0.0302, 0.0280, 0.0253, 0.0341, 0.0333, 0.0267, 0.0371], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 11:52:55,395 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 1.746e+02 2.093e+02 2.617e+02 5.212e+02, threshold=4.186e+02, percent-clipped=3.0 2023-03-09 11:53:07,976 INFO [train2.py:809] (0/4) Epoch 28, batch 3300, loss[ctc_loss=0.06273, att_loss=0.2097, loss=0.1803, over 15367.00 frames. utt_duration=1757 frames, utt_pad_proportion=0.01042, over 35.00 utterances.], tot_loss[ctc_loss=0.06315, att_loss=0.2304, loss=0.197, over 3279983.11 frames. utt_duration=1273 frames, utt_pad_proportion=0.04768, over 10315.92 utterances.], batch size: 35, lr: 3.81e-03, grad_scale: 8.0 2023-03-09 11:53:17,646 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110868.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:54:27,398 INFO [train2.py:809] (0/4) Epoch 28, batch 3350, loss[ctc_loss=0.07602, att_loss=0.2318, loss=0.2007, over 16324.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006582, over 45.00 utterances.], tot_loss[ctc_loss=0.06383, att_loss=0.2309, loss=0.1975, over 3287445.03 frames. utt_duration=1278 frames, utt_pad_proportion=0.04356, over 10300.84 utterances.], batch size: 45, lr: 3.80e-03, grad_scale: 8.0 2023-03-09 11:54:54,310 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=110929.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:55:10,279 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110939.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:55:36,026 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.217e+02 1.730e+02 2.154e+02 2.752e+02 9.084e+02, threshold=4.307e+02, percent-clipped=4.0 2023-03-09 11:55:41,061 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=110958.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:55:42,506 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110959.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:55:46,717 INFO [train2.py:809] (0/4) Epoch 28, batch 3400, loss[ctc_loss=0.07202, att_loss=0.2338, loss=0.2015, over 16328.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006247, over 45.00 utterances.], tot_loss[ctc_loss=0.0634, att_loss=0.2306, loss=0.1972, over 3282177.05 frames. utt_duration=1294 frames, utt_pad_proportion=0.04195, over 10158.90 utterances.], batch size: 45, lr: 3.80e-03, grad_scale: 8.0 2023-03-09 11:55:54,558 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=110967.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:56:05,565 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=110974.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:56:24,428 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-09 11:56:26,719 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=110987.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:56:38,525 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-03-09 11:56:58,887 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111007.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:57:06,333 INFO [train2.py:809] (0/4) Epoch 28, batch 3450, loss[ctc_loss=0.07827, att_loss=0.2455, loss=0.2121, over 17294.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01218, over 55.00 utterances.], tot_loss[ctc_loss=0.06415, att_loss=0.2314, loss=0.1979, over 3274312.44 frames. utt_duration=1256 frames, utt_pad_proportion=0.05353, over 10438.67 utterances.], batch size: 55, lr: 3.80e-03, grad_scale: 8.0 2023-03-09 11:57:17,517 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111019.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:57:21,907 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111022.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 11:58:14,706 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.190e+02 1.799e+02 2.104e+02 2.788e+02 6.212e+02, threshold=4.209e+02, percent-clipped=4.0 2023-03-09 11:58:25,585 INFO [train2.py:809] (0/4) Epoch 28, batch 3500, loss[ctc_loss=0.05685, att_loss=0.2236, loss=0.1903, over 15643.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008956, over 37.00 utterances.], tot_loss[ctc_loss=0.06431, att_loss=0.231, loss=0.1976, over 3274605.01 frames. utt_duration=1273 frames, utt_pad_proportion=0.04942, over 10299.78 utterances.], batch size: 37, lr: 3.80e-03, grad_scale: 8.0 2023-03-09 11:59:14,507 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111092.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 11:59:18,511 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-09 11:59:46,745 INFO [train2.py:809] (0/4) Epoch 28, batch 3550, loss[ctc_loss=0.1169, att_loss=0.265, loss=0.2354, over 13784.00 frames. utt_duration=376.6 frames, utt_pad_proportion=0.3393, over 147.00 utterances.], tot_loss[ctc_loss=0.06434, att_loss=0.2308, loss=0.1975, over 3251194.57 frames. utt_duration=1229 frames, utt_pad_proportion=0.0674, over 10593.11 utterances.], batch size: 147, lr: 3.80e-03, grad_scale: 8.0 2023-03-09 12:00:51,441 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.34 vs. limit=2.0 2023-03-09 12:00:55,331 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.124e+02 1.811e+02 2.154e+02 2.613e+02 6.278e+02, threshold=4.308e+02, percent-clipped=2.0 2023-03-09 12:01:05,927 INFO [train2.py:809] (0/4) Epoch 28, batch 3600, loss[ctc_loss=0.08841, att_loss=0.2551, loss=0.2217, over 16640.00 frames. utt_duration=1418 frames, utt_pad_proportion=0.004467, over 47.00 utterances.], tot_loss[ctc_loss=0.06497, att_loss=0.2315, loss=0.1982, over 3251499.32 frames. utt_duration=1216 frames, utt_pad_proportion=0.07111, over 10711.76 utterances.], batch size: 47, lr: 3.80e-03, grad_scale: 8.0 2023-03-09 12:02:20,209 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0461, 4.9471, 4.7918, 3.1152, 4.8192, 4.7784, 4.5159, 2.6697], device='cuda:0'), covar=tensor([0.0115, 0.0119, 0.0263, 0.0922, 0.0110, 0.0196, 0.0269, 0.1408], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0107, 0.0112, 0.0112, 0.0089, 0.0118, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:02:25,884 INFO [train2.py:809] (0/4) Epoch 28, batch 3650, loss[ctc_loss=0.08544, att_loss=0.2408, loss=0.2098, over 15961.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006175, over 41.00 utterances.], tot_loss[ctc_loss=0.06548, att_loss=0.2317, loss=0.1985, over 3259700.76 frames. utt_duration=1232 frames, utt_pad_proportion=0.0642, over 10596.33 utterances.], batch size: 41, lr: 3.80e-03, grad_scale: 8.0 2023-03-09 12:02:45,391 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111224.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:02:47,041 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111225.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:03:34,857 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.213e+02 1.792e+02 2.128e+02 2.694e+02 5.056e+02, threshold=4.257e+02, percent-clipped=2.0 2023-03-09 12:03:45,873 INFO [train2.py:809] (0/4) Epoch 28, batch 3700, loss[ctc_loss=0.05353, att_loss=0.2166, loss=0.184, over 15623.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009563, over 37.00 utterances.], tot_loss[ctc_loss=0.06503, att_loss=0.2311, loss=0.1979, over 3257326.72 frames. utt_duration=1227 frames, utt_pad_proportion=0.06594, over 10631.19 utterances.], batch size: 37, lr: 3.80e-03, grad_scale: 16.0 2023-03-09 12:03:54,455 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111267.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:04:10,433 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0824, 5.4545, 5.0172, 5.4691, 4.8719, 5.0423, 5.5519, 5.3365], device='cuda:0'), covar=tensor([0.0595, 0.0257, 0.0720, 0.0322, 0.0368, 0.0231, 0.0195, 0.0171], device='cuda:0'), in_proj_covar=tensor([0.0402, 0.0338, 0.0378, 0.0379, 0.0337, 0.0244, 0.0319, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 12:04:15,118 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9561, 5.2518, 5.4795, 5.3134, 5.4459, 5.9220, 5.2245, 6.0045], device='cuda:0'), covar=tensor([0.0735, 0.0777, 0.0883, 0.1483, 0.1873, 0.0884, 0.0800, 0.0661], device='cuda:0'), in_proj_covar=tensor([0.0925, 0.0536, 0.0655, 0.0692, 0.0926, 0.0668, 0.0517, 0.0652], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 12:04:24,217 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111286.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:05:06,060 INFO [train2.py:809] (0/4) Epoch 28, batch 3750, loss[ctc_loss=0.1029, att_loss=0.249, loss=0.2198, over 16555.00 frames. utt_duration=1473 frames, utt_pad_proportion=0.005327, over 45.00 utterances.], tot_loss[ctc_loss=0.06519, att_loss=0.2313, loss=0.1981, over 3262947.41 frames. utt_duration=1237 frames, utt_pad_proportion=0.06281, over 10563.92 utterances.], batch size: 45, lr: 3.80e-03, grad_scale: 16.0 2023-03-09 12:05:09,368 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111314.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:05:10,808 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111315.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:06:11,044 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4412, 3.4151, 2.8329, 2.9187, 3.4630, 3.2905, 2.4465, 3.4944], device='cuda:0'), covar=tensor([0.1383, 0.0510, 0.1144, 0.0871, 0.0918, 0.0727, 0.1185, 0.0498], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0230, 0.0233, 0.0210, 0.0291, 0.0251, 0.0206, 0.0300], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0004], device='cuda:0') 2023-03-09 12:06:13,701 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.228e+02 1.865e+02 2.265e+02 2.758e+02 6.117e+02, threshold=4.530e+02, percent-clipped=1.0 2023-03-09 12:06:25,565 INFO [train2.py:809] (0/4) Epoch 28, batch 3800, loss[ctc_loss=0.05952, att_loss=0.2001, loss=0.172, over 15884.00 frames. utt_duration=1631 frames, utt_pad_proportion=0.009269, over 39.00 utterances.], tot_loss[ctc_loss=0.06532, att_loss=0.2315, loss=0.1982, over 3268514.88 frames. utt_duration=1229 frames, utt_pad_proportion=0.0606, over 10650.37 utterances.], batch size: 39, lr: 3.80e-03, grad_scale: 16.0 2023-03-09 12:07:04,168 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7026, 5.0833, 4.9003, 5.1071, 5.2159, 4.8429, 3.5404, 5.1198], device='cuda:0'), covar=tensor([0.0133, 0.0116, 0.0144, 0.0081, 0.0088, 0.0108, 0.0732, 0.0175], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0093, 0.0119, 0.0073, 0.0080, 0.0091, 0.0107, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 12:07:13,925 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111392.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 12:07:44,851 INFO [train2.py:809] (0/4) Epoch 28, batch 3850, loss[ctc_loss=0.04265, att_loss=0.2154, loss=0.1808, over 16420.00 frames. utt_duration=1495 frames, utt_pad_proportion=0.006271, over 44.00 utterances.], tot_loss[ctc_loss=0.06534, att_loss=0.2325, loss=0.1991, over 3276939.49 frames. utt_duration=1215 frames, utt_pad_proportion=0.06099, over 10800.77 utterances.], batch size: 44, lr: 3.80e-03, grad_scale: 16.0 2023-03-09 12:07:54,349 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8394, 5.1505, 5.0034, 5.1726, 5.3201, 4.8798, 3.7678, 5.2377], device='cuda:0'), covar=tensor([0.0118, 0.0106, 0.0143, 0.0076, 0.0084, 0.0121, 0.0621, 0.0150], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0094, 0.0119, 0.0074, 0.0080, 0.0092, 0.0107, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 12:08:28,182 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111440.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:08:32,029 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-03-09 12:08:51,737 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.233e+02 1.814e+02 2.141e+02 2.602e+02 6.153e+02, threshold=4.282e+02, percent-clipped=2.0 2023-03-09 12:09:02,531 INFO [train2.py:809] (0/4) Epoch 28, batch 3900, loss[ctc_loss=0.07574, att_loss=0.2424, loss=0.2091, over 16882.00 frames. utt_duration=1380 frames, utt_pad_proportion=0.006612, over 49.00 utterances.], tot_loss[ctc_loss=0.06502, att_loss=0.2327, loss=0.1992, over 3278031.31 frames. utt_duration=1213 frames, utt_pad_proportion=0.06215, over 10824.62 utterances.], batch size: 49, lr: 3.80e-03, grad_scale: 16.0 2023-03-09 12:09:06,508 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.49 vs. limit=2.0 2023-03-09 12:09:12,020 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111468.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:10:19,197 INFO [train2.py:809] (0/4) Epoch 28, batch 3950, loss[ctc_loss=0.0573, att_loss=0.2164, loss=0.1845, over 15510.00 frames. utt_duration=1725 frames, utt_pad_proportion=0.007528, over 36.00 utterances.], tot_loss[ctc_loss=0.06529, att_loss=0.2323, loss=0.1989, over 3267792.03 frames. utt_duration=1186 frames, utt_pad_proportion=0.07112, over 11031.79 utterances.], batch size: 36, lr: 3.79e-03, grad_scale: 16.0 2023-03-09 12:10:37,688 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111524.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:10:45,258 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111529.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:10:59,762 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7489, 3.4172, 3.9230, 4.7326, 4.2760, 4.1750, 3.3016, 3.0987], device='cuda:0'), covar=tensor([0.0719, 0.1662, 0.0654, 0.0507, 0.0763, 0.0465, 0.1369, 0.1725], device='cuda:0'), in_proj_covar=tensor([0.0192, 0.0221, 0.0187, 0.0231, 0.0238, 0.0195, 0.0208, 0.0194], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:11:08,026 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-28.pt 2023-03-09 12:11:24,274 INFO [train2.py:809] (0/4) Epoch 29, batch 0, loss[ctc_loss=0.07626, att_loss=0.2433, loss=0.2099, over 17539.00 frames. utt_duration=1018 frames, utt_pad_proportion=0.04043, over 69.00 utterances.], tot_loss[ctc_loss=0.07626, att_loss=0.2433, loss=0.2099, over 17539.00 frames. utt_duration=1018 frames, utt_pad_proportion=0.04043, over 69.00 utterances.], batch size: 69, lr: 3.73e-03, grad_scale: 8.0 2023-03-09 12:11:24,276 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 12:11:36,590 INFO [train2.py:843] (0/4) Epoch 29, validation: ctc_loss=0.04125, att_loss=0.2346, loss=0.1959, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 12:11:36,591 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16119MB 2023-03-09 12:11:54,772 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 1.789e+02 2.164e+02 2.545e+02 5.866e+02, threshold=4.328e+02, percent-clipped=2.0 2023-03-09 12:12:19,802 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111572.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:12:34,354 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111581.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:12:44,464 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.27 vs. limit=2.0 2023-03-09 12:12:55,745 INFO [train2.py:809] (0/4) Epoch 29, batch 50, loss[ctc_loss=0.05488, att_loss=0.2196, loss=0.1866, over 16161.00 frames. utt_duration=1578 frames, utt_pad_proportion=0.008079, over 41.00 utterances.], tot_loss[ctc_loss=0.0636, att_loss=0.2296, loss=0.1964, over 737437.65 frames. utt_duration=1368 frames, utt_pad_proportion=0.0233, over 2158.67 utterances.], batch size: 41, lr: 3.73e-03, grad_scale: 8.0 2023-03-09 12:13:26,132 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.61 vs. limit=5.0 2023-03-09 12:13:26,784 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111614.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:14:02,164 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0712, 4.3260, 4.4776, 4.6175, 2.5445, 4.4240, 2.8834, 1.6892], device='cuda:0'), covar=tensor([0.0545, 0.0344, 0.0654, 0.0242, 0.1717, 0.0250, 0.1353, 0.1746], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0192, 0.0270, 0.0183, 0.0227, 0.0173, 0.0234, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:14:15,536 INFO [train2.py:809] (0/4) Epoch 29, batch 100, loss[ctc_loss=0.06362, att_loss=0.2361, loss=0.2016, over 16482.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.006459, over 46.00 utterances.], tot_loss[ctc_loss=0.06311, att_loss=0.23, loss=0.1966, over 1301299.43 frames. utt_duration=1255 frames, utt_pad_proportion=0.05153, over 4151.47 utterances.], batch size: 46, lr: 3.72e-03, grad_scale: 8.0 2023-03-09 12:14:33,659 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.291e+02 1.812e+02 2.121e+02 2.575e+02 9.495e+02, threshold=4.242e+02, percent-clipped=2.0 2023-03-09 12:14:40,230 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4629, 2.7248, 4.9373, 3.8768, 3.1080, 4.1777, 4.6666, 4.6464], device='cuda:0'), covar=tensor([0.0269, 0.1502, 0.0176, 0.0906, 0.1652, 0.0296, 0.0183, 0.0248], device='cuda:0'), in_proj_covar=tensor([0.0236, 0.0249, 0.0229, 0.0328, 0.0272, 0.0244, 0.0221, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:14:42,894 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111662.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:15:35,264 INFO [train2.py:809] (0/4) Epoch 29, batch 150, loss[ctc_loss=0.07492, att_loss=0.2521, loss=0.2166, over 17333.00 frames. utt_duration=1006 frames, utt_pad_proportion=0.05156, over 69.00 utterances.], tot_loss[ctc_loss=0.06515, att_loss=0.2308, loss=0.1977, over 1728968.72 frames. utt_duration=1208 frames, utt_pad_proportion=0.06586, over 5734.40 utterances.], batch size: 69, lr: 3.72e-03, grad_scale: 8.0 2023-03-09 12:16:20,847 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0474, 4.9843, 4.7481, 2.8907, 4.8352, 4.7859, 4.3564, 2.8439], device='cuda:0'), covar=tensor([0.0115, 0.0120, 0.0326, 0.1060, 0.0108, 0.0204, 0.0314, 0.1364], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0107, 0.0113, 0.0113, 0.0090, 0.0119, 0.0101, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:16:25,539 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9895, 4.0298, 4.0368, 4.0879, 4.1391, 4.1502, 3.8637, 3.8146], device='cuda:0'), covar=tensor([0.1032, 0.0704, 0.0938, 0.0524, 0.0383, 0.0405, 0.0559, 0.0395], device='cuda:0'), in_proj_covar=tensor([0.0539, 0.0379, 0.0377, 0.0380, 0.0446, 0.0448, 0.0380, 0.0417], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:16:57,986 INFO [train2.py:809] (0/4) Epoch 29, batch 200, loss[ctc_loss=0.06467, att_loss=0.2393, loss=0.2044, over 17127.00 frames. utt_duration=1225 frames, utt_pad_proportion=0.01451, over 56.00 utterances.], tot_loss[ctc_loss=0.06497, att_loss=0.2311, loss=0.1978, over 2065467.70 frames. utt_duration=1194 frames, utt_pad_proportion=0.07048, over 6930.84 utterances.], batch size: 56, lr: 3.72e-03, grad_scale: 8.0 2023-03-09 12:17:15,898 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 1.846e+02 2.300e+02 2.696e+02 5.049e+02, threshold=4.600e+02, percent-clipped=1.0 2023-03-09 12:17:45,422 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0282, 5.3871, 5.5644, 5.3738, 5.5318, 6.0339, 5.2903, 6.1176], device='cuda:0'), covar=tensor([0.0812, 0.0763, 0.0867, 0.1340, 0.2029, 0.0921, 0.0701, 0.0707], device='cuda:0'), in_proj_covar=tensor([0.0927, 0.0538, 0.0656, 0.0694, 0.0926, 0.0672, 0.0516, 0.0656], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 12:18:01,226 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0759, 5.0668, 4.8258, 3.2239, 4.9112, 4.7724, 4.4136, 2.9372], device='cuda:0'), covar=tensor([0.0122, 0.0100, 0.0307, 0.0825, 0.0096, 0.0184, 0.0278, 0.1183], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0107, 0.0112, 0.0112, 0.0090, 0.0118, 0.0101, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:18:12,017 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=111791.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 12:18:17,854 INFO [train2.py:809] (0/4) Epoch 29, batch 250, loss[ctc_loss=0.08696, att_loss=0.2551, loss=0.2215, over 17316.00 frames. utt_duration=1175 frames, utt_pad_proportion=0.02217, over 59.00 utterances.], tot_loss[ctc_loss=0.06398, att_loss=0.2303, loss=0.197, over 2335082.71 frames. utt_duration=1221 frames, utt_pad_proportion=0.06168, over 7661.64 utterances.], batch size: 59, lr: 3.72e-03, grad_scale: 8.0 2023-03-09 12:19:05,333 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=111824.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:19:37,433 INFO [train2.py:809] (0/4) Epoch 29, batch 300, loss[ctc_loss=0.06084, att_loss=0.2395, loss=0.2038, over 17322.00 frames. utt_duration=1261 frames, utt_pad_proportion=0.009976, over 55.00 utterances.], tot_loss[ctc_loss=0.06416, att_loss=0.2307, loss=0.1974, over 2541785.67 frames. utt_duration=1235 frames, utt_pad_proportion=0.05714, over 8241.50 utterances.], batch size: 55, lr: 3.72e-03, grad_scale: 8.0 2023-03-09 12:19:49,896 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=111852.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 12:19:55,514 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.235e+02 1.741e+02 2.135e+02 2.604e+02 1.132e+03, threshold=4.270e+02, percent-clipped=2.0 2023-03-09 12:20:35,402 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=111881.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:20:56,610 INFO [train2.py:809] (0/4) Epoch 29, batch 350, loss[ctc_loss=0.06238, att_loss=0.236, loss=0.2013, over 16859.00 frames. utt_duration=682.7 frames, utt_pad_proportion=0.1445, over 99.00 utterances.], tot_loss[ctc_loss=0.06446, att_loss=0.2307, loss=0.1974, over 2706251.67 frames. utt_duration=1243 frames, utt_pad_proportion=0.05417, over 8717.47 utterances.], batch size: 99, lr: 3.72e-03, grad_scale: 8.0 2023-03-09 12:21:50,815 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=111929.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:22:16,048 INFO [train2.py:809] (0/4) Epoch 29, batch 400, loss[ctc_loss=0.05553, att_loss=0.239, loss=0.2023, over 16616.