| { | |
| "best_metric": 0.1458934662818158, | |
| "best_model_checkpoint": "results15\\checkpoint-64000", | |
| "epoch": 1.151192, | |
| "eval_steps": 4000, | |
| "global_step": 68000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0008, | |
| "grad_norm": 12.430644989013672, | |
| "learning_rate": 1.8800000000000002e-06, | |
| "loss": 1.5617, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0016, | |
| "grad_norm": 11.342061042785645, | |
| "learning_rate": 3.88e-06, | |
| "loss": 0.7924, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0024, | |
| "grad_norm": 6.638418674468994, | |
| "learning_rate": 5.8800000000000005e-06, | |
| "loss": 0.3435, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.0032, | |
| "grad_norm": 19.131393432617188, | |
| "learning_rate": 7.88e-06, | |
| "loss": 0.2877, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.004, | |
| "grad_norm": 11.609911918640137, | |
| "learning_rate": 9.88e-06, | |
| "loss": 0.3076, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.0048, | |
| "grad_norm": 8.114494323730469, | |
| "learning_rate": 9.992449799196789e-06, | |
| "loss": 0.2996, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.0056, | |
| "grad_norm": 11.082118034362793, | |
| "learning_rate": 9.984417670682733e-06, | |
| "loss": 0.3146, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.0064, | |
| "grad_norm": 10.258833885192871, | |
| "learning_rate": 9.976385542168675e-06, | |
| "loss": 0.2866, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.0072, | |
| "grad_norm": 4.887807369232178, | |
| "learning_rate": 9.968353413654619e-06, | |
| "loss": 0.2909, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.008, | |
| "grad_norm": 6.694243431091309, | |
| "learning_rate": 9.960321285140563e-06, | |
| "loss": 0.3067, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.0088, | |
| "grad_norm": 9.998162269592285, | |
| "learning_rate": 9.952289156626507e-06, | |
| "loss": 0.305, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.0096, | |
| "grad_norm": 14.1092529296875, | |
| "learning_rate": 9.94425702811245e-06, | |
| "loss": 0.2772, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.0104, | |
| "grad_norm": 11.840794563293457, | |
| "learning_rate": 9.936224899598395e-06, | |
| "loss": 0.3012, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.0112, | |
| "grad_norm": 9.802504539489746, | |
| "learning_rate": 9.928192771084338e-06, | |
| "loss": 0.2899, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.012, | |
| "grad_norm": 21.836496353149414, | |
| "learning_rate": 9.920160642570282e-06, | |
| "loss": 0.3063, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.0128, | |
| "grad_norm": 8.876431465148926, | |
| "learning_rate": 9.912128514056226e-06, | |
| "loss": 0.3292, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.0136, | |
| "grad_norm": 13.00693416595459, | |
| "learning_rate": 9.904096385542169e-06, | |
| "loss": 0.305, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.0144, | |
| "grad_norm": 10.583467483520508, | |
| "learning_rate": 9.896064257028112e-06, | |
| "loss": 0.2992, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.0152, | |
| "grad_norm": 10.112874031066895, | |
| "learning_rate": 9.888032128514056e-06, | |
| "loss": 0.2896, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.016, | |
| "grad_norm": 7.511252403259277, | |
| "learning_rate": 9.88e-06, | |
| "loss": 0.3163, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.0168, | |
| "grad_norm": 9.708840370178223, | |
| "learning_rate": 9.871967871485944e-06, | |
| "loss": 0.274, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.0176, | |
| "grad_norm": 8.038116455078125, | |
| "learning_rate": 9.863935742971888e-06, | |
| "loss": 0.2818, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.0184, | |
| "grad_norm": 9.183858871459961, | |
| "learning_rate": 9.855903614457832e-06, | |
| "loss