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005962, over 47.00 utterances.], tot_loss[ctc_loss=0.06466, att_loss=0.2313, loss=0.1979, over 2840192.67 frames. utt_duration=1254 frames, utt_pad_proportion=0.04992, over 9068.84 utterances.], batch size: 47, lr: 3.72e-03, grad_scale: 8.0 2023-03-09 12:22:17,634 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9961, 6.2320, 5.6300, 5.9258, 5.8771, 5.4187, 5.6641, 5.3138], device='cuda:0'), covar=tensor([0.1106, 0.0808, 0.0933, 0.0843, 0.0823, 0.1547, 0.2015, 0.2141], device='cuda:0'), in_proj_covar=tensor([0.0563, 0.0639, 0.0490, 0.0475, 0.0454, 0.0483, 0.0641, 0.0547], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:22:31,884 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8690, 4.7809, 4.5761, 2.7842, 4.6530, 4.4799, 4.1171, 2.5944], device='cuda:0'), covar=tensor([0.0107, 0.0115, 0.0284, 0.1032, 0.0102, 0.0262, 0.0327, 0.1410], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0106, 0.0112, 0.0112, 0.0089, 0.0118, 0.0100, 0.0104], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:22:34,679 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.149e+02 1.870e+02 2.171e+02 2.633e+02 6.586e+02, threshold=4.343e+02, percent-clipped=4.0 2023-03-09 12:23:38,445 INFO [train2.py:809] (0/4) Epoch 29, batch 450, loss[ctc_loss=0.07118, att_loss=0.2485, loss=0.213, over 17056.00 frames. utt_duration=1314 frames, utt_pad_proportion=0.008549, over 52.00 utterances.], tot_loss[ctc_loss=0.0653, att_loss=0.2318, loss=0.1985, over 2938488.27 frames. utt_duration=1230 frames, utt_pad_proportion=0.05732, over 9567.15 utterances.], batch size: 52, lr: 3.72e-03, grad_scale: 4.0 2023-03-09 12:23:43,548 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8636, 2.6043, 2.6966, 2.6713, 2.9930, 2.8973, 2.6156, 3.0514], device='cuda:0'), covar=tensor([0.1651, 0.2101, 0.1761, 0.1213, 0.1350, 0.0890, 0.1782, 0.1289], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0149, 0.0146, 0.0140, 0.0158, 0.0135, 0.0158, 0.0135], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 12:23:47,400 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-112000.pt 2023-03-09 12:25:02,537 INFO [train2.py:809] (0/4) Epoch 29, batch 500, loss[ctc_loss=0.07703, att_loss=0.2481, loss=0.2138, over 17298.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02301, over 59.00 utterances.], tot_loss[ctc_loss=0.06515, att_loss=0.2315, loss=0.1982, over 3007613.41 frames. utt_duration=1226 frames, utt_pad_proportion=0.05966, over 9828.66 utterances.], batch size: 59, lr: 3.72e-03, grad_scale: 4.0 2023-03-09 12:25:22,139 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.395e+02 1.941e+02 2.244e+02 2.736e+02 5.323e+02, threshold=4.488e+02, percent-clipped=4.0 2023-03-09 12:25:37,091 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112066.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:26:06,151 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-03-09 12:26:22,018 INFO [train2.py:809] (0/4) Epoch 29, batch 550, loss[ctc_loss=0.05821, att_loss=0.2288, loss=0.1947, over 16292.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006534, over 43.00 utterances.], tot_loss[ctc_loss=0.06531, att_loss=0.2312, loss=0.198, over 3063045.49 frames. utt_duration=1224 frames, utt_pad_proportion=0.0611, over 10021.56 utterances.], batch size: 43, lr: 3.72e-03, grad_scale: 4.0 2023-03-09 12:27:10,026 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112124.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:27:14,743 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112127.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:27:42,789 INFO [train2.py:809] (0/4) Epoch 29, batch 600, loss[ctc_loss=0.06348, att_loss=0.2332, loss=0.1993, over 17044.00 frames. utt_duration=1338 frames, utt_pad_proportion=0.006492, over 51.00 utterances.], tot_loss[ctc_loss=0.06565, att_loss=0.2312, loss=0.1981, over 3109855.20 frames. utt_duration=1211 frames, utt_pad_proportion=0.0638, over 10286.48 utterances.], batch size: 51, lr: 3.72e-03, grad_scale: 4.0 2023-03-09 12:27:45,914 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112147.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 12:28:01,253 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.227e+02 2.016e+02 2.437e+02 2.973e+02 9.680e+02, threshold=4.873e+02, percent-clipped=6.0 2023-03-09 12:28:26,118 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112172.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:29:03,067 INFO [train2.py:809] (0/4) Epoch 29, batch 650, loss[ctc_loss=0.04455, att_loss=0.1992, loss=0.1683, over 14502.00 frames. utt_duration=1814 frames, utt_pad_proportion=0.03088, over 32.00 utterances.], tot_loss[ctc_loss=0.06535, att_loss=0.2311, loss=0.198, over 3142873.91 frames. utt_duration=1207 frames, utt_pad_proportion=0.066, over 10431.35 utterances.], batch size: 32, lr: 3.72e-03, grad_scale: 4.0 2023-03-09 12:30:18,648 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.93 vs. limit=5.0 2023-03-09 12:30:22,502 INFO [train2.py:809] (0/4) Epoch 29, batch 700, loss[ctc_loss=0.07299, att_loss=0.2258, loss=0.1952, over 16021.00 frames. utt_duration=1604 frames, utt_pad_proportion=0.006506, over 40.00 utterances.], tot_loss[ctc_loss=0.06439, att_loss=0.2299, loss=0.1968, over 3168619.61 frames. utt_duration=1254 frames, utt_pad_proportion=0.05579, over 10121.09 utterances.], batch size: 40, lr: 3.71e-03, grad_scale: 4.0 2023-03-09 12:30:41,002 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.343e+02 1.769e+02 2.156e+02 2.659e+02 7.287e+02, threshold=4.312e+02, percent-clipped=5.0 2023-03-09 12:31:34,947 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.81 vs. limit=2.0 2023-03-09 12:31:41,968 INFO [train2.py:809] (0/4) Epoch 29, batch 750, loss[ctc_loss=0.0853, att_loss=0.2588, loss=0.2241, over 17330.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02288, over 59.00 utterances.], tot_loss[ctc_loss=0.06481, att_loss=0.2309, loss=0.1977, over 3196837.34 frames. utt_duration=1229 frames, utt_pad_proportion=0.05828, over 10416.95 utterances.], batch size: 59, lr: 3.71e-03, grad_scale: 4.0 2023-03-09 12:32:23,409 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.74 vs. limit=5.0 2023-03-09 12:33:02,103 INFO [train2.py:809] (0/4) Epoch 29, batch 800, loss[ctc_loss=0.07981, att_loss=0.2482, loss=0.2145, over 17291.00 frames. utt_duration=877.1 frames, utt_pad_proportion=0.07671, over 79.00 utterances.], tot_loss[ctc_loss=0.06544, att_loss=0.2315, loss=0.1983, over 3215905.26 frames. utt_duration=1216 frames, utt_pad_proportion=0.06138, over 10593.07 utterances.], batch size: 79, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:33:21,114 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.436e+02 1.905e+02 2.158e+02 2.747e+02 4.535e+02, threshold=4.315e+02, percent-clipped=2.0 2023-03-09 12:34:13,341 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112389.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:34:22,115 INFO [train2.py:809] (0/4) Epoch 29, batch 850, loss[ctc_loss=0.0736, att_loss=0.2395, loss=0.2063, over 16472.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.007031, over 46.00 utterances.], tot_loss[ctc_loss=0.06503, att_loss=0.2318, loss=0.1985, over 3239937.75 frames. utt_duration=1239 frames, utt_pad_proportion=0.05302, over 10476.71 utterances.], batch size: 46, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:35:06,134 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112422.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:35:29,380 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.6993, 2.5013, 2.5873, 2.5531, 3.0140, 2.9054, 2.4636, 3.2119], device='cuda:0'), covar=tensor([0.1329, 0.2168, 0.1416, 0.1202, 0.1250, 0.0891, 0.1678, 0.0826], device='cuda:0'), in_proj_covar=tensor([0.0150, 0.0149, 0.0147, 0.0140, 0.0158, 0.0136, 0.0158, 0.0136], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 12:35:42,813 INFO [train2.py:809] (0/4) Epoch 29, batch 900, loss[ctc_loss=0.05048, att_loss=0.218, loss=0.1845, over 16177.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.007251, over 41.00 utterances.], tot_loss[ctc_loss=0.06467, att_loss=0.2319, loss=0.1984, over 3253960.50 frames. utt_duration=1248 frames, utt_pad_proportion=0.04998, over 10438.21 utterances.], batch size: 41, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:35:46,168 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112447.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 12:35:51,238 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112450.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:36:02,466 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.247e+02 1.980e+02 2.255e+02 2.971e+02 7.744e+02, threshold=4.510e+02, percent-clipped=6.0 2023-03-09 12:36:16,732 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.85 vs. limit=2.0 2023-03-09 12:36:34,579 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8277, 3.4612, 3.4144, 3.0118, 3.4399, 3.4866, 3.5455, 2.4065], device='cuda:0'), covar=tensor([0.1104, 0.1462, 0.1828, 0.2771, 0.1477, 0.1859, 0.0816, 0.3270], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0210, 0.0226, 0.0276, 0.0187, 0.0287, 0.0210, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:37:03,513 INFO [train2.py:809] (0/4) Epoch 29, batch 950, loss[ctc_loss=0.07306, att_loss=0.2331, loss=0.2011, over 16955.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.008079, over 50.00 utterances.], tot_loss[ctc_loss=0.06407, att_loss=0.2316, loss=0.1981, over 3262094.48 frames. utt_duration=1257 frames, utt_pad_proportion=0.04804, over 10393.84 utterances.], batch size: 50, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:37:03,588 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112495.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 12:38:17,828 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 12:38:19,294 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2302, 2.9174, 3.0590, 4.3416, 3.8709, 3.8252, 2.8760, 2.2718], device='cuda:0'), covar=tensor([0.0885, 0.1887, 0.1021, 0.0581, 0.0920, 0.0535, 0.1606, 0.2193], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0219, 0.0186, 0.0228, 0.0235, 0.0193, 0.0206, 0.0193], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:38:23,727 INFO [train2.py:809] (0/4) Epoch 29, batch 1000, loss[ctc_loss=0.05349, att_loss=0.2398, loss=0.2025, over 16629.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.005259, over 47.00 utterances.], tot_loss[ctc_loss=0.06344, att_loss=0.2311, loss=0.1976, over 3260318.96 frames. utt_duration=1271 frames, utt_pad_proportion=0.0446, over 10269.68 utterances.], batch size: 47, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:38:42,358 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.260e+02 1.873e+02 2.288e+02 2.757e+02 5.242e+02, threshold=4.577e+02, percent-clipped=3.0 2023-03-09 12:38:51,132 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112562.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:39:14,903 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112577.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:39:42,798 INFO [train2.py:809] (0/4) Epoch 29, batch 1050, loss[ctc_loss=0.07361, att_loss=0.2255, loss=0.1951, over 15947.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.006813, over 41.00 utterances.], tot_loss[ctc_loss=0.06317, att_loss=0.2311, loss=0.1975, over 3260558.68 frames. utt_duration=1293 frames, utt_pad_proportion=0.03994, over 10102.01 utterances.], batch size: 41, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:40:19,222 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112617.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:40:28,632 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112623.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:40:52,254 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112638.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 12:41:03,005 INFO [train2.py:809] (0/4) Epoch 29, batch 1100, loss[ctc_loss=0.04931, att_loss=0.2078, loss=0.1761, over 15764.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009044, over 38.00 utterances.], tot_loss[ctc_loss=0.06344, att_loss=0.231, loss=0.1975, over 3266046.62 frames. utt_duration=1285 frames, utt_pad_proportion=0.04141, over 10179.88 utterances.], batch size: 38, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:41:22,207 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.269e+02 1.863e+02 2.140e+02 2.769e+02 5.111e+02, threshold=4.279e+02, percent-clipped=1.0 2023-03-09 12:41:43,264 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1869, 5.0970, 4.9093, 3.2402, 4.8601, 4.7498, 4.4324, 2.9609], device='cuda:0'), covar=tensor([0.0108, 0.0123, 0.0301, 0.0954, 0.0118, 0.0216, 0.0311, 0.1288], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0107, 0.0112, 0.0113, 0.0091, 0.0119, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:41:55,440 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112678.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 12:42:21,112 INFO [train2.py:809] (0/4) Epoch 29, batch 1150, loss[ctc_loss=0.05447, att_loss=0.2286, loss=0.1938, over 16534.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006662, over 45.00 utterances.], tot_loss[ctc_loss=0.06346, att_loss=0.2308, loss=0.1973, over 3271012.16 frames. utt_duration=1299 frames, utt_pad_proportion=0.03899, over 10081.87 utterances.], batch size: 45, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:42:21,393 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8224, 5.1501, 4.6978, 5.1652, 4.5483, 4.8371, 5.2674, 5.0487], device='cuda:0'), covar=tensor([0.0593, 0.0262, 0.0793, 0.0321, 0.0412, 0.0326, 0.0206, 0.0202], device='cuda:0'), in_proj_covar=tensor([0.0408, 0.0343, 0.0383, 0.0386, 0.0341, 0.0247, 0.0324, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 12:43:01,934 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0445, 4.2171, 4.2071, 4.3755, 2.7214, 4.3028, 2.7038, 1.8073], device='cuda:0'), covar=tensor([0.0491, 0.0335, 0.0761, 0.0291, 0.1667, 0.0275, 0.1480, 0.1700], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0192, 0.0269, 0.0183, 0.0226, 0.0174, 0.0234, 0.0207], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:43:04,745 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=112722.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:43:40,480 INFO [train2.py:809] (0/4) Epoch 29, batch 1200, loss[ctc_loss=0.04103, att_loss=0.1968, loss=0.1656, over 15766.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008295, over 38.00 utterances.], tot_loss[ctc_loss=0.06391, att_loss=0.2313, loss=0.1978, over 3276300.52 frames. utt_duration=1293 frames, utt_pad_proportion=0.03964, over 10143.80 utterances.], batch size: 38, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:43:40,703 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112745.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:43:59,908 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.017e+02 1.786e+02 2.116e+02 2.747e+02 6.920e+02, threshold=4.232e+02, percent-clipped=7.0 2023-03-09 12:44:21,240 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=112770.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:45:00,866 INFO [train2.py:809] (0/4) Epoch 29, batch 1250, loss[ctc_loss=0.07279, att_loss=0.2418, loss=0.208, over 16869.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007493, over 49.00 utterances.], tot_loss[ctc_loss=0.06442, att_loss=0.2318, loss=0.1983, over 3272107.20 frames. utt_duration=1276 frames, utt_pad_proportion=0.04586, over 10273.06 utterances.], batch size: 49, lr: 3.71e-03, grad_scale: 8.0 2023-03-09 12:46:20,450 INFO [train2.py:809] (0/4) Epoch 29, batch 1300, loss[ctc_loss=0.07439, att_loss=0.2527, loss=0.217, over 17014.00 frames. utt_duration=1310 frames, utt_pad_proportion=0.00955, over 52.00 utterances.], tot_loss[ctc_loss=0.06408, att_loss=0.2311, loss=0.1977, over 3275680.25 frames. utt_duration=1288 frames, utt_pad_proportion=0.04249, over 10188.05 utterances.], batch size: 52, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:46:40,102 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.148e+02 1.906e+02 2.330e+02 2.870e+02 1.155e+03, threshold=4.661e+02, percent-clipped=5.0 2023-03-09 12:47:40,653 INFO [train2.py:809] (0/4) Epoch 29, batch 1350, loss[ctc_loss=0.05981, att_loss=0.2207, loss=0.1885, over 16133.00 frames. utt_duration=1538 frames, utt_pad_proportion=0.005695, over 42.00 utterances.], tot_loss[ctc_loss=0.06363, att_loss=0.231, loss=0.1975, over 3282620.16 frames. utt_duration=1283 frames, utt_pad_proportion=0.0423, over 10243.71 utterances.], batch size: 42, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:48:17,494 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112918.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:48:22,207 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=112921.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:48:30,035 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1686, 5.5406, 5.0809, 5.5406, 4.9082, 5.1393, 5.6428, 5.4333], device='cuda:0'), covar=tensor([0.0591, 0.0263, 0.0694, 0.0320, 0.0377, 0.0246, 0.0183, 0.0180], device='cuda:0'), in_proj_covar=tensor([0.0409, 0.0343, 0.0383, 0.0386, 0.0341, 0.0247, 0.0324, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 12:48:41,462 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112933.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 12:49:00,127 INFO [train2.py:809] (0/4) Epoch 29, batch 1400, loss[ctc_loss=0.07074, att_loss=0.2308, loss=0.1988, over 15779.00 frames. utt_duration=1662 frames, utt_pad_proportion=0.007463, over 38.00 utterances.], tot_loss[ctc_loss=0.06401, att_loss=0.2309, loss=0.1975, over 3280396.02 frames. utt_duration=1268 frames, utt_pad_proportion=0.04687, over 10360.48 utterances.], batch size: 38, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:49:19,015 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.160e+02 1.843e+02 2.238e+02 2.535e+02 6.230e+02, threshold=4.475e+02, percent-clipped=2.0 2023-03-09 12:49:29,251 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.5701, 2.8852, 3.6404, 4.5853, 4.0635, 4.1300, 3.0750, 2.5355], device='cuda:0'), covar=tensor([0.0661, 0.1919, 0.0801, 0.0528, 0.0872, 0.0467, 0.1496, 0.2037], device='cuda:0'), in_proj_covar=tensor([0.0189, 0.0220, 0.0185, 0.0227, 0.0235, 0.0192, 0.0205, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:49:41,493 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.2014, 5.4860, 5.6958, 5.5117, 5.7306, 6.1520, 5.3536, 6.2139], device='cuda:0'), covar=tensor([0.0615, 0.0671, 0.0766, 0.1303, 0.1542, 0.0778, 0.0645, 0.0615], device='cuda:0'), in_proj_covar=tensor([0.0912, 0.0528, 0.0638, 0.0680, 0.0905, 0.0663, 0.0509, 0.0636], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 12:49:44,575 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=112973.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 12:49:59,355 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=112982.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:50:19,500 INFO [train2.py:809] (0/4) Epoch 29, batch 1450, loss[ctc_loss=0.04773, att_loss=0.2386, loss=0.2004, over 16948.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.008605, over 50.00 utterances.], tot_loss[ctc_loss=0.06369, att_loss=0.2306, loss=0.1972, over 3285250.13 frames. utt_duration=1281 frames, utt_pad_proportion=0.0422, over 10268.67 utterances.], batch size: 50, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:51:17,167 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3451, 2.5809, 3.5656, 2.7342, 3.4481, 4.5416, 4.4795, 3.0127], device='cuda:0'), covar=tensor([0.0428, 0.2083, 0.1133, 0.1563, 0.1032, 0.0657, 0.0462, 0.1544], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0253, 0.0297, 0.0222, 0.0277, 0.0387, 0.0276, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 12:51:39,658 INFO [train2.py:809] (0/4) Epoch 29, batch 1500, loss[ctc_loss=0.05019, att_loss=0.2178, loss=0.1843, over 15962.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006824, over 41.00 utterances.], tot_loss[ctc_loss=0.0636, att_loss=0.2308, loss=0.1973, over 3284414.20 frames. utt_duration=1272 frames, utt_pad_proportion=0.04527, over 10343.02 utterances.], batch size: 41, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:51:39,920 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113045.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:51:58,818 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.135e+02 1.764e+02 2.089e+02 2.547e+02 5.820e+02, threshold=4.178e+02, percent-clipped=2.0 2023-03-09 12:52:55,206 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113093.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:52:58,847 INFO [train2.py:809] (0/4) Epoch 29, batch 1550, loss[ctc_loss=0.07515, att_loss=0.2449, loss=0.2109, over 17275.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01282, over 55.00 utterances.], tot_loss[ctc_loss=0.06453, att_loss=0.2316, loss=0.1982, over 3287490.22 frames. utt_duration=1265 frames, utt_pad_proportion=0.04686, over 10404.09 utterances.], batch size: 55, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:53:16,020 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-03-09 12:53:30,797 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0463, 4.2866, 4.3749, 4.4780, 2.4452, 4.4620, 2.7757, 1.7747], device='cuda:0'), covar=tensor([0.0486, 0.0313, 0.0646, 0.0253, 0.1763, 0.0253, 0.1430, 0.1665], device='cuda:0'), in_proj_covar=tensor([0.0222, 0.0191, 0.0267, 0.0182, 0.0225, 0.0172, 0.0233, 0.0204], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:53:44,286 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.89 vs. limit=5.0 2023-03-09 12:54:19,352 INFO [train2.py:809] (0/4) Epoch 29, batch 1600, loss[ctc_loss=0.05804, att_loss=0.2185, loss=0.1864, over 16268.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.006756, over 43.00 utterances.], tot_loss[ctc_loss=0.06354, att_loss=0.2307, loss=0.1973, over 3288182.93 frames. utt_duration=1271 frames, utt_pad_proportion=0.04533, over 10358.72 utterances.], batch size: 43, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:54:37,605 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0910, 5.1253, 4.8269, 2.8794, 4.8157, 4.7252, 4.3712, 2.7020], device='cuda:0'), covar=tensor([0.0110, 0.0109, 0.0328, 0.1167, 0.0133, 0.0239, 0.0340, 0.1519], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0107, 0.0113, 0.0113, 0.0091, 0.0120, 0.0102, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 12:54:38,746 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.261e+02 1.848e+02 2.168e+02 2.655e+02 4.318e+02, threshold=4.336e+02, percent-clipped=2.0 2023-03-09 12:55:22,770 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5809, 5.0248, 4.8634, 4.9669, 5.0658, 4.6458, 3.6789, 5.0067], device='cuda:0'), covar=tensor([0.0157, 0.0113, 0.0160, 0.0084, 0.0109, 0.0133, 0.0642, 0.0237], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0093, 0.0118, 0.0074, 0.0081, 0.0091, 0.0107, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 12:55:38,881 INFO [train2.py:809] (0/4) Epoch 29, batch 1650, loss[ctc_loss=0.06035, att_loss=0.2349, loss=0.2, over 16127.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006264, over 42.00 utterances.], tot_loss[ctc_loss=0.06352, att_loss=0.2303, loss=0.197, over 3281115.41 frames. utt_duration=1272 frames, utt_pad_proportion=0.04775, over 10332.55 utterances.], batch size: 42, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:55:40,850 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4656, 4.5127, 4.6549, 4.5978, 5.1991, 4.3892, 4.5528, 2.6493], device='cuda:0'), covar=tensor([0.0324, 0.0426, 0.0343, 0.0347, 0.0731, 0.0308, 0.0345, 0.1667], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0230, 0.0226, 0.0241, 0.0389, 0.0198, 0.0217, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:55:53,048 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.23 vs. limit=2.0 2023-03-09 12:56:06,432 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-03-09 12:56:14,939 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113218.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:56:38,804 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113233.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:56:40,388 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113234.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:56:58,263 INFO [train2.py:809] (0/4) Epoch 29, batch 1700, loss[ctc_loss=0.0437, att_loss=0.2059, loss=0.1735, over 15868.00 frames. utt_duration=1629 frames, utt_pad_proportion=0.01045, over 39.00 utterances.], tot_loss[ctc_loss=0.06275, att_loss=0.2298, loss=0.1964, over 3272149.73 frames. utt_duration=1268 frames, utt_pad_proportion=0.05055, over 10334.12 utterances.], batch size: 39, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:57:17,570 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.337e+02 1.796e+02 2.106e+02 2.556e+02 4.663e+02, threshold=4.212e+02, percent-clipped=1.0 2023-03-09 12:57:18,094 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3112, 4.3903, 4.5180, 4.4791, 4.9942, 4.3502, 4.3699, 2.4881], device='cuda:0'), covar=tensor([0.0372, 0.0433, 0.0397, 0.0344, 0.0752, 0.0298, 0.0434, 0.1830], device='cuda:0'), in_proj_covar=tensor([0.0201, 0.0228, 0.0225, 0.0239, 0.0387, 0.0197, 0.0216, 0.0223], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 12:57:31,616 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113266.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:57:43,275 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113273.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 12:57:46,848 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.68 vs. limit=5.0 2023-03-09 12:57:49,280 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113277.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:57:55,492 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113281.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:58:18,081 INFO [train2.py:809] (0/4) Epoch 29, batch 1750, loss[ctc_loss=0.06718, att_loss=0.2403, loss=0.2057, over 17279.00 frames. utt_duration=1258 frames, utt_pad_proportion=0.01244, over 55.00 utterances.], tot_loss[ctc_loss=0.06348, att_loss=0.2307, loss=0.1972, over 3266124.99 frames. utt_duration=1231 frames, utt_pad_proportion=0.06068, over 10623.96 utterances.], batch size: 55, lr: 3.70e-03, grad_scale: 8.0 2023-03-09 12:58:18,452 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113295.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:58:37,245 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2360, 5.5463, 5.1348, 5.5772, 4.9510, 5.1408, 5.6508, 5.4459], device='cuda:0'), covar=tensor([0.0479, 0.0293, 0.0650, 0.0312, 0.0360, 0.0223, 0.0201, 0.0176], device='cuda:0'), in_proj_covar=tensor([0.0407, 0.0343, 0.0383, 0.0386, 0.0341, 0.0248, 0.0325, 0.0306], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 12:58:59,047 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113321.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 12:59:06,406 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.55 vs. limit=5.0 2023-03-09 12:59:37,376 INFO [train2.py:809] (0/4) Epoch 29, batch 1800, loss[ctc_loss=0.05664, att_loss=0.2306, loss=0.1958, over 16344.00 frames. utt_duration=1454 frames, utt_pad_proportion=0.005335, over 45.00 utterances.], tot_loss[ctc_loss=0.06269, att_loss=0.2302, loss=0.1967, over 3270587.23 frames. utt_duration=1259 frames, utt_pad_proportion=0.05289, over 10404.76 utterances.], batch size: 45, lr: 3.70e-03, grad_scale: 4.0 2023-03-09 12:59:39,118 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.47 vs. limit=2.0 2023-03-09 12:59:57,720 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.115e+02 1.881e+02 2.263e+02 2.651e+02 5.025e+02, threshold=4.526e+02, percent-clipped=2.0 2023-03-09 13:00:07,796 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-03-09 13:00:18,115 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2241, 5.6258, 4.7530, 5.7216, 5.0393, 5.3316, 5.6166, 5.4843], device='cuda:0'), covar=tensor([0.0588, 0.0345, 0.1173, 0.0353, 0.0373, 0.0246, 0.0304, 0.0236], device='cuda:0'), in_proj_covar=tensor([0.0406, 0.0342, 0.0383, 0.0385, 0.0340, 0.0247, 0.0324, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 13:00:37,690 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8954, 3.5769, 3.5561, 3.0518, 3.5815, 3.6931, 3.6264, 2.5193], device='cuda:0'), covar=tensor([0.0976, 0.1051, 0.1723, 0.2905, 0.0945, 0.2138, 0.0755, 0.3096], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0210, 0.0227, 0.0277, 0.0186, 0.0288, 0.0209, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:00:55,299 INFO [train2.py:809] (0/4) Epoch 29, batch 1850, loss[ctc_loss=0.06229, att_loss=0.2133, loss=0.1831, over 14512.00 frames. utt_duration=1815 frames, utt_pad_proportion=0.04655, over 32.00 utterances.], tot_loss[ctc_loss=0.06321, att_loss=0.2298, loss=0.1965, over 3262209.64 frames. utt_duration=1263 frames, utt_pad_proportion=0.05412, over 10347.22 utterances.], batch size: 32, lr: 3.70e-03, grad_scale: 4.0 2023-03-09 13:02:00,029 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8305, 5.0870, 4.7318, 5.1263, 4.5374, 4.7812, 5.2139, 4.9963], device='cuda:0'), covar=tensor([0.0528, 0.0290, 0.0697, 0.0337, 0.0407, 0.0374, 0.0209, 0.0197], device='cuda:0'), in_proj_covar=tensor([0.0404, 0.0341, 0.0381, 0.0383, 0.0339, 0.0245, 0.0323, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 13:02:15,433 INFO [train2.py:809] (0/4) Epoch 29, batch 1900, loss[ctc_loss=0.06887, att_loss=0.2373, loss=0.2036, over 16531.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006231, over 45.00 utterances.], tot_loss[ctc_loss=0.06339, att_loss=0.2306, loss=0.1971, over 3265130.03 frames. utt_duration=1226 frames, utt_pad_proportion=0.06195, over 10668.32 utterances.], batch size: 45, lr: 3.70e-03, grad_scale: 4.0 2023-03-09 13:02:35,441 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 1.919e+02 2.347e+02 2.814e+02 3.658e+03, threshold=4.695e+02, percent-clipped=4.0 2023-03-09 13:03:34,130 INFO [train2.py:809] (0/4) Epoch 29, batch 1950, loss[ctc_loss=0.05606, att_loss=0.2005, loss=0.1716, over 15661.00 frames. utt_duration=1695 frames, utt_pad_proportion=0.007236, over 37.00 utterances.], tot_loss[ctc_loss=0.06332, att_loss=0.2304, loss=0.197, over 3266771.34 frames. utt_duration=1250 frames, utt_pad_proportion=0.05589, over 10463.35 utterances.], batch size: 37, lr: 3.69e-03, grad_scale: 4.0 2023-03-09 13:04:41,092 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1919, 5.5200, 5.1241, 5.5624, 4.9370, 5.1854, 5.6540, 5.4131], device='cuda:0'), covar=tensor([0.0574, 0.0281, 0.0697, 0.0321, 0.0399, 0.0212, 0.0190, 0.0186], device='cuda:0'), in_proj_covar=tensor([0.0403, 0.0340, 0.0379, 0.0382, 0.0338, 0.0244, 0.0322, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 13:04:45,101 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.27 vs. limit=5.0 2023-03-09 13:04:54,505 INFO [train2.py:809] (0/4) Epoch 29, batch 2000, loss[ctc_loss=0.06026, att_loss=0.2435, loss=0.2069, over 17434.00 frames. utt_duration=1012 frames, utt_pad_proportion=0.04511, over 69.00 utterances.], tot_loss[ctc_loss=0.06299, att_loss=0.2303, loss=0.1968, over 3266558.96 frames. utt_duration=1260 frames, utt_pad_proportion=0.05318, over 10385.68 utterances.], batch size: 69, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:05:15,508 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.032e+02 1.751e+02 2.072e+02 2.530e+02 5.087e+02, threshold=4.144e+02, percent-clipped=1.0 2023-03-09 13:05:45,770 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113577.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:06:05,666 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113590.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:06:05,803 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4058, 2.8600, 3.5998, 2.9531, 3.4876, 4.5185, 4.3700, 3.2721], device='cuda:0'), covar=tensor([0.0441, 0.2000, 0.1334, 0.1437, 0.1187, 0.1047, 0.0634, 0.1242], device='cuda:0'), in_proj_covar=tensor([0.0255, 0.0254, 0.0296, 0.0223, 0.0277, 0.0388, 0.0276, 0.0237], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 13:06:10,757 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.54 vs. limit=5.0 2023-03-09 13:06:14,830 INFO [train2.py:809] (0/4) Epoch 29, batch 2050, loss[ctc_loss=0.06309, att_loss=0.2333, loss=0.1992, over 17379.00 frames. utt_duration=1009 frames, utt_pad_proportion=0.04908, over 69.00 utterances.], tot_loss[ctc_loss=0.06316, att_loss=0.2301, loss=0.1967, over 3264051.52 frames. utt_duration=1229 frames, utt_pad_proportion=0.06051, over 10636.60 utterances.], batch size: 69, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:06:49,200 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113617.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:07:02,054 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113625.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:07:23,407 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1458, 3.7572, 3.1309, 3.3611, 3.9606, 3.6405, 3.0593, 4.1062], device='cuda:0'), covar=tensor([0.0938, 0.0516, 0.1093, 0.0784, 0.0835, 0.0707, 0.0863, 0.0507], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0232, 0.0234, 0.0212, 0.0296, 0.0254, 0.0209, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 13:07:35,389 INFO [train2.py:809] (0/4) Epoch 29, batch 2100, loss[ctc_loss=0.06388, att_loss=0.227, loss=0.1944, over 16537.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.005811, over 45.00 utterances.], tot_loss[ctc_loss=0.06388, att_loss=0.2311, loss=0.1977, over 3273959.24 frames. utt_duration=1220 frames, utt_pad_proportion=0.06044, over 10742.99 utterances.], batch size: 45, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:07:37,255 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=113646.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:07:55,039 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.156e+02 1.850e+02 2.211e+02 2.646e+02 5.386e+02, threshold=4.422e+02, percent-clipped=4.0 2023-03-09 13:08:26,909 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113678.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:08:55,293 INFO [train2.py:809] (0/4) Epoch 29, batch 2150, loss[ctc_loss=0.04094, att_loss=0.2109, loss=0.1769, over 14591.00 frames. utt_duration=1825 frames, utt_pad_proportion=0.04334, over 32.00 utterances.], tot_loss[ctc_loss=0.06422, att_loss=0.2307, loss=0.1974, over 3270809.33 frames. utt_duration=1222 frames, utt_pad_proportion=0.06096, over 10719.19 utterances.], batch size: 32, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:09:13,874 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=113707.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:10:13,175 INFO [train2.py:809] (0/4) Epoch 29, batch 2200, loss[ctc_loss=0.06037, att_loss=0.2369, loss=0.2016, over 17325.00 frames. utt_duration=1262 frames, utt_pad_proportion=0.00969, over 55.00 utterances.], tot_loss[ctc_loss=0.06516, att_loss=0.2315, loss=0.1982, over 3267363.82 frames. utt_duration=1194 frames, utt_pad_proportion=0.06793, over 10963.60 utterances.], batch size: 55, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:10:30,453 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.98 vs. limit=5.0 2023-03-09 13:10:32,604 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.987e+02 2.347e+02 2.821e+02 4.899e+02, threshold=4.694e+02, percent-clipped=4.0 2023-03-09 13:11:32,256 INFO [train2.py:809] (0/4) Epoch 29, batch 2250, loss[ctc_loss=0.06636, att_loss=0.2264, loss=0.1944, over 15937.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.008208, over 41.00 utterances.], tot_loss[ctc_loss=0.06456, att_loss=0.2313, loss=0.198, over 3270740.36 frames. utt_duration=1205 frames, utt_pad_proportion=0.06399, over 10866.24 utterances.], batch size: 41, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:12:01,335 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.77 vs. limit=2.0 2023-03-09 13:12:18,342 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.86 vs. limit=2.0 2023-03-09 13:12:51,813 INFO [train2.py:809] (0/4) Epoch 29, batch 2300, loss[ctc_loss=0.06622, att_loss=0.2257, loss=0.1938, over 16280.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.007204, over 43.00 utterances.], tot_loss[ctc_loss=0.06359, att_loss=0.2312, loss=0.1977, over 3271870.16 frames. utt_duration=1224 frames, utt_pad_proportion=0.0607, over 10701.77 utterances.], batch size: 43, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:13:12,812 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 1.846e+02 2.164e+02 2.611e+02 9.969e+02, threshold=4.329e+02, percent-clipped=2.0 2023-03-09 13:13:13,835 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=2.89 vs. limit=5.0 2023-03-09 13:14:03,892 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=113890.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:14:11,237 INFO [train2.py:809] (0/4) Epoch 29, batch 2350, loss[ctc_loss=0.05572, att_loss=0.2177, loss=0.1853, over 15998.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.007946, over 40.00 utterances.], tot_loss[ctc_loss=0.0636, att_loss=0.231, loss=0.1975, over 3267619.99 frames. utt_duration=1208 frames, utt_pad_proportion=0.0661, over 10835.07 utterances.], batch size: 40, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:14:27,152 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.81 vs. limit=5.0 2023-03-09 13:14:32,527 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6010, 2.7884, 3.8677, 3.5215, 2.9791, 3.5959, 3.5682, 3.6922], device='cuda:0'), covar=tensor([0.0394, 0.1094, 0.0263, 0.0702, 0.1237, 0.0388, 0.0346, 0.0389], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0248, 0.0230, 0.0324, 0.0271, 0.0244, 0.0224, 0.0244], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:15:19,183 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=113938.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:15:29,842 INFO [train2.py:809] (0/4) Epoch 29, batch 2400, loss[ctc_loss=0.0905, att_loss=0.2583, loss=0.2248, over 17057.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009267, over 53.00 utterances.], tot_loss[ctc_loss=0.06395, att_loss=0.2315, loss=0.198, over 3268008.34 frames. utt_duration=1208 frames, utt_pad_proportion=0.06658, over 10837.65 utterances.], batch size: 53, lr: 3.69e-03, grad_scale: 8.0 2023-03-09 13:15:39,392 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0635, 5.2617, 5.2757, 5.2319, 5.3384, 5.2987, 4.9780, 4.7699], device='cuda:0'), covar=tensor([0.1002, 0.0569, 0.0301, 0.0491, 0.0295, 0.0298, 0.0411, 0.0310], device='cuda:0'), in_proj_covar=tensor([0.0540, 0.0382, 0.0378, 0.0384, 0.0446, 0.0451, 0.0382, 0.0421], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:15:50,422 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 1.796e+02 2.124e+02 2.708e+02 5.253e+02, threshold=4.249e+02, percent-clipped=4.0 2023-03-09 13:16:13,533 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=113973.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:16:49,307 INFO [train2.py:809] (0/4) Epoch 29, batch 2450, loss[ctc_loss=0.04987, att_loss=0.2142, loss=0.1813, over 16172.00 frames. utt_duration=1579 frames, utt_pad_proportion=0.006109, over 41.00 utterances.], tot_loss[ctc_loss=0.06473, att_loss=0.2318, loss=0.1984, over 3263964.42 frames. utt_duration=1202 frames, utt_pad_proportion=0.06864, over 10872.99 utterances.], batch size: 41, lr: 3.69e-03, grad_scale: 4.0 2023-03-09 13:16:57,473 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-114000.pt 2023-03-09 13:17:04,625 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114002.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:17:20,636 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5660, 2.8083, 5.0120, 4.1050, 3.2531, 4.3595, 4.8621, 4.7605], device='cuda:0'), covar=tensor([0.0295, 0.1346, 0.0238, 0.0798, 0.1439, 0.0263, 0.0200, 0.0267], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0248, 0.0231, 0.0325, 0.0271, 0.0244, 0.0225, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:17:37,067 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1797, 5.1475, 4.9554, 3.1790, 4.9597, 4.8818, 4.5731, 2.8777], device='cuda:0'), covar=tensor([0.0120, 0.0105, 0.0274, 0.0925, 0.0109, 0.0177, 0.0255, 0.1234], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0106, 0.0112, 0.0112, 0.0091, 0.0120, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:17:42,256 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9204, 5.3122, 5.3721, 5.3294, 5.3719, 5.3873, 5.1234, 4.9462], device='cuda:0'), covar=tensor([0.1358, 0.0599, 0.0362, 0.0545, 0.0440, 0.0370, 0.0417, 0.0347], device='cuda:0'), in_proj_covar=tensor([0.0536, 0.0380, 0.0378, 0.0383, 0.0446, 0.0449, 0.0381, 0.0419], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:18:12,390 INFO [train2.py:809] (0/4) Epoch 29, batch 2500, loss[ctc_loss=0.075, att_loss=0.255, loss=0.219, over 17409.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.0324, over 63.00 utterances.], tot_loss[ctc_loss=0.06425, att_loss=0.2312, loss=0.1978, over 3267046.10 frames. utt_duration=1238 frames, utt_pad_proportion=0.05879, over 10572.66 utterances.], batch size: 63, lr: 3.69e-03, grad_scale: 4.0 2023-03-09 13:18:34,819 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.371e+02 1.923e+02 2.146e+02 2.615e+02 3.828e+02, threshold=4.291e+02, percent-clipped=0.0 2023-03-09 13:18:50,496 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3353, 2.9472, 3.4978, 4.5222, 4.0100, 4.0525, 3.0696, 2.5191], device='cuda:0'), covar=tensor([0.0802, 0.1851, 0.0798, 0.0488, 0.0833, 0.0471, 0.1408, 0.1877], device='cuda:0'), in_proj_covar=tensor([0.0190, 0.0221, 0.0186, 0.0230, 0.0236, 0.0194, 0.0206, 0.0191], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:19:32,906 INFO [train2.py:809] (0/4) Epoch 29, batch 2550, loss[ctc_loss=0.07454, att_loss=0.2494, loss=0.2144, over 17303.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01176, over 55.00 utterances.], tot_loss[ctc_loss=0.06382, att_loss=0.2304, loss=0.1971, over 3265689.25 frames. utt_duration=1235 frames, utt_pad_proportion=0.0609, over 10592.57 utterances.], batch size: 55, lr: 3.68e-03, grad_scale: 4.0 2023-03-09 13:20:23,959 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114127.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:20:52,673 INFO [train2.py:809] (0/4) Epoch 29, batch 2600, loss[ctc_loss=0.04773, att_loss=0.2075, loss=0.1755, over 15662.00 frames. utt_duration=1695 frames, utt_pad_proportion=0.00788, over 37.00 utterances.], tot_loss[ctc_loss=0.06309, att_loss=0.2299, loss=0.1965, over 3271438.27 frames. utt_duration=1262 frames, utt_pad_proportion=0.05286, over 10382.02 utterances.], batch size: 37, lr: 3.68e-03, grad_scale: 4.0 2023-03-09 13:21:15,099 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.450e+01 1.839e+02 2.178e+02 2.659e+02 5.439e+02, threshold=4.356e+02, percent-clipped=3.0 2023-03-09 13:21:28,779 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4288, 4.7417, 4.6223, 4.7462, 4.8640, 4.4698, 3.3134, 4.6933], device='cuda:0'), covar=tensor([0.0130, 0.0132, 0.0149, 0.0086, 0.0095, 0.0147, 0.0750, 0.0206], device='cuda:0'), in_proj_covar=tensor([0.0097, 0.0093, 0.0118, 0.0073, 0.0080, 0.0091, 0.0106, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 13:22:01,471 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114188.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:22:11,798 INFO [train2.py:809] (0/4) Epoch 29, batch 2650, loss[ctc_loss=0.07745, att_loss=0.2481, loss=0.214, over 16948.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.007924, over 50.00 utterances.], tot_loss[ctc_loss=0.06371, att_loss=0.2307, loss=0.1973, over 3279355.93 frames. utt_duration=1249 frames, utt_pad_proportion=0.05237, over 10512.88 utterances.], batch size: 50, lr: 3.68e-03, grad_scale: 4.0 2023-03-09 13:23:31,705 INFO [train2.py:809] (0/4) Epoch 29, batch 2700, loss[ctc_loss=0.06454, att_loss=0.2189, loss=0.188, over 15959.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006576, over 41.00 utterances.], tot_loss[ctc_loss=0.06405, att_loss=0.2313, loss=0.1978, over 3289335.15 frames. utt_duration=1261 frames, utt_pad_proportion=0.04689, over 10447.10 utterances.], batch size: 41, lr: 3.68e-03, grad_scale: 4.0 2023-03-09 13:23:53,981 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.272e+02 1.816e+02 2.102e+02 2.530e+02 4.306e+02, threshold=4.205e+02, percent-clipped=0.0 2023-03-09 13:24:16,605 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114273.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:24:30,919 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.76 vs. limit=2.0 2023-03-09 13:24:51,547 INFO [train2.py:809] (0/4) Epoch 29, batch 2750, loss[ctc_loss=0.06999, att_loss=0.2526, loss=0.2161, over 17043.00 frames. utt_duration=1312 frames, utt_pad_proportion=0.009536, over 52.00 utterances.], tot_loss[ctc_loss=0.06437, att_loss=0.2314, loss=0.198, over 3287476.22 frames. utt_duration=1226 frames, utt_pad_proportion=0.05561, over 10741.09 utterances.], batch size: 52, lr: 3.68e-03, grad_scale: 4.0 2023-03-09 13:25:03,004 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114302.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:25:09,786 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.82 vs. limit=2.0 2023-03-09 13:25:18,417 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7930, 3.4238, 3.3548, 2.9362, 3.4044, 3.5044, 3.4849, 2.4375], device='cuda:0'), covar=tensor([0.0965, 0.1164, 0.2497, 0.2743, 0.0888, 0.1762, 0.0808, 0.2882], device='cuda:0'), in_proj_covar=tensor([0.0206, 0.0209, 0.0224, 0.0274, 0.0185, 0.0285, 0.0208, 0.0229], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:25:32,526 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114321.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:25:50,168 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.72 vs. limit=2.0 2023-03-09 13:26:11,317 INFO [train2.py:809] (0/4) Epoch 29, batch 2800, loss[ctc_loss=0.05909, att_loss=0.2374, loss=0.2017, over 17432.00 frames. utt_duration=884.1 frames, utt_pad_proportion=0.07232, over 79.00 utterances.], tot_loss[ctc_loss=0.06434, att_loss=0.2307, loss=0.1974, over 3279875.15 frames. utt_duration=1232 frames, utt_pad_proportion=0.05586, over 10662.74 utterances.], batch size: 79, lr: 3.68e-03, grad_scale: 8.0 2023-03-09 13:26:16,286 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.7893, 3.4991, 3.6991, 4.6116, 4.1926, 4.1556, 3.2610, 2.7878], device='cuda:0'), covar=tensor([0.0586, 0.1556, 0.0767, 0.0476, 0.0894, 0.0498, 0.1337, 0.1851], device='cuda:0'), in_proj_covar=tensor([0.0191, 0.0222, 0.0187, 0.0231, 0.0238, 0.0195, 0.0207, 0.0192], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:26:19,158 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114350.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:26:33,215 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.204e+02 1.822e+02 2.190e+02 2.901e+02 6.134e+02, threshold=4.379e+02, percent-clipped=9.0 2023-03-09 13:27:30,405 INFO [train2.py:809] (0/4) Epoch 29, batch 2850, loss[ctc_loss=0.06819, att_loss=0.244, loss=0.2088, over 16474.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.006056, over 46.00 utterances.], tot_loss[ctc_loss=0.06408, att_loss=0.2305, loss=0.1972, over 3275556.61 frames. utt_duration=1241 frames, utt_pad_proportion=0.05476, over 10574.60 utterances.], batch size: 46, lr: 3.68e-03, grad_scale: 8.0 2023-03-09 13:27:44,245 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.4769, 5.3476, 5.1855, 3.6691, 5.2288, 5.0426, 4.7559, 3.1071], device='cuda:0'), covar=tensor([0.0090, 0.0094, 0.0245, 0.0784, 0.0081, 0.0173, 0.0254, 0.1217], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0106, 0.0112, 0.0112, 0.0091, 0.0120, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:28:50,606 INFO [train2.py:809] (0/4) Epoch 29, batch 2900, loss[ctc_loss=0.1056, att_loss=0.257, loss=0.2267, over 14542.00 frames. utt_duration=402.9 frames, utt_pad_proportion=0.2993, over 145.00 utterances.], tot_loss[ctc_loss=0.06379, att_loss=0.2308, loss=0.1974, over 3275114.89 frames. utt_duration=1218 frames, utt_pad_proportion=0.0601, over 10772.41 utterances.], batch size: 145, lr: 3.68e-03, grad_scale: 8.0 2023-03-09 13:29:12,892 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.369e+02 1.896e+02 2.215e+02 2.552e+02 6.599e+02, threshold=4.430e+02, percent-clipped=4.0 2023-03-09 13:29:52,208 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=114483.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:29:52,481 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4687, 2.5761, 4.8567, 3.7883, 2.9837, 4.1415, 4.4938, 4.5880], device='cuda:0'), covar=tensor([0.0291, 0.1588, 0.0228, 0.0870, 0.1669, 0.0298, 0.0251, 0.0286], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0248, 0.0231, 0.0326, 0.0272, 0.0244, 0.0227, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:30:10,924 INFO [train2.py:809] (0/4) Epoch 29, batch 2950, loss[ctc_loss=0.08646, att_loss=0.2524, loss=0.2192, over 13415.00 frames. utt_duration=369 frames, utt_pad_proportion=0.356, over 146.00 utterances.], tot_loss[ctc_loss=0.06419, att_loss=0.231, loss=0.1977, over 3275196.98 frames. utt_duration=1211 frames, utt_pad_proportion=0.06127, over 10829.69 utterances.], batch size: 146, lr: 3.68e-03, grad_scale: 8.0 2023-03-09 13:31:30,953 INFO [train2.py:809] (0/4) Epoch 29, batch 3000, loss[ctc_loss=0.0473, att_loss=0.2213, loss=0.1865, over 16687.00 frames. utt_duration=1452 frames, utt_pad_proportion=0.006602, over 46.00 utterances.], tot_loss[ctc_loss=0.06444, att_loss=0.2306, loss=0.1973, over 3261953.24 frames. utt_duration=1196 frames, utt_pad_proportion=0.06979, over 10924.99 utterances.], batch size: 46, lr: 3.68e-03, grad_scale: 8.0 2023-03-09 13:31:30,957 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 13:31:44,956 INFO [train2.py:843] (0/4) Epoch 29, validation: ctc_loss=0.04115, att_loss=0.2348, loss=0.1961, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 13:31:44,957 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16157MB 2023-03-09 13:32:06,763 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.233e+02 1.877e+02 2.271e+02 2.617e+02 4.276e+02, threshold=4.543e+02, percent-clipped=0.0 2023-03-09 13:33:04,602 INFO [train2.py:809] (0/4) Epoch 29, batch 3050, loss[ctc_loss=0.06807, att_loss=0.2434, loss=0.2083, over 16775.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.006067, over 48.00 utterances.], tot_loss[ctc_loss=0.06499, att_loss=0.2305, loss=0.1974, over 3259933.75 frames. utt_duration=1211 frames, utt_pad_proportion=0.06693, over 10784.99 utterances.], batch size: 48, lr: 3.68e-03, grad_scale: 8.0 2023-03-09 13:34:24,746 INFO [train2.py:809] (0/4) Epoch 29, batch 3100, loss[ctc_loss=0.06121, att_loss=0.2327, loss=0.1984, over 16478.00 frames. utt_duration=1434 frames, utt_pad_proportion=0.005906, over 46.00 utterances.], tot_loss[ctc_loss=0.06477, att_loss=0.2314, loss=0.1981, over 3269709.63 frames. utt_duration=1205 frames, utt_pad_proportion=0.06645, over 10870.39 utterances.], batch size: 46, lr: 3.68e-03, grad_scale: 8.0 2023-03-09 13:34:31,380 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3255, 4.3084, 4.4909, 4.4635, 4.9941, 4.4088, 4.3057, 2.8197], device='cuda:0'), covar=tensor([0.0336, 0.0557, 0.0361, 0.0450, 0.0629, 0.0268, 0.0435, 0.1547], device='cuda:0'), in_proj_covar=tensor([0.0202, 0.0229, 0.0224, 0.0241, 0.0387, 0.0199, 0.0218, 0.0224], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:34:46,588 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.249e+02 1.785e+02 2.252e+02 2.835e+02 6.156e+02, threshold=4.505e+02, percent-clipped=2.0 2023-03-09 13:34:56,684 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.3374, 5.2365, 5.0064, 3.4036, 5.0997, 4.9524, 4.5668, 2.9326], device='cuda:0'), covar=tensor([0.0100, 0.0103, 0.0316, 0.0803, 0.0100, 0.0171, 0.0283, 0.1269], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0106, 0.0112, 0.0112, 0.0090, 0.0120, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:35:44,837 INFO [train2.py:809] (0/4) Epoch 29, batch 3150, loss[ctc_loss=0.07194, att_loss=0.2353, loss=0.2026, over 17048.00 frames. utt_duration=1313 frames, utt_pad_proportion=0.008882, over 52.00 utterances.], tot_loss[ctc_loss=0.0643, att_loss=0.2312, loss=0.1978, over 3277000.50 frames. utt_duration=1207 frames, utt_pad_proportion=0.06398, over 10874.22 utterances.], batch size: 52, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:36:18,398 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114716.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:36:42,517 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.9523, 3.6076, 3.6733, 3.1394, 3.6446, 3.6881, 3.6881, 2.6251], device='cuda:0'), covar=tensor([0.1050, 0.1165, 0.1276, 0.2582, 0.0987, 0.2012, 0.0778, 0.2846], device='cuda:0'), in_proj_covar=tensor([0.0209, 0.0210, 0.0227, 0.0277, 0.0189, 0.0288, 0.0211, 0.0232], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:37:04,183 INFO [train2.py:809] (0/4) Epoch 29, batch 3200, loss[ctc_loss=0.0656, att_loss=0.2514, loss=0.2143, over 17317.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02355, over 59.00 utterances.], tot_loss[ctc_loss=0.06342, att_loss=0.2306, loss=0.1972, over 3277922.61 frames. utt_duration=1244 frames, utt_pad_proportion=0.0552, over 10553.73 utterances.], batch size: 59, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:37:27,012 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.042e+02 1.814e+02 2.180e+02 2.558e+02 5.727e+02, threshold=4.361e+02, percent-clipped=1.0 2023-03-09 13:37:52,431 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9825, 4.1946, 4.1533, 4.5014, 2.5402, 4.3724, 2.6724, 2.1709], device='cuda:0'), covar=tensor([0.0618, 0.0375, 0.0761, 0.0272, 0.1746, 0.0288, 0.1484, 0.1488], device='cuda:0'), in_proj_covar=tensor([0.0221, 0.0193, 0.0265, 0.0182, 0.0223, 0.0174, 0.0234, 0.0203], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:37:54,488 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114777.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:38:04,232 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=114783.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:38:22,854 INFO [train2.py:809] (0/4) Epoch 29, batch 3250, loss[ctc_loss=0.05209, att_loss=0.2254, loss=0.1907, over 16761.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.006263, over 48.00 utterances.], tot_loss[ctc_loss=0.06378, att_loss=0.2302, loss=0.1969, over 3273561.62 frames. utt_duration=1250 frames, utt_pad_proportion=0.05544, over 10489.50 utterances.], batch size: 48, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:38:23,283 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4849, 2.7073, 4.9061, 3.8798, 3.1043, 4.3873, 4.7862, 4.7373], device='cuda:0'), covar=tensor([0.0300, 0.1425, 0.0268, 0.0916, 0.1624, 0.0243, 0.0205, 0.0264], device='cuda:0'), in_proj_covar=tensor([0.0237, 0.0248, 0.0231, 0.0325, 0.0271, 0.0244, 0.0226, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:38:26,160 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0480, 5.4331, 5.5652, 5.4251, 5.5740, 6.0465, 5.3618, 6.1073], device='cuda:0'), covar=tensor([0.0741, 0.0736, 0.0924, 0.1328, 0.1763, 0.0781, 0.0691, 0.0645], device='cuda:0'), in_proj_covar=tensor([0.0911, 0.0529, 0.0639, 0.0676, 0.0903, 0.0659, 0.0509, 0.0634], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 13:38:41,000 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.36 vs. limit=2.0 2023-03-09 13:39:20,493 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=114831.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:39:42,519 INFO [train2.py:809] (0/4) Epoch 29, batch 3300, loss[ctc_loss=0.04867, att_loss=0.2286, loss=0.1926, over 17030.00 frames. utt_duration=1337 frames, utt_pad_proportion=0.007957, over 51.00 utterances.], tot_loss[ctc_loss=0.06359, att_loss=0.2299, loss=0.1966, over 3261424.76 frames. utt_duration=1237 frames, utt_pad_proportion=0.06158, over 10559.96 utterances.], batch size: 51, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:40:05,653 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.268e+02 1.794e+02 2.073e+02 2.411e+02 5.454e+02, threshold=4.146e+02, percent-clipped=1.0 2023-03-09 13:40:09,041 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4805, 4.8069, 4.6619, 4.7691, 4.8212, 4.5198, 3.4003, 4.7401], device='cuda:0'), covar=tensor([0.0133, 0.0114, 0.0152, 0.0090, 0.0125, 0.0149, 0.0728, 0.0205], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0094, 0.0119, 0.0074, 0.0081, 0.0092, 0.0107, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 13:40:22,996 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=114871.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:41:01,520 INFO [train2.py:809] (0/4) Epoch 29, batch 3350, loss[ctc_loss=0.06365, att_loss=0.237, loss=0.2023, over 16629.00 frames. utt_duration=1417 frames, utt_pad_proportion=0.00511, over 47.00 utterances.], tot_loss[ctc_loss=0.06399, att_loss=0.2304, loss=0.1971, over 3264949.92 frames. utt_duration=1231 frames, utt_pad_proportion=0.06278, over 10625.49 utterances.], batch size: 47, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:41:45,652 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1831, 3.8042, 3.2602, 3.3770, 4.0692, 3.7271, 2.9462, 4.3084], device='cuda:0'), covar=tensor([0.1015, 0.0552, 0.1165, 0.0788, 0.0772, 0.0760, 0.0977, 0.0571], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0230, 0.0233, 0.0209, 0.0293, 0.0252, 0.0207, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 13:42:00,803 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=114932.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:42:16,369 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0712, 5.3904, 4.9458, 5.4083, 4.8269, 5.0231, 5.4736, 5.2687], device='cuda:0'), covar=tensor([0.0521, 0.0250, 0.0718, 0.0301, 0.0369, 0.0258, 0.0215, 0.0187], device='cuda:0'), in_proj_covar=tensor([0.0403, 0.0339, 0.0379, 0.0380, 0.0335, 0.0244, 0.0321, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 13:42:20,957 INFO [train2.py:809] (0/4) Epoch 29, batch 3400, loss[ctc_loss=0.06145, att_loss=0.2361, loss=0.2011, over 17046.00 frames. utt_duration=1288 frames, utt_pad_proportion=0.009978, over 53.00 utterances.], tot_loss[ctc_loss=0.06415, att_loss=0.2304, loss=0.1972, over 3263206.15 frames. utt_duration=1221 frames, utt_pad_proportion=0.06585, over 10703.41 utterances.], batch size: 53, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:42:44,853 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.258e+02 1.858e+02 2.229e+02 2.688e+02 7.167e+02, threshold=4.457e+02, percent-clipped=6.0 2023-03-09 13:43:41,690 INFO [train2.py:809] (0/4) Epoch 29, batch 3450, loss[ctc_loss=0.05827, att_loss=0.2289, loss=0.1948, over 17294.00 frames. utt_duration=877 frames, utt_pad_proportion=0.07979, over 79.00 utterances.], tot_loss[ctc_loss=0.06334, att_loss=0.2299, loss=0.1966, over 3262883.68 frames. utt_duration=1245 frames, utt_pad_proportion=0.06008, over 10497.70 utterances.], batch size: 79, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:45:02,581 INFO [train2.py:809] (0/4) Epoch 29, batch 3500, loss[ctc_loss=0.08743, att_loss=0.2521, loss=0.2192, over 17064.00 frames. utt_duration=1289 frames, utt_pad_proportion=0.009137, over 53.00 utterances.], tot_loss[ctc_loss=0.06284, att_loss=0.2295, loss=0.1962, over 3265385.15 frames. utt_duration=1254 frames, utt_pad_proportion=0.05788, over 10424.70 utterances.], batch size: 53, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:45:26,259 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.209e+02 1.733e+02 2.269e+02 2.703e+02 9.473e+02, threshold=4.539e+02, percent-clipped=4.0 2023-03-09 13:45:46,155 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115072.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:46:02,312 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8451, 6.0777, 5.4945, 5.7957, 5.7610, 5.2598, 5.5485, 5.1962], device='cuda:0'), covar=tensor([0.1145, 0.0824, 0.1089, 0.0867, 0.0919, 0.1468, 0.2029, 0.2370], device='cuda:0'), in_proj_covar=tensor([0.0562, 0.0636, 0.0492, 0.0473, 0.0455, 0.0485, 0.0637, 0.0545], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:46:09,457 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115086.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:46:22,696 INFO [train2.py:809] (0/4) Epoch 29, batch 3550, loss[ctc_loss=0.09517, att_loss=0.2576, loss=0.2251, over 17075.00 frames. utt_duration=1221 frames, utt_pad_proportion=0.01698, over 56.00 utterances.], tot_loss[ctc_loss=0.06372, att_loss=0.2298, loss=0.1966, over 3273975.53 frames. utt_duration=1266 frames, utt_pad_proportion=0.0518, over 10360.59 utterances.], batch size: 56, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:46:45,257 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6931, 5.0773, 4.8968, 4.9711, 5.1208, 4.7637, 3.6636, 4.9965], device='cuda:0'), covar=tensor([0.0118, 0.0103, 0.0151, 0.0087, 0.0093, 0.0129, 0.0650, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0095, 0.0120, 0.0075, 0.0082, 0.0093, 0.0109, 0.0114], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0004], device='cuda:0') 2023-03-09 13:46:51,205 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8617, 6.0919, 5.5708, 5.7875, 5.8147, 5.2982, 5.5691, 5.2834], device='cuda:0'), covar=tensor([0.1282, 0.0879, 0.1025, 0.0895, 0.1017, 0.1537, 0.2262, 0.2273], device='cuda:0'), in_proj_covar=tensor([0.0562, 0.0637, 0.0491, 0.0473, 0.0454, 0.0484, 0.0637, 0.0545], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:47:42,043 INFO [train2.py:809] (0/4) Epoch 29, batch 3600, loss[ctc_loss=0.05459, att_loss=0.2445, loss=0.2065, over 17293.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01229, over 55.00 utterances.], tot_loss[ctc_loss=0.06361, att_loss=0.2301, loss=0.1968, over 3275617.36 frames. utt_duration=1239 frames, utt_pad_proportion=0.05788, over 10590.68 utterances.], batch size: 55, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:47:45,560 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115147.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:48:05,805 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.296e+02 1.777e+02 2.035e+02 2.403e+02 5.227e+02, threshold=4.069e+02, percent-clipped=1.0 2023-03-09 13:49:02,355 INFO [train2.py:809] (0/4) Epoch 29, batch 3650, loss[ctc_loss=0.05933, att_loss=0.1985, loss=0.1706, over 14508.00 frames. utt_duration=1815 frames, utt_pad_proportion=0.04561, over 32.00 utterances.], tot_loss[ctc_loss=0.06425, att_loss=0.2306, loss=0.1974, over 3274155.18 frames. utt_duration=1214 frames, utt_pad_proportion=0.06337, over 10802.34 utterances.], batch size: 32, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:49:51,375 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0652, 5.4399, 4.5065, 5.5715, 4.9146, 5.1672, 5.4557, 5.3018], device='cuda:0'), covar=tensor([0.0586, 0.0331, 0.1179, 0.0342, 0.0326, 0.0241, 0.0300, 0.0200], device='cuda:0'), in_proj_covar=tensor([0.0404, 0.0340, 0.0380, 0.0382, 0.0336, 0.0244, 0.0320, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 13:49:54,432 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115227.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:50:03,652 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.61 vs. limit=5.0 2023-03-09 13:50:23,436 INFO [train2.py:809] (0/4) Epoch 29, batch 3700, loss[ctc_loss=0.05215, att_loss=0.2122, loss=0.1802, over 15770.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.008185, over 38.00 utterances.], tot_loss[ctc_loss=0.06428, att_loss=0.2308, loss=0.1975, over 3272920.49 frames. utt_duration=1214 frames, utt_pad_proportion=0.06409, over 10796.53 utterances.], batch size: 38, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:50:47,789 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.294e+02 1.820e+02 2.082e+02 2.598e+02 4.716e+02, threshold=4.165e+02, percent-clipped=2.0 2023-03-09 13:51:05,897 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0416, 6.3005, 5.6823, 5.9297, 5.9746, 5.3920, 5.8320, 5.3594], device='cuda:0'), covar=tensor([0.1415, 0.0942, 0.1129, 0.0993, 0.0888, 0.1791, 0.2242, 0.2466], device='cuda:0'), in_proj_covar=tensor([0.0568, 0.0641, 0.0497, 0.0479, 0.0458, 0.0489, 0.0645, 0.0551], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:51:44,492 INFO [train2.py:809] (0/4) Epoch 29, batch 3750, loss[ctc_loss=0.07196, att_loss=0.2452, loss=0.2106, over 17093.00 frames. utt_duration=692.2 frames, utt_pad_proportion=0.1294, over 99.00 utterances.], tot_loss[ctc_loss=0.06307, att_loss=0.2306, loss=0.1971, over 3276004.95 frames. utt_duration=1209 frames, utt_pad_proportion=0.06245, over 10848.05 utterances.], batch size: 99, lr: 3.67e-03, grad_scale: 8.0 2023-03-09 13:52:11,599 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2868, 3.8190, 3.3403, 3.5271, 4.0255, 3.7645, 3.1098, 4.2663], device='cuda:0'), covar=tensor([0.0925, 0.0493, 0.1043, 0.0702, 0.0724, 0.0716, 0.0869, 0.0469], device='cuda:0'), in_proj_covar=tensor([0.0210, 0.0230, 0.0233, 0.0210, 0.0292, 0.0252, 0.0207, 0.0301], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 13:52:49,244 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.80 vs. limit=2.0 2023-03-09 13:53:04,022 INFO [train2.py:809] (0/4) Epoch 29, batch 3800, loss[ctc_loss=0.04795, att_loss=0.1971, loss=0.1672, over 15885.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009502, over 39.00 utterances.], tot_loss[ctc_loss=0.06269, att_loss=0.2298, loss=0.1964, over 3262906.04 frames. utt_duration=1222 frames, utt_pad_proportion=0.06202, over 10695.02 utterances.], batch size: 39, lr: 3.66e-03, grad_scale: 8.0 2023-03-09 13:53:27,652 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.195e+02 1.885e+02 2.239e+02 2.673e+02 5.789e+02, threshold=4.479e+02, percent-clipped=6.0 2023-03-09 13:53:37,742 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7927, 5.0532, 4.6233, 5.0629, 4.5430, 4.7139, 5.1335, 4.9288], device='cuda:0'), covar=tensor([0.0576, 0.0291, 0.0795, 0.0353, 0.0394, 0.0302, 0.0223, 0.0201], device='cuda:0'), in_proj_covar=tensor([0.0408, 0.0344, 0.0384, 0.0385, 0.0340, 0.0246, 0.0323, 0.0306], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 13:53:43,980 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115370.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:53:46,174 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115371.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:53:47,651 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115372.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:54:23,176 INFO [train2.py:809] (0/4) Epoch 29, batch 3850, loss[ctc_loss=0.05635, att_loss=0.2146, loss=0.183, over 15939.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.007912, over 41.00 utterances.], tot_loss[ctc_loss=0.06325, att_loss=0.2301, loss=0.1968, over 3263053.49 frames. utt_duration=1228 frames, utt_pad_proportion=0.06067, over 10644.32 utterances.], batch size: 41, lr: 3.66e-03, grad_scale: 8.0 2023-03-09 13:54:48,122 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1024, 5.2981, 5.2532, 5.2929, 5.3667, 5.3182, 4.9622, 4.7818], device='cuda:0'), covar=tensor([0.0983, 0.0526, 0.0293, 0.0418, 0.0280, 0.0320, 0.0454, 0.0344], device='cuda:0'), in_proj_covar=tensor([0.0540, 0.0385, 0.0379, 0.0385, 0.0451, 0.0455, 0.0384, 0.0424], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 13:55:01,794 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115420.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:55:18,363 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115431.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:55:20,002 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115432.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:55:34,958 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115442.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:55:39,327 INFO [train2.py:809] (0/4) Epoch 29, batch 3900, loss[ctc_loss=0.07339, att_loss=0.2455, loss=0.2111, over 17340.00 frames. utt_duration=1177 frames, utt_pad_proportion=0.02217, over 59.00 utterances.], tot_loss[ctc_loss=0.06334, att_loss=0.2305, loss=0.1971, over 3266170.18 frames. utt_duration=1238 frames, utt_pad_proportion=0.05813, over 10564.07 utterances.], batch size: 59, lr: 3.66e-03, grad_scale: 8.0 2023-03-09 13:56:00,047 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5907, 4.9875, 4.7717, 4.9199, 5.0280, 4.7308, 3.5358, 4.9478], device='cuda:0'), covar=tensor([0.0139, 0.0121, 0.0169, 0.0095, 0.0107, 0.0123, 0.0709, 0.0198], device='cuda:0'), in_proj_covar=tensor([0.0099, 0.0094, 0.0119, 0.0075, 0.0081, 0.0092, 0.0108, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 13:56:02,813 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.234e+02 1.900e+02 2.175e+02 2.636e+02 4.480e+02, threshold=4.350e+02, percent-clipped=1.0 2023-03-09 13:56:41,271 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2215, 3.0249, 3.2440, 4.3940, 3.9120, 3.8929, 2.9616, 2.1268], device='cuda:0'), covar=tensor([0.0893, 0.1732, 0.0954, 0.0459, 0.0843, 0.0470, 0.1418, 0.2299], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0221, 0.0188, 0.0231, 0.0240, 0.0195, 0.0208, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 13:56:55,925 INFO [train2.py:809] (0/4) Epoch 29, batch 3950, loss[ctc_loss=0.0855, att_loss=0.2433, loss=0.2117, over 13760.00 frames. utt_duration=376 frames, utt_pad_proportion=0.3415, over 147.00 utterances.], tot_loss[ctc_loss=0.06402, att_loss=0.2307, loss=0.1973, over 3267125.08 frames. utt_duration=1222 frames, utt_pad_proportion=0.0626, over 10703.34 utterances.], batch size: 147, lr: 3.66e-03, grad_scale: 8.0 2023-03-09 13:57:22,069 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8738, 2.3673, 2.5769, 2.8120, 2.7080, 2.7955, 2.3575, 3.0060], device='cuda:0'), covar=tensor([0.1333, 0.2190, 0.1641, 0.1232, 0.1976, 0.1073, 0.1979, 0.1400], device='cuda:0'), in_proj_covar=tensor([0.0153, 0.0150, 0.0148, 0.0143, 0.0161, 0.0138, 0.0163, 0.0139], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 13:57:42,179 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-03-09 13:57:44,598 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115527.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:57:47,118 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-29.pt 2023-03-09 13:58:12,588 INFO [train2.py:809] (0/4) Epoch 30, batch 0, loss[ctc_loss=0.0569, att_loss=0.2243, loss=0.1908, over 16535.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006377, over 45.00 utterances.], tot_loss[ctc_loss=0.0569, att_loss=0.2243, loss=0.1908, over 16535.00 frames. utt_duration=1472 frames, utt_pad_proportion=0.006377, over 45.00 utterances.], batch size: 45, lr: 3.60e-03, grad_scale: 8.0 2023-03-09 13:58:12,590 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 13:58:17,435 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0373, 5.1148, 4.9328, 2.4471, 2.2320, 3.3378, 2.5308, 3.8293], device='cuda:0'), covar=tensor([0.0739, 0.0359, 0.0334, 0.5069, 0.5499, 0.2185, 0.4127, 0.1578], device='cuda:0'), in_proj_covar=tensor([0.0365, 0.0307, 0.0280, 0.0253, 0.0340, 0.0332, 0.0265, 0.0373], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 13:58:24,680 INFO [train2.py:843] (0/4) Epoch 30, validation: ctc_loss=0.03959, att_loss=0.2341, loss=0.1952, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 13:58:24,680 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16157MB 2023-03-09 13:58:49,313 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3661, 2.9293, 3.3801, 4.4615, 3.9485, 3.9278, 2.9170, 2.2095], device='cuda:0'), covar=tensor([0.0858, 0.2027, 0.0963, 0.0582, 0.0955, 0.0544, 0.1651, 0.2426], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0223, 0.0189, 0.0232, 0.0241, 0.0196, 0.0209, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-03-09 13:59:05,303 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.24 vs. limit=2.0 2023-03-09 13:59:13,838 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.172e+02 1.763e+02 2.122e+02 2.494e+02 5.580e+02, threshold=4.245e+02, percent-clipped=4.0 2023-03-09 13:59:38,519 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115575.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 13:59:44,601 INFO [train2.py:809] (0/4) Epoch 30, batch 50, loss[ctc_loss=0.06487, att_loss=0.2436, loss=0.2079, over 17567.00 frames. utt_duration=1020 frames, utt_pad_proportion=0.03776, over 69.00 utterances.], tot_loss[ctc_loss=0.06001, att_loss=0.2273, loss=0.1938, over 739929.07 frames. utt_duration=1307 frames, utt_pad_proportion=0.0414, over 2267.24 utterances.], batch size: 69, lr: 3.60e-03, grad_scale: 8.0 2023-03-09 14:01:04,301 INFO [train2.py:809] (0/4) Epoch 30, batch 100, loss[ctc_loss=0.05418, att_loss=0.2231, loss=0.1893, over 16400.00 frames. utt_duration=1492 frames, utt_pad_proportion=0.007047, over 44.00 utterances.], tot_loss[ctc_loss=0.06269, att_loss=0.2307, loss=0.1971, over 1302148.79 frames. utt_duration=1272 frames, utt_pad_proportion=0.04886, over 4098.24 utterances.], batch size: 44, lr: 3.60e-03, grad_scale: 8.0 2023-03-09 14:01:06,171 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3853, 2.9365, 3.3169, 4.3960, 3.8942, 3.8502, 2.8729, 2.2771], device='cuda:0'), covar=tensor([0.0820, 0.1920, 0.0968, 0.0611, 0.0998, 0.0571, 0.1593, 0.2267], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0222, 0.0189, 0.0232, 0.0241, 0.0196, 0.0209, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-03-09 14:01:54,045 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.157e+02 1.758e+02 2.072e+02 2.595e+02 5.402e+02, threshold=4.143e+02, percent-clipped=3.0 2023-03-09 14:02:13,418 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2705, 2.7815, 3.2159, 4.3517, 3.7841, 3.8281, 2.7505, 2.2632], device='cuda:0'), covar=tensor([0.0888, 0.1909, 0.0933, 0.0569, 0.1045, 0.0548, 0.1728, 0.2186], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0223, 0.0190, 0.0233, 0.0242, 0.0197, 0.0210, 0.0196], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0003, 0.0002], device='cuda:0') 2023-03-09 14:02:22,591 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9272, 5.3890, 4.5015, 5.5124, 4.8528, 5.1716, 5.4064, 5.2248], device='cuda:0'), covar=tensor([0.0594, 0.0297, 0.1208, 0.0319, 0.0330, 0.0223, 0.0328, 0.0213], device='cuda:0'), in_proj_covar=tensor([0.0410, 0.0344, 0.0386, 0.0387, 0.0341, 0.0248, 0.0323, 0.0308], device='cuda:0'), out_proj_covar=tensor([0.0007, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 14:02:23,889 INFO [train2.py:809] (0/4) Epoch 30, batch 150, loss[ctc_loss=0.07358, att_loss=0.2337, loss=0.2017, over 15965.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.005786, over 41.00 utterances.], tot_loss[ctc_loss=0.0644, att_loss=0.2323, loss=0.1987, over 1742292.09 frames. utt_duration=1222 frames, utt_pad_proportion=0.06004, over 5708.83 utterances.], batch size: 41, lr: 3.60e-03, grad_scale: 8.0 2023-03-09 14:03:04,242 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.8485, 6.0907, 5.5966, 5.7775, 5.7536, 5.2094, 5.4645, 5.2185], device='cuda:0'), covar=tensor([0.1199, 0.0868, 0.0931, 0.0882, 0.1063, 0.1684, 0.2331, 0.2118], device='cuda:0'), in_proj_covar=tensor([0.0563, 0.0638, 0.0492, 0.0476, 0.0454, 0.0486, 0.0641, 0.0543], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:03:39,620 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115726.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:03:39,872 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5612, 2.6630, 4.8823, 3.9228, 3.0836, 4.3226, 4.8114, 4.7467], device='cuda:0'), covar=tensor([0.0286, 0.1482, 0.0321, 0.0933, 0.1582, 0.0259, 0.0193, 0.0246], device='cuda:0'), in_proj_covar=tensor([0.0239, 0.0247, 0.0233, 0.0324, 0.0272, 0.0246, 0.0226, 0.0245], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 14:03:41,131 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=115727.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:03:44,043 INFO [train2.py:809] (0/4) Epoch 30, batch 200, loss[ctc_loss=0.04886, att_loss=0.202, loss=0.1714, over 14570.00 frames. utt_duration=1823 frames, utt_pad_proportion=0.03591, over 32.00 utterances.], tot_loss[ctc_loss=0.06391, att_loss=0.2324, loss=0.1987, over 2082183.90 frames. utt_duration=1174 frames, utt_pad_proportion=0.07174, over 7105.84 utterances.], batch size: 32, lr: 3.60e-03, grad_scale: 8.0 2023-03-09 14:04:04,996 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=115742.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:04:23,451 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.8860, 3.6418, 3.6377, 3.0044, 3.6510, 3.7336, 3.7194, 2.5513], device='cuda:0'), covar=tensor([0.1101, 0.0994, 0.1388, 0.3431, 0.0890, 0.1842, 0.0778, 0.3089], device='cuda:0'), in_proj_covar=tensor([0.0211, 0.0214, 0.0227, 0.0279, 0.0190, 0.0289, 0.0212, 0.0235], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 14:04:32,781 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.110e+02 1.790e+02 2.190e+02 2.640e+02 5.774e+02, threshold=4.380e+02, percent-clipped=3.0 2023-03-09 14:05:03,782 INFO [train2.py:809] (0/4) Epoch 30, batch 250, loss[ctc_loss=0.06036, att_loss=0.2452, loss=0.2082, over 16945.00 frames. utt_duration=1357 frames, utt_pad_proportion=0.00878, over 50.00 utterances.], tot_loss[ctc_loss=0.06434, att_loss=0.2328, loss=0.1991, over 2354068.14 frames. utt_duration=1183 frames, utt_pad_proportion=0.06859, over 7970.92 utterances.], batch size: 50, lr: 3.60e-03, grad_scale: 8.0 2023-03-09 14:05:15,047 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3869, 2.4001, 4.8108, 3.7808, 2.9573, 4.0730, 4.5845, 4.5636], device='cuda:0'), covar=tensor([0.0327, 0.1684, 0.0240, 0.0871, 0.1687, 0.0301, 0.0225, 0.0278], device='cuda:0'), in_proj_covar=tensor([0.0240, 0.0247, 0.0233, 0.0324, 0.0273, 0.0246, 0.0226, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 14:05:18,717 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9382, 5.0589, 4.8939, 2.2340, 2.0484, 3.0945, 2.4320, 3.9186], device='cuda:0'), covar=tensor([0.0820, 0.0308, 0.0298, 0.5376, 0.5576, 0.2291, 0.4030, 0.1503], device='cuda:0'), in_proj_covar=tensor([0.0367, 0.0308, 0.0281, 0.0254, 0.0341, 0.0332, 0.0265, 0.0374], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 14:05:21,456 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=115790.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:06:23,344 INFO [train2.py:809] (0/4) Epoch 30, batch 300, loss[ctc_loss=0.0586, att_loss=0.2449, loss=0.2076, over 17352.00 frames. utt_duration=1103 frames, utt_pad_proportion=0.03319, over 63.00 utterances.], tot_loss[ctc_loss=0.06334, att_loss=0.2322, loss=0.1984, over 2562263.66 frames. utt_duration=1206 frames, utt_pad_proportion=0.06171, over 8506.72 utterances.], batch size: 63, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:07:12,541 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.192e+02 1.839e+02 2.070e+02 2.435e+02 4.420e+02, threshold=4.140e+02, percent-clipped=2.0 2023-03-09 14:07:43,750 INFO [train2.py:809] (0/4) Epoch 30, batch 350, loss[ctc_loss=0.05671, att_loss=0.2237, loss=0.1903, over 16288.00 frames. utt_duration=1517 frames, utt_pad_proportion=0.006777, over 43.00 utterances.], tot_loss[ctc_loss=0.06256, att_loss=0.2313, loss=0.1976, over 2716839.91 frames. utt_duration=1222 frames, utt_pad_proportion=0.05987, over 8904.90 utterances.], batch size: 43, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:08:05,280 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=115892.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 14:09:04,301 INFO [train2.py:809] (0/4) Epoch 30, batch 400, loss[ctc_loss=0.06874, att_loss=0.226, loss=0.1945, over 15988.00 frames. utt_duration=1601 frames, utt_pad_proportion=0.008333, over 40.00 utterances.], tot_loss[ctc_loss=0.06232, att_loss=0.231, loss=0.1973, over 2849348.17 frames. utt_duration=1244 frames, utt_pad_proportion=0.05173, over 9174.12 utterances.], batch size: 40, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:09:19,774 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.28 vs. limit=2.0 2023-03-09 14:09:43,428 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=115953.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 14:09:53,665 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.410e+02 1.924e+02 2.363e+02 2.895e+02 9.842e+02, threshold=4.725e+02, percent-clipped=4.0 2023-03-09 14:10:24,227 INFO [train2.py:809] (0/4) Epoch 30, batch 450, loss[ctc_loss=0.06403, att_loss=0.2159, loss=0.1855, over 14615.00 frames. utt_duration=1829 frames, utt_pad_proportion=0.03097, over 32.00 utterances.], tot_loss[ctc_loss=0.06288, att_loss=0.2317, loss=0.1979, over 2947613.64 frames. utt_duration=1216 frames, utt_pad_proportion=0.05871, over 9711.36 utterances.], batch size: 32, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:10:57,394 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-116000.pt 2023-03-09 14:11:24,419 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116014.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:11:43,437 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116026.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:11:45,001 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116027.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:11:48,385 INFO [train2.py:809] (0/4) Epoch 30, batch 500, loss[ctc_loss=0.06814, att_loss=0.2209, loss=0.1903, over 15940.00 frames. utt_duration=1556 frames, utt_pad_proportion=0.007373, over 41.00 utterances.], tot_loss[ctc_loss=0.06274, att_loss=0.231, loss=0.1973, over 3026186.88 frames. utt_duration=1228 frames, utt_pad_proportion=0.05322, over 9870.56 utterances.], batch size: 41, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:11:56,278 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4525, 2.5859, 4.7092, 3.8243, 3.1411, 4.1930, 4.2679, 4.4885], device='cuda:0'), covar=tensor([0.0195, 0.1374, 0.0178, 0.0769, 0.1382, 0.0231, 0.0232, 0.0218], device='cuda:0'), in_proj_covar=tensor([0.0239, 0.0246, 0.0232, 0.0322, 0.0271, 0.0245, 0.0226, 0.0246], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 14:12:37,217 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.075e+02 1.798e+02 2.149e+02 2.710e+02 6.262e+02, threshold=4.298e+02, percent-clipped=5.0 2023-03-09 14:12:59,516 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116074.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:13:01,051 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116075.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:13:01,331 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116075.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 14:13:07,561 INFO [train2.py:809] (0/4) Epoch 30, batch 550, loss[ctc_loss=0.06806, att_loss=0.2352, loss=0.2018, over 16479.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.00574, over 46.00 utterances.], tot_loss[ctc_loss=0.06252, att_loss=0.2305, loss=0.1969, over 3087685.96 frames. utt_duration=1223 frames, utt_pad_proportion=0.05314, over 10113.99 utterances.], batch size: 46, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:13:47,221 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8499, 4.9808, 4.4030, 2.8285, 4.7513, 4.6375, 3.9837, 2.2662], device='cuda:0'), covar=tensor([0.0233, 0.0134, 0.0482, 0.1378, 0.0141, 0.0262, 0.0550, 0.2384], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0108, 0.0113, 0.0114, 0.0091, 0.0121, 0.0103, 0.0107], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:13:48,798 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116105.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:14:06,606 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116116.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:14:27,137 INFO [train2.py:809] (0/4) Epoch 30, batch 600, loss[ctc_loss=0.08262, att_loss=0.2195, loss=0.1921, over 15646.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.008703, over 37.00 utterances.], tot_loss[ctc_loss=0.06214, att_loss=0.2289, loss=0.1956, over 3119665.83 frames. utt_duration=1275 frames, utt_pad_proportion=0.0452, over 9796.51 utterances.], batch size: 37, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:15:15,791 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.620e+01 1.813e+02 2.125e+02 2.957e+02 6.068e+02, threshold=4.250e+02, percent-clipped=9.0 2023-03-09 14:15:26,204 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116166.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:15:44,290 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116177.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:15:47,048 INFO [train2.py:809] (0/4) Epoch 30, batch 650, loss[ctc_loss=0.0449, att_loss=0.2094, loss=0.1765, over 15873.00 frames. utt_duration=1630 frames, utt_pad_proportion=0.009939, over 39.00 utterances.], tot_loss[ctc_loss=0.06167, att_loss=0.2288, loss=0.1954, over 3153870.71 frames. utt_duration=1260 frames, utt_pad_proportion=0.04984, over 10023.86 utterances.], batch size: 39, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:17:07,966 INFO [train2.py:809] (0/4) Epoch 30, batch 700, loss[ctc_loss=0.04388, att_loss=0.2108, loss=0.1774, over 14554.00 frames. utt_duration=1821 frames, utt_pad_proportion=0.03841, over 32.00 utterances.], tot_loss[ctc_loss=0.062, att_loss=0.2292, loss=0.1957, over 3174808.37 frames. utt_duration=1237 frames, utt_pad_proportion=0.05854, over 10277.54 utterances.], batch size: 32, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:17:38,373 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116248.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 14:17:57,485 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.158e+02 1.730e+02 1.988e+02 2.420e+02 4.405e+02, threshold=3.977e+02, percent-clipped=2.0 2023-03-09 14:18:28,327 INFO [train2.py:809] (0/4) Epoch 30, batch 750, loss[ctc_loss=0.06923, att_loss=0.2272, loss=0.1956, over 16125.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006495, over 42.00 utterances.], tot_loss[ctc_loss=0.0625, att_loss=0.2299, loss=0.1964, over 3200961.39 frames. utt_duration=1237 frames, utt_pad_proportion=0.05752, over 10364.69 utterances.], batch size: 42, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:19:37,956 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116323.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:19:47,388 INFO [train2.py:809] (0/4) Epoch 30, batch 800, loss[ctc_loss=0.05973, att_loss=0.2326, loss=0.198, over 16521.00 frames. utt_duration=1470 frames, utt_pad_proportion=0.007232, over 45.00 utterances.], tot_loss[ctc_loss=0.06259, att_loss=0.2299, loss=0.1964, over 3215455.01 frames. utt_duration=1247 frames, utt_pad_proportion=0.05587, over 10323.44 utterances.], batch size: 45, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:20:11,251 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4579, 2.9924, 3.6285, 3.1290, 3.5143, 4.5504, 4.4197, 3.2207], device='cuda:0'), covar=tensor([0.0422, 0.1770, 0.1349, 0.1331, 0.1167, 0.1120, 0.0624, 0.1377], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0254, 0.0295, 0.0222, 0.0274, 0.0385, 0.0276, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 14:20:30,374 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5704, 2.9749, 3.6920, 3.2740, 3.5647, 4.6309, 4.4876, 3.2423], device='cuda:0'), covar=tensor([0.0372, 0.1937, 0.1424, 0.1274, 0.1134, 0.0941, 0.0552, 0.1450], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0254, 0.0295, 0.0222, 0.0274, 0.0385, 0.0276, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 14:20:36,149 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.359e+02 1.849e+02 2.173e+02 2.485e+02 4.246e+02, threshold=4.345e+02, percent-clipped=3.0 2023-03-09 14:20:36,993 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.21 vs. limit=2.0 2023-03-09 14:20:52,673 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116370.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 14:21:06,894 INFO [train2.py:809] (0/4) Epoch 30, batch 850, loss[ctc_loss=0.06146, att_loss=0.2391, loss=0.2036, over 16963.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007874, over 50.00 utterances.], tot_loss[ctc_loss=0.06283, att_loss=0.2302, loss=0.1967, over 3234765.00 frames. utt_duration=1256 frames, utt_pad_proportion=0.0501, over 10310.26 utterances.], batch size: 50, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:21:15,093 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116384.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:22:27,521 INFO [train2.py:809] (0/4) Epoch 30, batch 900, loss[ctc_loss=0.04381, att_loss=0.2113, loss=0.1778, over 15950.00 frames. utt_duration=1558 frames, utt_pad_proportion=0.006533, over 41.00 utterances.], tot_loss[ctc_loss=0.06238, att_loss=0.2301, loss=0.1965, over 3247961.73 frames. utt_duration=1263 frames, utt_pad_proportion=0.04814, over 10300.10 utterances.], batch size: 41, lr: 3.59e-03, grad_scale: 8.0 2023-03-09 14:22:54,714 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2734, 3.9340, 3.3808, 3.5964, 4.1136, 3.8603, 3.2815, 4.3809], device='cuda:0'), covar=tensor([0.0951, 0.0455, 0.1070, 0.0750, 0.0770, 0.0668, 0.0812, 0.0460], device='cuda:0'), in_proj_covar=tensor([0.0212, 0.0230, 0.0234, 0.0211, 0.0294, 0.0253, 0.0208, 0.0302], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 14:22:54,723 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116446.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:23:16,725 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.211e+02 1.807e+02 2.134e+02 2.722e+02 5.477e+02, threshold=4.269e+02, percent-clipped=4.0 2023-03-09 14:23:18,522 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116461.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:23:36,248 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116472.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:23:47,732 INFO [train2.py:809] (0/4) Epoch 30, batch 950, loss[ctc_loss=0.06456, att_loss=0.2076, loss=0.179, over 15634.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009463, over 37.00 utterances.], tot_loss[ctc_loss=0.0622, att_loss=0.2293, loss=0.1959, over 3244106.76 frames. utt_duration=1269 frames, utt_pad_proportion=0.04991, over 10233.95 utterances.], batch size: 37, lr: 3.58e-03, grad_scale: 8.0 2023-03-09 14:24:29,865 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116505.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:24:33,189 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116507.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 14:25:07,881 INFO [train2.py:809] (0/4) Epoch 30, batch 1000, loss[ctc_loss=0.07136, att_loss=0.2473, loss=0.2121, over 17087.00 frames. utt_duration=1291 frames, utt_pad_proportion=0.007556, over 53.00 utterances.], tot_loss[ctc_loss=0.06161, att_loss=0.2293, loss=0.1958, over 3254069.85 frames. utt_duration=1274 frames, utt_pad_proportion=0.04779, over 10224.85 utterances.], batch size: 53, lr: 3.58e-03, grad_scale: 8.0 2023-03-09 14:25:38,083 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116548.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 14:25:56,767 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.166e+02 1.757e+02 2.029e+02 2.635e+02 6.922e+02, threshold=4.057e+02, percent-clipped=2.0 2023-03-09 14:26:07,201 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116566.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:26:21,831 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0800, 5.1419, 4.9344, 2.5387, 2.2428, 3.4381, 2.6266, 3.8820], device='cuda:0'), covar=tensor([0.0771, 0.0392, 0.0329, 0.5540, 0.4951, 0.2030, 0.4082, 0.1752], device='cuda:0'), in_proj_covar=tensor([0.0362, 0.0304, 0.0277, 0.0250, 0.0335, 0.0330, 0.0262, 0.0370], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 14:26:27,614 INFO [train2.py:809] (0/4) Epoch 30, batch 1050, loss[ctc_loss=0.05628, att_loss=0.2211, loss=0.1881, over 16309.00 frames. utt_duration=1519 frames, utt_pad_proportion=0.005544, over 43.00 utterances.], tot_loss[ctc_loss=0.06182, att_loss=0.2298, loss=0.1962, over 3260523.48 frames. utt_duration=1265 frames, utt_pad_proportion=0.04968, over 10321.34 utterances.], batch size: 43, lr: 3.58e-03, grad_scale: 8.0 2023-03-09 14:26:37,689 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1226, 5.3265, 5.2733, 5.3170, 5.3786, 5.3601, 4.9637, 4.8193], device='cuda:0'), covar=tensor([0.0994, 0.0522, 0.0326, 0.0433, 0.0267, 0.0301, 0.0460, 0.0338], device='cuda:0'), in_proj_covar=tensor([0.0540, 0.0387, 0.0382, 0.0387, 0.0452, 0.0456, 0.0385, 0.0424], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:26:54,734 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116596.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 14:27:30,171 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116618.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:27:47,630 INFO [train2.py:809] (0/4) Epoch 30, batch 1100, loss[ctc_loss=0.05307, att_loss=0.2274, loss=0.1925, over 16527.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006992, over 45.00 utterances.], tot_loss[ctc_loss=0.063, att_loss=0.2309, loss=0.1973, over 3268768.90 frames. utt_duration=1258 frames, utt_pad_proportion=0.05094, over 10407.69 utterances.], batch size: 45, lr: 3.58e-03, grad_scale: 8.0 2023-03-09 14:27:53,137 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9791, 6.1912, 5.7247, 5.8458, 5.8600, 5.2924, 5.6601, 5.3837], device='cuda:0'), covar=tensor([0.1202, 0.0862, 0.0767, 0.0950, 0.0939, 0.1636, 0.2147, 0.2142], device='cuda:0'), in_proj_covar=tensor([0.0568, 0.0641, 0.0495, 0.0477, 0.0458, 0.0486, 0.0649, 0.0545], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:28:34,268 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.8240, 5.0261, 4.9374, 5.0153, 5.0589, 5.0605, 4.6713, 4.5235], device='cuda:0'), covar=tensor([0.0897, 0.0584, 0.0407, 0.0469, 0.0315, 0.0341, 0.0470, 0.0373], device='cuda:0'), in_proj_covar=tensor([0.0539, 0.0385, 0.0381, 0.0386, 0.0451, 0.0454, 0.0384, 0.0422], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:28:37,009 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.262e+02 1.803e+02 2.274e+02 2.882e+02 6.074e+02, threshold=4.549e+02, percent-clipped=3.0 2023-03-09 14:28:54,126 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116670.0, num_to_drop=1, layers_to_drop={1} 2023-03-09 14:29:07,782 INFO [train2.py:809] (0/4) Epoch 30, batch 1150, loss[ctc_loss=0.05157, att_loss=0.232, loss=0.1959, over 16331.00 frames. utt_duration=1453 frames, utt_pad_proportion=0.006278, over 45.00 utterances.], tot_loss[ctc_loss=0.06376, att_loss=0.2305, loss=0.1971, over 3245951.24 frames. utt_duration=1243 frames, utt_pad_proportion=0.06002, over 10454.25 utterances.], batch size: 45, lr: 3.58e-03, grad_scale: 8.0 2023-03-09 14:29:08,010 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:29:08,180 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116679.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:29:15,841 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.4924, 2.2651, 2.4397, 2.6776, 2.7738, 2.3805, 2.2031, 2.7897], device='cuda:0'), covar=tensor([0.1757, 0.2317, 0.1989, 0.1016, 0.1821, 0.1442, 0.1854, 0.1154], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0153, 0.0149, 0.0145, 0.0161, 0.0140, 0.0163, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 14:29:47,256 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116704.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:30:09,471 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116718.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:30:26,869 INFO [train2.py:809] (0/4) Epoch 30, batch 1200, loss[ctc_loss=0.06008, att_loss=0.227, loss=0.1936, over 16124.00 frames. utt_duration=1537 frames, utt_pad_proportion=0.006172, over 42.00 utterances.], tot_loss[ctc_loss=0.06312, att_loss=0.2301, loss=0.1967, over 3246433.31 frames. utt_duration=1236 frames, utt_pad_proportion=0.06135, over 10515.99 utterances.], batch size: 42, lr: 3.58e-03, grad_scale: 8.0 2023-03-09 14:30:33,508 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1011, 5.0871, 4.8689, 3.0856, 4.9093, 4.7507, 4.5234, 2.9650], device='cuda:0'), covar=tensor([0.0129, 0.0105, 0.0266, 0.1000, 0.0109, 0.0215, 0.0278, 0.1253], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0109, 0.0114, 0.0115, 0.0092, 0.0122, 0.0103, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:31:12,888 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7730, 2.5226, 2.5562, 2.8870, 2.9514, 2.8700, 2.5110, 3.3086], device='cuda:0'), covar=tensor([0.1425, 0.2311, 0.1785, 0.0977, 0.1501, 0.1202, 0.1785, 0.0841], device='cuda:0'), in_proj_covar=tensor([0.0154, 0.0153, 0.0149, 0.0145, 0.0161, 0.0140, 0.0163, 0.0140], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 14:31:16,204 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.120e+02 1.678e+02 2.013e+02 2.421e+02 9.091e+02, threshold=4.026e+02, percent-clipped=5.0 2023-03-09 14:31:18,096 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116761.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:31:25,086 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116765.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:31:35,981 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116772.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:31:39,182 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.1369, 3.7911, 3.2482, 3.4475, 4.0495, 3.6349, 3.0878, 4.2663], device='cuda:0'), covar=tensor([0.1032, 0.0474, 0.1114, 0.0748, 0.0715, 0.0819, 0.0865, 0.0479], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0232, 0.0236, 0.0212, 0.0297, 0.0256, 0.0209, 0.0305], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 14:31:47,181 INFO [train2.py:809] (0/4) Epoch 30, batch 1250, loss[ctc_loss=0.07707, att_loss=0.2435, loss=0.2103, over 16624.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.004695, over 47.00 utterances.], tot_loss[ctc_loss=0.06269, att_loss=0.23, loss=0.1965, over 3258158.08 frames. utt_duration=1239 frames, utt_pad_proportion=0.05946, over 10533.39 utterances.], batch size: 47, lr: 3.58e-03, grad_scale: 16.0 2023-03-09 14:32:23,516 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116802.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 14:32:23,664 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=116802.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:32:35,242 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116809.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:32:53,096 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=116820.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:33:07,498 INFO [train2.py:809] (0/4) Epoch 30, batch 1300, loss[ctc_loss=0.05839, att_loss=0.2101, loss=0.1798, over 15634.00 frames. utt_duration=1692 frames, utt_pad_proportion=0.009025, over 37.00 utterances.], tot_loss[ctc_loss=0.06295, att_loss=0.2301, loss=0.1967, over 3262729.06 frames. utt_duration=1243 frames, utt_pad_proportion=0.05806, over 10514.69 utterances.], batch size: 37, lr: 3.58e-03, grad_scale: 16.0 2023-03-09 14:33:12,599 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6396, 4.9829, 4.8639, 4.9653, 5.0333, 4.6116, 3.2895, 4.9803], device='cuda:0'), covar=tensor([0.0127, 0.0102, 0.0141, 0.0065, 0.0093, 0.0139, 0.0773, 0.0178], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0093, 0.0120, 0.0074, 0.0081, 0.0092, 0.0107, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 14:33:57,067 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.212e+02 1.796e+02 2.133e+02 2.719e+02 8.820e+02, threshold=4.267e+02, percent-clipped=2.0 2023-03-09 14:33:58,857 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116861.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:34:02,718 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=116863.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:34:28,218 INFO [train2.py:809] (0/4) Epoch 30, batch 1350, loss[ctc_loss=0.04345, att_loss=0.2327, loss=0.1948, over 16867.00 frames. utt_duration=1379 frames, utt_pad_proportion=0.007508, over 49.00 utterances.], tot_loss[ctc_loss=0.06238, att_loss=0.2295, loss=0.1961, over 3262524.25 frames. utt_duration=1250 frames, utt_pad_proportion=0.05552, over 10455.80 utterances.], batch size: 49, lr: 3.58e-03, grad_scale: 16.0 2023-03-09 14:35:48,164 INFO [train2.py:809] (0/4) Epoch 30, batch 1400, loss[ctc_loss=0.07217, att_loss=0.2433, loss=0.2091, over 17368.00 frames. utt_duration=1008 frames, utt_pad_proportion=0.04606, over 69.00 utterances.], tot_loss[ctc_loss=0.06206, att_loss=0.2289, loss=0.1955, over 3256489.71 frames. utt_duration=1232 frames, utt_pad_proportion=0.06043, over 10587.84 utterances.], batch size: 69, lr: 3.58e-03, grad_scale: 16.0 2023-03-09 14:36:37,549 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.312e+02 1.819e+02 2.157e+02 2.707e+02 5.079e+02, threshold=4.315e+02, percent-clipped=2.0 2023-03-09 14:37:00,999 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=116974.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:37:08,667 INFO [train2.py:809] (0/4) Epoch 30, batch 1450, loss[ctc_loss=0.07193, att_loss=0.2514, loss=0.2155, over 17281.00 frames. utt_duration=1173 frames, utt_pad_proportion=0.02482, over 59.00 utterances.], tot_loss[ctc_loss=0.06256, att_loss=0.2296, loss=0.1962, over 3265395.36 frames. utt_duration=1262 frames, utt_pad_proportion=0.05176, over 10363.74 utterances.], batch size: 59, lr: 3.58e-03, grad_scale: 16.0 2023-03-09 14:37:08,997 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=116979.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:37:10,605 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7750, 2.3937, 2.6909, 2.7731, 2.9794, 2.8490, 2.5680, 3.2444], device='cuda:0'), covar=tensor([0.1410, 0.2176, 0.1517, 0.1084, 0.1513, 0.0999, 0.1508, 0.0798], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0154, 0.0151, 0.0147, 0.0163, 0.0141, 0.0164, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 14:37:39,293 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.32 vs. limit=2.0 2023-03-09 14:38:25,655 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117027.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:38:28,586 INFO [train2.py:809] (0/4) Epoch 30, batch 1500, loss[ctc_loss=0.04595, att_loss=0.2135, loss=0.18, over 16274.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.007041, over 43.00 utterances.], tot_loss[ctc_loss=0.06263, att_loss=0.2297, loss=0.1963, over 3261380.35 frames. utt_duration=1242 frames, utt_pad_proportion=0.05798, over 10512.69 utterances.], batch size: 43, lr: 3.58e-03, grad_scale: 16.0 2023-03-09 14:39:18,349 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.257e+02 1.806e+02 2.089e+02 2.584e+02 4.922e+02, threshold=4.178e+02, percent-clipped=1.0 2023-03-09 14:39:18,638 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117060.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:39:48,456 INFO [train2.py:809] (0/4) Epoch 30, batch 1550, loss[ctc_loss=0.06019, att_loss=0.2392, loss=0.2034, over 16976.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.006983, over 50.00 utterances.], tot_loss[ctc_loss=0.06206, att_loss=0.2301, loss=0.1965, over 3272206.49 frames. utt_duration=1271 frames, utt_pad_proportion=0.04909, over 10312.47 utterances.], batch size: 50, lr: 3.58e-03, grad_scale: 16.0 2023-03-09 14:40:25,222 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117102.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:41:08,938 INFO [train2.py:809] (0/4) Epoch 30, batch 1600, loss[ctc_loss=0.07281, att_loss=0.2398, loss=0.2064, over 16775.00 frames. utt_duration=686.1 frames, utt_pad_proportion=0.1402, over 98.00 utterances.], tot_loss[ctc_loss=0.06196, att_loss=0.2296, loss=0.1961, over 3259810.99 frames. utt_duration=1250 frames, utt_pad_proportion=0.05852, over 10444.31 utterances.], batch size: 98, lr: 3.57e-03, grad_scale: 16.0 2023-03-09 14:41:43,251 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117150.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:41:56,312 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117158.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:41:59,199 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.078e+02 1.817e+02 2.106e+02 2.742e+02 6.475e+02, threshold=4.211e+02, percent-clipped=3.0 2023-03-09 14:42:00,923 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117161.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:42:29,628 INFO [train2.py:809] (0/4) Epoch 30, batch 1650, loss[ctc_loss=0.04203, att_loss=0.1963, loss=0.1655, over 15521.00 frames. utt_duration=1726 frames, utt_pad_proportion=0.007587, over 36.00 utterances.], tot_loss[ctc_loss=0.06263, att_loss=0.2305, loss=0.1969, over 3262970.73 frames. utt_duration=1216 frames, utt_pad_proportion=0.06564, over 10749.16 utterances.], batch size: 36, lr: 3.57e-03, grad_scale: 16.0 2023-03-09 14:42:34,858 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4757, 2.9312, 4.8846, 3.8388, 3.2151, 4.2202, 4.7761, 4.6461], device='cuda:0'), covar=tensor([0.0300, 0.1384, 0.0268, 0.0892, 0.1549, 0.0276, 0.0215, 0.0297], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0248, 0.0235, 0.0324, 0.0272, 0.0247, 0.0229, 0.0249], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 14:42:39,605 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0558, 4.1735, 4.1932, 4.2131, 4.6888, 4.2504, 4.1061, 2.6100], device='cuda:0'), covar=tensor([0.0421, 0.0564, 0.0477, 0.0396, 0.0818, 0.0338, 0.0471, 0.1708], device='cuda:0'), in_proj_covar=tensor([0.0207, 0.0235, 0.0227, 0.0246, 0.0390, 0.0202, 0.0222, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 14:43:19,240 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117209.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:43:39,165 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117221.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:43:51,234 INFO [train2.py:809] (0/4) Epoch 30, batch 1700, loss[ctc_loss=0.06777, att_loss=0.2266, loss=0.1949, over 16106.00 frames. utt_duration=1535 frames, utt_pad_proportion=0.007449, over 42.00 utterances.], tot_loss[ctc_loss=0.06225, att_loss=0.2305, loss=0.1969, over 3270486.68 frames. utt_duration=1233 frames, utt_pad_proportion=0.05954, over 10626.90 utterances.], batch size: 42, lr: 3.57e-03, grad_scale: 16.0 2023-03-09 14:44:26,095 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1569, 5.4306, 5.3732, 5.3984, 5.4835, 5.4254, 5.1369, 4.9162], device='cuda:0'), covar=tensor([0.1087, 0.0541, 0.0313, 0.0477, 0.0292, 0.0327, 0.0415, 0.0301], device='cuda:0'), in_proj_covar=tensor([0.0547, 0.0390, 0.0388, 0.0392, 0.0455, 0.0461, 0.0390, 0.0425], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:44:43,556 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.065e+02 1.821e+02 2.138e+02 2.705e+02 3.860e+02, threshold=4.276e+02, percent-clipped=0.0 2023-03-09 14:45:04,965 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117274.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:45:12,033 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.42 vs. limit=2.0 2023-03-09 14:45:12,577 INFO [train2.py:809] (0/4) Epoch 30, batch 1750, loss[ctc_loss=0.06434, att_loss=0.2457, loss=0.2094, over 17324.00 frames. utt_duration=1176 frames, utt_pad_proportion=0.02172, over 59.00 utterances.], tot_loss[ctc_loss=0.06292, att_loss=0.2305, loss=0.197, over 3260423.69 frames. utt_duration=1221 frames, utt_pad_proportion=0.06385, over 10695.19 utterances.], batch size: 59, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:45:17,618 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117282.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:46:03,599 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7805, 5.1430, 5.0484, 5.1393, 5.2177, 4.8083, 3.6139, 5.1678], device='cuda:0'), covar=tensor([0.0116, 0.0099, 0.0124, 0.0068, 0.0087, 0.0121, 0.0642, 0.0144], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0093, 0.0120, 0.0074, 0.0081, 0.0092, 0.0107, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 14:46:23,159 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117322.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:46:29,176 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.78 vs. limit=2.0 2023-03-09 14:46:29,969 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0895, 5.1412, 4.9013, 3.3135, 4.9961, 4.9206, 4.4196, 2.9069], device='cuda:0'), covar=tensor([0.0157, 0.0104, 0.0313, 0.0879, 0.0104, 0.0170, 0.0317, 0.1313], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0109, 0.0114, 0.0115, 0.0092, 0.0121, 0.0103, 0.0106], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:46:34,466 INFO [train2.py:809] (0/4) Epoch 30, batch 1800, loss[ctc_loss=0.05511, att_loss=0.229, loss=0.1942, over 17011.00 frames. utt_duration=1336 frames, utt_pad_proportion=0.008094, over 51.00 utterances.], tot_loss[ctc_loss=0.06226, att_loss=0.2302, loss=0.1966, over 3262838.50 frames. utt_duration=1230 frames, utt_pad_proportion=0.06122, over 10620.85 utterances.], batch size: 51, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:46:34,853 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=117329.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:47:25,067 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117360.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:47:26,235 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.207e+02 1.761e+02 2.042e+02 2.458e+02 5.519e+02, threshold=4.084e+02, percent-clipped=2.0 2023-03-09 14:47:54,422 INFO [train2.py:809] (0/4) Epoch 30, batch 1850, loss[ctc_loss=0.03491, att_loss=0.2097, loss=0.1747, over 16284.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.00634, over 43.00 utterances.], tot_loss[ctc_loss=0.06217, att_loss=0.2304, loss=0.1968, over 3266613.78 frames. utt_duration=1222 frames, utt_pad_proportion=0.06293, over 10708.44 utterances.], batch size: 43, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:47:59,388 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5303, 2.9563, 3.6872, 2.9438, 3.6004, 4.6538, 4.4748, 3.1308], device='cuda:0'), covar=tensor([0.0386, 0.1770, 0.1143, 0.1367, 0.1062, 0.0870, 0.0535, 0.1408], device='cuda:0'), in_proj_covar=tensor([0.0254, 0.0254, 0.0295, 0.0221, 0.0276, 0.0387, 0.0278, 0.0241], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 14:48:12,245 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=117390.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:48:41,519 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117408.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:49:14,299 INFO [train2.py:809] (0/4) Epoch 30, batch 1900, loss[ctc_loss=0.05522, att_loss=0.23, loss=0.195, over 16967.00 frames. utt_duration=1359 frames, utt_pad_proportion=0.007524, over 50.00 utterances.], tot_loss[ctc_loss=0.06184, att_loss=0.2295, loss=0.196, over 3261283.19 frames. utt_duration=1238 frames, utt_pad_proportion=0.06189, over 10552.51 utterances.], batch size: 50, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:49:16,589 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-09 14:50:00,403 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117458.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:50:00,444 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6291, 3.3121, 3.7217, 3.1741, 3.6501, 4.7034, 4.5386, 3.4814], device='cuda:0'), covar=tensor([0.0346, 0.1423, 0.1289, 0.1244, 0.1112, 0.0841, 0.0554, 0.1117], device='cuda:0'), in_proj_covar=tensor([0.0251, 0.0252, 0.0293, 0.0218, 0.0273, 0.0383, 0.0275, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 14:50:04,885 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.123e+02 1.786e+02 2.185e+02 2.601e+02 6.892e+02, threshold=4.370e+02, percent-clipped=2.0 2023-03-09 14:50:33,484 INFO [train2.py:809] (0/4) Epoch 30, batch 1950, loss[ctc_loss=0.0678, att_loss=0.2449, loss=0.2095, over 17287.00 frames. utt_duration=1259 frames, utt_pad_proportion=0.01125, over 55.00 utterances.], tot_loss[ctc_loss=0.06235, att_loss=0.2298, loss=0.1963, over 3246653.11 frames. utt_duration=1240 frames, utt_pad_proportion=0.06391, over 10485.56 utterances.], batch size: 55, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:51:17,142 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117506.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:51:53,202 INFO [train2.py:809] (0/4) Epoch 30, batch 2000, loss[ctc_loss=0.07749, att_loss=0.254, loss=0.2187, over 16760.00 frames. utt_duration=685.5 frames, utt_pad_proportion=0.141, over 98.00 utterances.], tot_loss[ctc_loss=0.06229, att_loss=0.2298, loss=0.1963, over 3254992.79 frames. utt_duration=1248 frames, utt_pad_proportion=0.05922, over 10444.72 utterances.], batch size: 98, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:52:29,677 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5194, 2.7415, 4.8748, 3.8773, 3.1135, 4.1837, 4.7367, 4.6892], device='cuda:0'), covar=tensor([0.0322, 0.1451, 0.0366, 0.0907, 0.1626, 0.0306, 0.0251, 0.0323], device='cuda:0'), in_proj_covar=tensor([0.0247, 0.0252, 0.0240, 0.0330, 0.0277, 0.0252, 0.0234, 0.0254], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0003], device='cuda:0') 2023-03-09 14:52:43,010 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.243e+02 1.738e+02 2.282e+02 2.861e+02 7.444e+02, threshold=4.565e+02, percent-clipped=5.0 2023-03-09 14:53:08,585 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117577.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:53:11,458 INFO [train2.py:809] (0/4) Epoch 30, batch 2050, loss[ctc_loss=0.05623, att_loss=0.2321, loss=0.1969, over 16956.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007354, over 50.00 utterances.], tot_loss[ctc_loss=0.0623, att_loss=0.23, loss=0.1965, over 3252364.91 frames. utt_duration=1255 frames, utt_pad_proportion=0.05803, over 10382.34 utterances.], batch size: 50, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:54:32,206 INFO [train2.py:809] (0/4) Epoch 30, batch 2100, loss[ctc_loss=0.08287, att_loss=0.2164, loss=0.1897, over 15382.00 frames. utt_duration=1759 frames, utt_pad_proportion=0.009872, over 35.00 utterances.], tot_loss[ctc_loss=0.06272, att_loss=0.2303, loss=0.1968, over 3252108.43 frames. utt_duration=1239 frames, utt_pad_proportion=0.06169, over 10515.94 utterances.], batch size: 35, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:55:24,881 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.295e+02 1.716e+02 1.992e+02 2.490e+02 6.469e+02, threshold=3.983e+02, percent-clipped=1.0 2023-03-09 14:55:52,866 INFO [train2.py:809] (0/4) Epoch 30, batch 2150, loss[ctc_loss=0.1178, att_loss=0.2625, loss=0.2336, over 14706.00 frames. utt_duration=404.4 frames, utt_pad_proportion=0.2967, over 146.00 utterances.], tot_loss[ctc_loss=0.06275, att_loss=0.2303, loss=0.1968, over 3260380.08 frames. utt_duration=1239 frames, utt_pad_proportion=0.05973, over 10540.14 utterances.], batch size: 146, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:56:02,886 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=117685.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 14:56:40,206 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.2395, 3.7189, 3.2765, 3.4412, 3.9330, 3.6288, 2.9843, 4.2084], device='cuda:0'), covar=tensor([0.0941, 0.0524, 0.1075, 0.0733, 0.0801, 0.0781, 0.0942, 0.0563], device='cuda:0'), in_proj_covar=tensor([0.0213, 0.0231, 0.0236, 0.0212, 0.0296, 0.0256, 0.0209, 0.0303], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 14:57:00,149 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1201, 4.3391, 4.6267, 4.7393, 2.7697, 4.3975, 3.1543, 2.1247], device='cuda:0'), covar=tensor([0.0547, 0.0392, 0.0574, 0.0280, 0.1550, 0.0326, 0.1257, 0.1594], device='cuda:0'), in_proj_covar=tensor([0.0223, 0.0196, 0.0265, 0.0184, 0.0224, 0.0175, 0.0234, 0.0205], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 14:57:12,173 INFO [train2.py:809] (0/4) Epoch 30, batch 2200, loss[ctc_loss=0.07704, att_loss=0.2368, loss=0.2049, over 16410.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.005826, over 44.00 utterances.], tot_loss[ctc_loss=0.06229, att_loss=0.2299, loss=0.1964, over 3267621.20 frames. utt_duration=1259 frames, utt_pad_proportion=0.05276, over 10392.95 utterances.], batch size: 44, lr: 3.57e-03, grad_scale: 8.0 2023-03-09 14:58:03,079 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.299e+02 1.825e+02 2.078e+02 2.480e+02 5.999e+02, threshold=4.155e+02, percent-clipped=3.0 2023-03-09 14:58:30,570 INFO [train2.py:809] (0/4) Epoch 30, batch 2250, loss[ctc_loss=0.0356, att_loss=0.2068, loss=0.1726, over 15945.00 frames. utt_duration=1557 frames, utt_pad_proportion=0.006195, over 41.00 utterances.], tot_loss[ctc_loss=0.06184, att_loss=0.2296, loss=0.1961, over 3269588.36 frames. utt_duration=1287 frames, utt_pad_proportion=0.04548, over 10170.54 utterances.], batch size: 41, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 14:58:30,725 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9041, 6.1432, 5.5796, 5.8148, 5.7664, 5.2140, 5.6629, 5.3535], device='cuda:0'), covar=tensor([0.1292, 0.0920, 0.1153, 0.0907, 0.1128, 0.1656, 0.2277, 0.2319], device='cuda:0'), in_proj_covar=tensor([0.0566, 0.0644, 0.0496, 0.0482, 0.0460, 0.0489, 0.0650, 0.0550], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 14:59:37,757 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.87 vs. limit=2.0 2023-03-09 14:59:50,949 INFO [train2.py:809] (0/4) Epoch 30, batch 2300, loss[ctc_loss=0.06242, att_loss=0.2339, loss=0.1996, over 16759.00 frames. utt_duration=1398 frames, utt_pad_proportion=0.007147, over 48.00 utterances.], tot_loss[ctc_loss=0.062, att_loss=0.2297, loss=0.1962, over 3268641.82 frames. utt_duration=1272 frames, utt_pad_proportion=0.04927, over 10289.74 utterances.], batch size: 48, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 15:00:42,225 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.175e+02 1.803e+02 2.047e+02 2.475e+02 6.158e+02, threshold=4.093e+02, percent-clipped=3.0 2023-03-09 15:00:46,945 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.7748, 5.1428, 5.3753, 5.1301, 5.3633, 5.7495, 5.0858, 5.8366], device='cuda:0'), covar=tensor([0.0823, 0.0777, 0.0862, 0.1567, 0.1829, 0.0935, 0.0874, 0.0641], device='cuda:0'), in_proj_covar=tensor([0.0947, 0.0549, 0.0660, 0.0700, 0.0930, 0.0687, 0.0526, 0.0662], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 15:01:07,098 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117877.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:01:09,867 INFO [train2.py:809] (0/4) Epoch 30, batch 2350, loss[ctc_loss=0.05331, att_loss=0.2222, loss=0.1884, over 15967.00 frames. utt_duration=1559 frames, utt_pad_proportion=0.006187, over 41.00 utterances.], tot_loss[ctc_loss=0.0619, att_loss=0.2298, loss=0.1963, over 3277510.01 frames. utt_duration=1281 frames, utt_pad_proportion=0.04492, over 10246.86 utterances.], batch size: 41, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 15:01:54,308 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([2.7395, 2.4030, 2.5253, 2.7140, 3.0486, 2.8318, 2.3975, 2.9963], device='cuda:0'), covar=tensor([0.2100, 0.2370, 0.1687, 0.1226, 0.1466, 0.1281, 0.1782, 0.1029], device='cuda:0'), in_proj_covar=tensor([0.0156, 0.0153, 0.0150, 0.0147, 0.0162, 0.0141, 0.0162, 0.0141], device='cuda:0'), out_proj_covar=tensor([0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001], device='cuda:0') 2023-03-09 15:02:12,211 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.96 vs. limit=2.0 2023-03-09 15:02:23,391 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=117925.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:02:29,492 INFO [train2.py:809] (0/4) Epoch 30, batch 2400, loss[ctc_loss=0.06147, att_loss=0.239, loss=0.2035, over 16764.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.00579, over 48.00 utterances.], tot_loss[ctc_loss=0.06205, att_loss=0.2297, loss=0.1962, over 3278794.35 frames. utt_duration=1285 frames, utt_pad_proportion=0.04401, over 10222.14 utterances.], batch size: 48, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 15:03:20,930 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.310e+02 1.808e+02 2.132e+02 2.494e+02 4.997e+02, threshold=4.263e+02, percent-clipped=2.0 2023-03-09 15:03:48,709 INFO [train2.py:809] (0/4) Epoch 30, batch 2450, loss[ctc_loss=0.08942, att_loss=0.2568, loss=0.2233, over 17290.00 frames. utt_duration=1099 frames, utt_pad_proportion=0.03897, over 63.00 utterances.], tot_loss[ctc_loss=0.06283, att_loss=0.2303, loss=0.1968, over 3276784.69 frames. utt_duration=1271 frames, utt_pad_proportion=0.04801, over 10327.56 utterances.], batch size: 63, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 15:03:58,949 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=117985.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:04:23,461 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/checkpoint-118000.pt 2023-03-09 15:05:01,293 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9131, 5.1884, 5.1080, 5.1529, 5.2321, 5.1485, 4.8307, 4.6736], device='cuda:0'), covar=tensor([0.1078, 0.0504, 0.0344, 0.0428, 0.0264, 0.0352, 0.0455, 0.0335], device='cuda:0'), in_proj_covar=tensor([0.0541, 0.0383, 0.0382, 0.0385, 0.0446, 0.0453, 0.0386, 0.0421], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:05:13,630 INFO [train2.py:809] (0/4) Epoch 30, batch 2500, loss[ctc_loss=0.0918, att_loss=0.2592, loss=0.2258, over 17148.00 frames. utt_duration=1226 frames, utt_pad_proportion=0.01327, over 56.00 utterances.], tot_loss[ctc_loss=0.0627, att_loss=0.2305, loss=0.1969, over 3276593.77 frames. utt_duration=1275 frames, utt_pad_proportion=0.04635, over 10292.84 utterances.], batch size: 56, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 15:05:21,120 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118033.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:05:32,744 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118040.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:06:07,288 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 1.938e+02 2.187e+02 2.706e+02 8.579e+02, threshold=4.374e+02, percent-clipped=4.0 2023-03-09 15:06:35,075 INFO [train2.py:809] (0/4) Epoch 30, batch 2550, loss[ctc_loss=0.05054, att_loss=0.22, loss=0.1861, over 16539.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.005766, over 45.00 utterances.], tot_loss[ctc_loss=0.06248, att_loss=0.23, loss=0.1965, over 3274483.11 frames. utt_duration=1276 frames, utt_pad_proportion=0.04623, over 10279.37 utterances.], batch size: 45, lr: 3.56e-03, grad_scale: 4.0 2023-03-09 15:07:11,126 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118101.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:07:54,649 INFO [train2.py:809] (0/4) Epoch 30, batch 2600, loss[ctc_loss=0.0833, att_loss=0.2477, loss=0.2148, over 17300.00 frames. utt_duration=1260 frames, utt_pad_proportion=0.01182, over 55.00 utterances.], tot_loss[ctc_loss=0.06219, att_loss=0.23, loss=0.1965, over 3280709.19 frames. utt_duration=1281 frames, utt_pad_proportion=0.0441, over 10259.28 utterances.], batch size: 55, lr: 3.56e-03, grad_scale: 4.0 2023-03-09 15:08:46,943 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.196e+02 1.796e+02 2.196e+02 2.749e+02 4.358e+02, threshold=4.391e+02, percent-clipped=0.0 2023-03-09 15:08:56,500 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118168.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:09:14,092 INFO [train2.py:809] (0/4) Epoch 30, batch 2650, loss[ctc_loss=0.04568, att_loss=0.2089, loss=0.1762, over 16378.00 frames. utt_duration=1491 frames, utt_pad_proportion=0.007999, over 44.00 utterances.], tot_loss[ctc_loss=0.06271, att_loss=0.2297, loss=0.1963, over 3272763.38 frames. utt_duration=1276 frames, utt_pad_proportion=0.04736, over 10272.78 utterances.], batch size: 44, lr: 3.56e-03, grad_scale: 4.0 2023-03-09 15:09:43,388 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9415, 3.9719, 3.8127, 2.8309, 3.7254, 3.7877, 3.5569, 2.7573], device='cuda:0'), covar=tensor([0.0137, 0.0149, 0.0296, 0.0927, 0.0163, 0.0413, 0.0344, 0.1233], device='cuda:0'), in_proj_covar=tensor([0.0080, 0.0108, 0.0113, 0.0114, 0.0091, 0.0120, 0.0103, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:09:43,605 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.6038, 3.6760, 3.7725, 2.5380, 2.4224, 3.0159, 2.6223, 3.4917], device='cuda:0'), covar=tensor([0.0711, 0.0529, 0.0454, 0.3655, 0.3426, 0.1779, 0.2525, 0.1313], device='cuda:0'), in_proj_covar=tensor([0.0369, 0.0313, 0.0284, 0.0257, 0.0344, 0.0338, 0.0268, 0.0377], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 15:10:33,465 INFO [train2.py:809] (0/4) Epoch 30, batch 2700, loss[ctc_loss=0.05711, att_loss=0.2031, loss=0.1739, over 15601.00 frames. utt_duration=1688 frames, utt_pad_proportion=0.01039, over 37.00 utterances.], tot_loss[ctc_loss=0.06339, att_loss=0.2302, loss=0.1968, over 3267378.77 frames. utt_duration=1251 frames, utt_pad_proportion=0.05561, over 10462.55 utterances.], batch size: 37, lr: 3.56e-03, grad_scale: 4.0 2023-03-09 15:10:33,870 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118229.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 15:10:39,336 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.68 vs. limit=2.0 2023-03-09 15:11:25,944 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.097e+02 1.836e+02 2.266e+02 2.685e+02 4.437e+02, threshold=4.533e+02, percent-clipped=1.0 2023-03-09 15:11:37,220 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118269.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:11:52,836 INFO [train2.py:809] (0/4) Epoch 30, batch 2750, loss[ctc_loss=0.0451, att_loss=0.2201, loss=0.1851, over 16613.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006066, over 47.00 utterances.], tot_loss[ctc_loss=0.06367, att_loss=0.2304, loss=0.1971, over 3267089.86 frames. utt_duration=1220 frames, utt_pad_proportion=0.06225, over 10726.33 utterances.], batch size: 47, lr: 3.56e-03, grad_scale: 4.0 2023-03-09 15:12:10,137 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118289.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:12:23,931 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.4778, 3.1163, 3.4555, 4.6640, 4.0765, 4.1264, 3.1300, 2.5899], device='cuda:0'), covar=tensor([0.0811, 0.1945, 0.0913, 0.0525, 0.0851, 0.0473, 0.1416, 0.2042], device='cuda:0'), in_proj_covar=tensor([0.0193, 0.0222, 0.0187, 0.0230, 0.0242, 0.0197, 0.0205, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:12:26,643 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.45 vs. limit=2.0 2023-03-09 15:12:28,635 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.7868, 5.1840, 5.0237, 5.1353, 5.1955, 4.8810, 3.6720, 5.0964], device='cuda:0'), covar=tensor([0.0123, 0.0109, 0.0133, 0.0069, 0.0086, 0.0114, 0.0643, 0.0166], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0094, 0.0121, 0.0075, 0.0082, 0.0093, 0.0108, 0.0113], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 15:13:12,414 INFO [train2.py:809] (0/4) Epoch 30, batch 2800, loss[ctc_loss=0.09373, att_loss=0.2395, loss=0.2103, over 16835.00 frames. utt_duration=681.7 frames, utt_pad_proportion=0.1436, over 99.00 utterances.], tot_loss[ctc_loss=0.06382, att_loss=0.2301, loss=0.1969, over 3271599.69 frames. utt_duration=1233 frames, utt_pad_proportion=0.05797, over 10626.26 utterances.], batch size: 99, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 15:13:12,758 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118329.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:13:14,394 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118330.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:13:47,157 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118350.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:14:05,712 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.224e+02 1.819e+02 2.314e+02 2.830e+02 4.183e+02, threshold=4.629e+02, percent-clipped=0.0 2023-03-09 15:14:32,836 INFO [train2.py:809] (0/4) Epoch 30, batch 2850, loss[ctc_loss=0.05348, att_loss=0.2202, loss=0.1869, over 16003.00 frames. utt_duration=1602 frames, utt_pad_proportion=0.007652, over 40.00 utterances.], tot_loss[ctc_loss=0.06379, att_loss=0.2309, loss=0.1974, over 3283489.96 frames. utt_duration=1230 frames, utt_pad_proportion=0.05548, over 10694.55 utterances.], batch size: 40, lr: 3.56e-03, grad_scale: 8.0 2023-03-09 15:14:43,844 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.1149, 5.3819, 5.3640, 5.3259, 5.4086, 5.3809, 4.9977, 4.8140], device='cuda:0'), covar=tensor([0.0995, 0.0526, 0.0287, 0.0506, 0.0290, 0.0304, 0.0416, 0.0351], device='cuda:0'), in_proj_covar=tensor([0.0542, 0.0384, 0.0383, 0.0386, 0.0449, 0.0455, 0.0386, 0.0423], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:14:51,893 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118390.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:15:01,132 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118396.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:15:53,109 INFO [train2.py:809] (0/4) Epoch 30, batch 2900, loss[ctc_loss=0.06943, att_loss=0.2195, loss=0.1895, over 15375.00 frames. utt_duration=1758 frames, utt_pad_proportion=0.01099, over 35.00 utterances.], tot_loss[ctc_loss=0.0631, att_loss=0.2303, loss=0.1968, over 3278647.41 frames. utt_duration=1244 frames, utt_pad_proportion=0.05296, over 10553.95 utterances.], batch size: 35, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:16:01,978 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3188, 3.9208, 3.4292, 3.6341, 4.1104, 3.8533, 3.2542, 4.4147], device='cuda:0'), covar=tensor([0.0987, 0.0557, 0.1122, 0.0693, 0.0778, 0.0709, 0.0835, 0.0540], device='cuda:0'), in_proj_covar=tensor([0.0214, 0.0233, 0.0236, 0.0213, 0.0296, 0.0256, 0.0209, 0.0304], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004, 0.0003, 0.0005], device='cuda:0') 2023-03-09 15:16:46,257 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.238e+02 1.677e+02 2.045e+02 2.519e+02 5.925e+02, threshold=4.089e+02, percent-clipped=1.0 2023-03-09 15:17:12,897 INFO [train2.py:809] (0/4) Epoch 30, batch 2950, loss[ctc_loss=0.0461, att_loss=0.2161, loss=0.1821, over 16401.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.006759, over 44.00 utterances.], tot_loss[ctc_loss=0.06268, att_loss=0.2301, loss=0.1966, over 3279783.36 frames. utt_duration=1237 frames, utt_pad_proportion=0.05452, over 10620.43 utterances.], batch size: 44, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:17:28,477 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3381, 2.3192, 3.6190, 2.5609, 3.4421, 4.5885, 4.5110, 2.7606], device='cuda:0'), covar=tensor([0.0523, 0.2503, 0.0993, 0.1971, 0.0975, 0.0671, 0.0489, 0.1823], device='cuda:0'), in_proj_covar=tensor([0.0252, 0.0252, 0.0293, 0.0220, 0.0274, 0.0385, 0.0276, 0.0238], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 15:18:23,846 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118524.0, num_to_drop=1, layers_to_drop={3} 2023-03-09 15:18:31,910 INFO [train2.py:809] (0/4) Epoch 30, batch 3000, loss[ctc_loss=0.05016, att_loss=0.2234, loss=0.1887, over 16405.00 frames. utt_duration=1493 frames, utt_pad_proportion=0.007359, over 44.00 utterances.], tot_loss[ctc_loss=0.06267, att_loss=0.2303, loss=0.1968, over 3282342.21 frames. utt_duration=1269 frames, utt_pad_proportion=0.04665, over 10356.94 utterances.], batch size: 44, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:18:31,912 INFO [train2.py:834] (0/4) Computing validation loss 2023-03-09 15:18:46,181 INFO [train2.py:843] (0/4) Epoch 30, validation: ctc_loss=0.04128, att_loss=0.2346, loss=0.196, over 944034.00 frames. utt_duration=679.8 frames, utt_pad_proportion=0.1349, over 5567.00 utterances. 2023-03-09 15:18:46,182 INFO [train2.py:844] (0/4) Maximum memory allocated so far is 16157MB 2023-03-09 15:18:52,048 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.39 vs. limit=2.0 2023-03-09 15:18:58,546 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.30 vs. limit=5.0 2023-03-09 15:19:09,439 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.2107, 5.4287, 5.4039, 5.3772, 5.4610, 5.4282, 5.0397, 4.8228], device='cuda:0'), covar=tensor([0.0963, 0.0518, 0.0269, 0.0482, 0.0301, 0.0319, 0.0436, 0.0350], device='cuda:0'), in_proj_covar=tensor([0.0538, 0.0383, 0.0382, 0.0385, 0.0448, 0.0454, 0.0384, 0.0421], device='cuda:0'), out_proj_covar=tensor([0.0005, 0.0004, 0.0004, 0.0004, 0.0005, 0.0005, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:19:15,760 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.5379, 5.7714, 5.2223, 5.4437, 5.3868, 4.9092, 5.2641, 4.9413], device='cuda:0'), covar=tensor([0.1324, 0.0930, 0.1104, 0.1011, 0.1105, 0.1554, 0.2237, 0.2369], device='cuda:0'), in_proj_covar=tensor([0.0569, 0.0645, 0.0494, 0.0484, 0.0460, 0.0486, 0.0649, 0.0551], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:19:38,943 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.222e+02 1.708e+02 2.016e+02 2.374e+02 4.392e+02, threshold=4.031e+02, percent-clipped=1.0 2023-03-09 15:20:05,704 INFO [train2.py:809] (0/4) Epoch 30, batch 3050, loss[ctc_loss=0.05829, att_loss=0.2419, loss=0.2052, over 17298.00 frames. utt_duration=1174 frames, utt_pad_proportion=0.02461, over 59.00 utterances.], tot_loss[ctc_loss=0.06233, att_loss=0.23, loss=0.1965, over 3272912.42 frames. utt_duration=1266 frames, utt_pad_proportion=0.04969, over 10350.68 utterances.], batch size: 59, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:21:20,125 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118625.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:21:25,500 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.9704, 5.0005, 4.6612, 2.9627, 4.7697, 4.6956, 4.1874, 2.6359], device='cuda:0'), covar=tensor([0.0134, 0.0138, 0.0336, 0.1170, 0.0146, 0.0221, 0.0408, 0.1650], device='cuda:0'), in_proj_covar=tensor([0.0079, 0.0108, 0.0113, 0.0114, 0.0091, 0.0120, 0.0102, 0.0105], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0004, 0.0004, 0.0004, 0.0003, 0.0004, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:21:26,775 INFO [train2.py:809] (0/4) Epoch 30, batch 3100, loss[ctc_loss=0.05373, att_loss=0.2351, loss=0.1988, over 16962.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007757, over 50.00 utterances.], tot_loss[ctc_loss=0.06251, att_loss=0.2304, loss=0.1968, over 3284361.22 frames. utt_duration=1261 frames, utt_pad_proportion=0.04787, over 10430.16 utterances.], batch size: 50, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:21:30,247 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=118631.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:21:53,101 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118645.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:22:19,882 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.328e+02 1.896e+02 2.204e+02 2.593e+02 6.390e+02, threshold=4.408e+02, percent-clipped=1.0 2023-03-09 15:22:47,069 INFO [train2.py:809] (0/4) Epoch 30, batch 3150, loss[ctc_loss=0.04637, att_loss=0.2019, loss=0.1708, over 15490.00 frames. utt_duration=1723 frames, utt_pad_proportion=0.008902, over 36.00 utterances.], tot_loss[ctc_loss=0.06243, att_loss=0.2301, loss=0.1966, over 3281744.60 frames. utt_duration=1263 frames, utt_pad_proportion=0.048, over 10407.70 utterances.], batch size: 36, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:22:56,697 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118685.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:23:08,836 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=118692.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:23:14,670 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118696.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:23:36,826 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.0480, 5.3778, 4.9701, 5.4560, 4.8474, 5.1144, 5.5198, 5.3259], device='cuda:0'), covar=tensor([0.0581, 0.0302, 0.0783, 0.0329, 0.0377, 0.0245, 0.0243, 0.0185], device='cuda:0'), in_proj_covar=tensor([0.0408, 0.0344, 0.0384, 0.0388, 0.0342, 0.0248, 0.0324, 0.0307], device='cuda:0'), out_proj_covar=tensor([0.0006, 0.0006, 0.0006, 0.0006, 0.0005, 0.0004, 0.0005, 0.0005], device='cuda:0') 2023-03-09 15:23:50,252 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.31 vs. limit=5.0 2023-03-09 15:24:06,684 INFO [train2.py:809] (0/4) Epoch 30, batch 3200, loss[ctc_loss=0.06739, att_loss=0.2303, loss=0.1977, over 16535.00 frames. utt_duration=1471 frames, utt_pad_proportion=0.006752, over 45.00 utterances.], tot_loss[ctc_loss=0.06181, att_loss=0.2294, loss=0.1959, over 3275768.97 frames. utt_duration=1268 frames, utt_pad_proportion=0.04892, over 10346.14 utterances.], batch size: 45, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:24:26,576 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2517, 2.9028, 3.5194, 2.8877, 3.4641, 4.3147, 4.2124, 3.1321], device='cuda:0'), covar=tensor([0.0368, 0.1692, 0.1240, 0.1428, 0.1100, 0.0934, 0.0650, 0.1320], device='cuda:0'), in_proj_covar=tensor([0.0250, 0.0250, 0.0292, 0.0220, 0.0273, 0.0384, 0.0276, 0.0237], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003], device='cuda:0') 2023-03-09 15:24:28,213 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4487, 2.5421, 4.9158, 3.8490, 2.9628, 4.1306, 4.5990, 4.5913], device='cuda:0'), covar=tensor([0.0293, 0.1602, 0.0190, 0.0871, 0.1733, 0.0301, 0.0215, 0.0251], device='cuda:0'), in_proj_covar=tensor([0.0244, 0.0248, 0.0237, 0.0328, 0.0274, 0.0249, 0.0232, 0.0251], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:24:30,970 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118744.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:24:59,442 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.119e+02 1.732e+02 2.054e+02 2.597e+02 6.136e+02, threshold=4.108e+02, percent-clipped=5.0 2023-03-09 15:25:14,394 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-03-09 15:25:26,532 INFO [train2.py:809] (0/4) Epoch 30, batch 3250, loss[ctc_loss=0.0464, att_loss=0.1967, loss=0.1667, over 15622.00 frames. utt_duration=1691 frames, utt_pad_proportion=0.009595, over 37.00 utterances.], tot_loss[ctc_loss=0.06197, att_loss=0.2294, loss=0.1959, over 3276664.19 frames. utt_duration=1281 frames, utt_pad_proportion=0.04589, over 10242.89 utterances.], batch size: 37, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:25:41,908 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.31 vs. limit=2.0 2023-03-09 15:26:33,175 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=3.64 vs. limit=5.0 2023-03-09 15:26:39,457 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118824.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:26:46,843 INFO [train2.py:809] (0/4) Epoch 30, batch 3300, loss[ctc_loss=0.04443, att_loss=0.2028, loss=0.1711, over 15505.00 frames. utt_duration=1724 frames, utt_pad_proportion=0.007959, over 36.00 utterances.], tot_loss[ctc_loss=0.06233, att_loss=0.2295, loss=0.1961, over 3271778.48 frames. utt_duration=1247 frames, utt_pad_proportion=0.05625, over 10504.76 utterances.], batch size: 36, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:27:39,412 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.286e+02 1.786e+02 1.999e+02 2.429e+02 6.941e+02, threshold=3.998e+02, percent-clipped=3.0 2023-03-09 15:27:55,320 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118872.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:28:06,049 INFO [train2.py:809] (0/4) Epoch 30, batch 3350, loss[ctc_loss=0.06847, att_loss=0.249, loss=0.2129, over 16776.00 frames. utt_duration=1399 frames, utt_pad_proportion=0.005405, over 48.00 utterances.], tot_loss[ctc_loss=0.06257, att_loss=0.2289, loss=0.1957, over 3259216.47 frames. utt_duration=1221 frames, utt_pad_proportion=0.06616, over 10691.78 utterances.], batch size: 48, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:28:34,636 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.5764, 4.9255, 4.8055, 4.9112, 4.9992, 4.7073, 3.2568, 4.9468], device='cuda:0'), covar=tensor([0.0139, 0.0135, 0.0153, 0.0098, 0.0112, 0.0125, 0.0824, 0.0181], device='cuda:0'), in_proj_covar=tensor([0.0098, 0.0094, 0.0121, 0.0075, 0.0081, 0.0092, 0.0108, 0.0112], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0004, 0.0002, 0.0002, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 15:28:55,097 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.3539, 4.5277, 4.6369, 4.6637, 5.2889, 4.5623, 4.5830, 2.9937], device='cuda:0'), covar=tensor([0.0382, 0.0479, 0.0412, 0.0402, 0.0617, 0.0307, 0.0416, 0.1500], device='cuda:0'), in_proj_covar=tensor([0.0208, 0.0237, 0.0231, 0.0248, 0.0390, 0.0204, 0.0223, 0.0225], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:29:20,568 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118925.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:29:26,691 INFO [train2.py:809] (0/4) Epoch 30, batch 3400, loss[ctc_loss=0.06994, att_loss=0.2432, loss=0.2086, over 16610.00 frames. utt_duration=1415 frames, utt_pad_proportion=0.006201, over 47.00 utterances.], tot_loss[ctc_loss=0.06191, att_loss=0.2286, loss=0.1952, over 3259685.12 frames. utt_duration=1249 frames, utt_pad_proportion=0.06009, over 10455.24 utterances.], batch size: 47, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:29:52,904 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118945.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:29:54,901 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.6823, 4.8900, 4.4304, 4.7422, 4.5111, 4.0814, 4.4051, 4.1800], device='cuda:0'), covar=tensor([0.1336, 0.1338, 0.1074, 0.1099, 0.1469, 0.1733, 0.2444, 0.2586], device='cuda:0'), in_proj_covar=tensor([0.0561, 0.0641, 0.0492, 0.0479, 0.0457, 0.0480, 0.0646, 0.0547], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:30:16,622 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.2899, 2.5578, 4.6494, 3.7701, 2.9347, 4.0121, 4.3051, 4.3445], device='cuda:0'), covar=tensor([0.0274, 0.1456, 0.0200, 0.0789, 0.1550, 0.0315, 0.0252, 0.0287], device='cuda:0'), in_proj_covar=tensor([0.0243, 0.0247, 0.0236, 0.0326, 0.0273, 0.0248, 0.0232, 0.0250], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:30:19,113 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.48 vs. limit=2.0 2023-03-09 15:30:19,288 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.021e+02 1.939e+02 2.194e+02 2.759e+02 6.591e+02, threshold=4.388e+02, percent-clipped=2.0 2023-03-09 15:30:30,393 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.07 vs. limit=5.0 2023-03-09 15:30:37,158 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118973.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:30:46,806 INFO [train2.py:809] (0/4) Epoch 30, batch 3450, loss[ctc_loss=0.06095, att_loss=0.2351, loss=0.2003, over 16625.00 frames. utt_duration=1416 frames, utt_pad_proportion=0.005484, over 47.00 utterances.], tot_loss[ctc_loss=0.06265, att_loss=0.2295, loss=0.1962, over 3260830.33 frames. utt_duration=1225 frames, utt_pad_proportion=0.06405, over 10662.11 utterances.], batch size: 47, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:30:57,732 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=118985.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:31:00,751 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=118987.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:31:09,770 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=118993.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:32:06,374 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.1994, 5.1084, 4.9293, 2.5309, 2.1745, 3.4220, 2.4647, 4.0189], device='cuda:0'), covar=tensor([0.0684, 0.0482, 0.0362, 0.5274, 0.5214, 0.1996, 0.4066, 0.1584], device='cuda:0'), in_proj_covar=tensor([0.0366, 0.0312, 0.0283, 0.0256, 0.0342, 0.0336, 0.0268, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 15:32:07,442 INFO [train2.py:809] (0/4) Epoch 30, batch 3500, loss[ctc_loss=0.04958, att_loss=0.2109, loss=0.1786, over 14545.00 frames. utt_duration=1819 frames, utt_pad_proportion=0.03937, over 32.00 utterances.], tot_loss[ctc_loss=0.0632, att_loss=0.2298, loss=0.1965, over 3260585.66 frames. utt_duration=1212 frames, utt_pad_proportion=0.06704, over 10777.41 utterances.], batch size: 32, lr: 3.55e-03, grad_scale: 8.0 2023-03-09 15:32:15,048 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119033.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:32:48,180 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.9567, 4.9941, 4.7888, 2.3384, 2.1516, 3.0420, 2.2590, 3.8902], device='cuda:0'), covar=tensor([0.0798, 0.0325, 0.0337, 0.5220, 0.5296, 0.2388, 0.4315, 0.1563], device='cuda:0'), in_proj_covar=tensor([0.0366, 0.0312, 0.0283, 0.0256, 0.0342, 0.0336, 0.0267, 0.0375], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0001, 0.0002], device='cuda:0') 2023-03-09 15:32:59,920 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.231e+02 1.837e+02 2.167e+02 2.597e+02 5.563e+02, threshold=4.335e+02, percent-clipped=3.0 2023-03-09 15:33:20,390 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119075.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:33:27,091 INFO [train2.py:809] (0/4) Epoch 30, batch 3550, loss[ctc_loss=0.05683, att_loss=0.2181, loss=0.1858, over 16276.00 frames. utt_duration=1516 frames, utt_pad_proportion=0.006873, over 43.00 utterances.], tot_loss[ctc_loss=0.06306, att_loss=0.2297, loss=0.1964, over 3263895.00 frames. utt_duration=1235 frames, utt_pad_proportion=0.06072, over 10583.79 utterances.], batch size: 43, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:33:50,469 INFO [zipformer.py:625] (0/4) warmup_begin=2666.7, warmup_end=3333.3, batch_count=119093.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:34:46,591 INFO [train2.py:809] (0/4) Epoch 30, batch 3600, loss[ctc_loss=0.06096, att_loss=0.2447, loss=0.208, over 17364.00 frames. utt_duration=1179 frames, utt_pad_proportion=0.02083, over 59.00 utterances.], tot_loss[ctc_loss=0.06241, att_loss=0.2296, loss=0.1962, over 3266920.76 frames. utt_duration=1264 frames, utt_pad_proportion=0.05289, over 10346.70 utterances.], batch size: 59, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:34:51,469 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.0868, 4.2745, 4.2115, 4.6493, 2.9069, 4.3022, 2.8573, 1.6508], device='cuda:0'), covar=tensor([0.0558, 0.0334, 0.0701, 0.0263, 0.1471, 0.0305, 0.1381, 0.1820], device='cuda:0'), in_proj_covar=tensor([0.0226, 0.0199, 0.0266, 0.0187, 0.0226, 0.0178, 0.0236, 0.0206], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0003, 0.0002, 0.0002, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:34:57,966 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119136.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:35:02,999 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=96, metric=1.30 vs. limit=2.0 2023-03-09 15:35:26,382 INFO [zipformer.py:625] (0/4) warmup_begin=3333.3, warmup_end=4000.0, batch_count=119154.0, num_to_drop=1, layers_to_drop={2} 2023-03-09 15:35:38,265 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.140e+02 1.787e+02 2.179e+02 2.834e+02 1.118e+03, threshold=4.357e+02, percent-clipped=5.0 2023-03-09 15:36:05,602 INFO [train2.py:809] (0/4) Epoch 30, batch 3650, loss[ctc_loss=0.06577, att_loss=0.242, loss=0.2068, over 17402.00 frames. utt_duration=1107 frames, utt_pad_proportion=0.03184, over 63.00 utterances.], tot_loss[ctc_loss=0.06247, att_loss=0.2292, loss=0.1958, over 3269568.92 frames. utt_duration=1280 frames, utt_pad_proportion=0.04861, over 10227.35 utterances.], batch size: 63, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:37:24,890 INFO [train2.py:809] (0/4) Epoch 30, batch 3700, loss[ctc_loss=0.07345, att_loss=0.2095, loss=0.1823, over 15646.00 frames. utt_duration=1693 frames, utt_pad_proportion=0.007636, over 37.00 utterances.], tot_loss[ctc_loss=0.06266, att_loss=0.2288, loss=0.1956, over 3256301.09 frames. utt_duration=1282 frames, utt_pad_proportion=0.05087, over 10169.81 utterances.], batch size: 37, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:38:13,219 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.0127, 3.7402, 3.7669, 3.2593, 3.6624, 3.7666, 3.7758, 2.8337], device='cuda:0'), covar=tensor([0.0988, 0.1177, 0.1441, 0.2680, 0.1602, 0.2442, 0.0910, 0.3009], device='cuda:0'), in_proj_covar=tensor([0.0216, 0.0221, 0.0234, 0.0283, 0.0198, 0.0295, 0.0218, 0.0237], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0002, 0.0003, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:38:18,046 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 9.417e+01 1.774e+02 2.178e+02 2.630e+02 5.010e+02, threshold=4.357e+02, percent-clipped=3.0 2023-03-09 15:38:32,596 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3521, 2.9673, 3.4892, 4.5804, 4.0896, 4.0729, 3.0752, 2.2453], device='cuda:0'), covar=tensor([0.0882, 0.1999, 0.0940, 0.0545, 0.0837, 0.0562, 0.1427, 0.2395], device='cuda:0'), in_proj_covar=tensor([0.0194, 0.0223, 0.0188, 0.0232, 0.0243, 0.0198, 0.0206, 0.0195], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0003, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:38:34,077 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([5.9002, 5.2373, 5.5323, 5.2853, 5.4064, 5.8881, 5.2449, 5.9878], device='cuda:0'), covar=tensor([0.0794, 0.0789, 0.0772, 0.1372, 0.1890, 0.0954, 0.0729, 0.0632], device='cuda:0'), in_proj_covar=tensor([0.0926, 0.0540, 0.0652, 0.0689, 0.0915, 0.0678, 0.0516, 0.0654], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0002, 0.0003, 0.0003, 0.0004, 0.0003, 0.0003, 0.0003], device='cuda:0') 2023-03-09 15:38:45,451 INFO [train2.py:809] (0/4) Epoch 30, batch 3750, loss[ctc_loss=0.06181, att_loss=0.2282, loss=0.1949, over 16274.00 frames. utt_duration=1515 frames, utt_pad_proportion=0.006995, over 43.00 utterances.], tot_loss[ctc_loss=0.06268, att_loss=0.2296, loss=0.1962, over 3266438.12 frames. utt_duration=1255 frames, utt_pad_proportion=0.05545, over 10424.41 utterances.], batch size: 43, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:38:50,205 INFO [scaling.py:679] (0/4) Whitening: num_groups=8, num_channels=192, metric=1.91 vs. limit=2.0 2023-03-09 15:38:59,218 INFO [zipformer.py:625] (0/4) warmup_begin=2000.0, warmup_end=2666.7, batch_count=119287.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:40:05,290 INFO [train2.py:809] (0/4) Epoch 30, batch 3800, loss[ctc_loss=0.05414, att_loss=0.2149, loss=0.1828, over 15351.00 frames. utt_duration=1756 frames, utt_pad_proportion=0.01204, over 35.00 utterances.], tot_loss[ctc_loss=0.06246, att_loss=0.2299, loss=0.1964, over 3272162.53 frames. utt_duration=1278 frames, utt_pad_proportion=0.04763, over 10250.18 utterances.], batch size: 35, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:40:15,894 INFO [zipformer.py:625] (0/4) warmup_begin=666.7, warmup_end=1333.3, batch_count=119335.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:40:58,012 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.145e+02 1.731e+02 2.132e+02 2.565e+02 6.876e+02, threshold=4.265e+02, percent-clipped=4.0 2023-03-09 15:41:24,908 INFO [train2.py:809] (0/4) Epoch 30, batch 3850, loss[ctc_loss=0.06276, att_loss=0.223, loss=0.1909, over 16485.00 frames. utt_duration=1435 frames, utt_pad_proportion=0.005514, over 46.00 utterances.], tot_loss[ctc_loss=0.06257, att_loss=0.2311, loss=0.1974, over 3286041.57 frames. utt_duration=1266 frames, utt_pad_proportion=0.04643, over 10395.06 utterances.], batch size: 46, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:41:26,338 INFO [scaling.py:679] (0/4) Whitening: num_groups=1, num_channels=384, metric=4.29 vs. limit=5.0 2023-03-09 15:42:43,264 INFO [train2.py:809] (0/4) Epoch 30, batch 3900, loss[ctc_loss=0.05303, att_loss=0.2101, loss=0.1787, over 15763.00 frames. utt_duration=1661 frames, utt_pad_proportion=0.009044, over 38.00 utterances.], tot_loss[ctc_loss=0.06279, att_loss=0.2307, loss=0.1972, over 3277295.91 frames. utt_duration=1252 frames, utt_pad_proportion=0.05235, over 10482.05 utterances.], batch size: 38, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:42:46,454 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119431.0, num_to_drop=0, layers_to_drop=set() 2023-03-09 15:43:14,054 INFO [zipformer.py:625] (0/4) warmup_begin=1333.3, warmup_end=2000.0, batch_count=119449.0, num_to_drop=1, layers_to_drop={0} 2023-03-09 15:43:33,706 INFO [optim.py:369] (0/4) Clipping_scale=2.0, grad-norm quartiles 1.098e+02 1.815e+02 2.229e+02 2.704e+02 7.416e+02, threshold=4.458e+02, percent-clipped=4.0 2023-03-09 15:43:50,630 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([6.0124, 6.2563, 5.6874, 5.9071, 5.9397, 5.4361, 5.7428, 5.3282], device='cuda:0'), covar=tensor([0.1438, 0.0885, 0.1134, 0.0860, 0.0951, 0.1515, 0.2197, 0.2158], device='cuda:0'), in_proj_covar=tensor([0.0567, 0.0650, 0.0500, 0.0484, 0.0459, 0.0483, 0.0655, 0.0553], device='cuda:0'), out_proj_covar=tensor([0.0004, 0.0004, 0.0003, 0.0003, 0.0003, 0.0003, 0.0004, 0.0004], device='cuda:0') 2023-03-09 15:43:58,539 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([4.4010, 2.6964, 4.7944, 3.7901, 3.0287, 4.1322, 4.5682, 4.5368], device='cuda:0'), covar=tensor([0.0308, 0.1497, 0.0241, 0.0865, 0.1559, 0.0306, 0.0293, 0.0283], device='cuda:0'), in_proj_covar=tensor([0.0242, 0.0245, 0.0236, 0.0325, 0.0271, 0.0248, 0.0232, 0.0250], device='cuda:0'), out_proj_covar=tensor([0.0002, 0.0002, 0.0002, 0.0003, 0.0003, 0.0002, 0.0002, 0.0002], device='cuda:0') 2023-03-09 15:43:59,618 INFO [train2.py:809] (0/4) Epoch 30, batch 3950, loss[ctc_loss=0.06791, att_loss=0.2372, loss=0.2034, over 16961.00 frames. utt_duration=1358 frames, utt_pad_proportion=0.007787, over 50.00 utterances.], tot_loss[ctc_loss=0.06365, att_loss=0.2311, loss=0.1976, over 3280038.98 frames. utt_duration=1263 frames, utt_pad_proportion=0.04963, over 10403.43 utterances.], batch size: 50, lr: 3.54e-03, grad_scale: 8.0 2023-03-09 15:44:40,455 INFO [zipformer.py:1447] (0/4) attn_weights_entropy = tensor([3.3002, 2.5122, 3.0734, 2.6094, 3.0884, 3.4076, 3.3341, 2.6792], device='cuda:0'), covar=tensor([0.0508, 0.1553, 0.1148, 0.1102, 0.0935, 0.1214, 0.0763, 0.1259], device='cuda:0'), in_proj_covar=tensor([0.0253, 0.0252, 0.0297, 0.0220, 0.0277, 0.0388, 0.0277, 0.0240], device='cuda:0'), out_proj_covar=tensor([0.0003, 0.0003, 0.0003, 0.0003, 0.0003, 0.0005, 0.0003, 0.0003], device='cuda:0') 2023-03-09 15:44:52,251 INFO [checkpoint.py:75] (0/4) Saving checkpoint to zipformer_ctc_att/exp/v0/epoch-30.pt 2023-03-09 15:44:55,863 INFO [train2.py:1037] (0/4) Done!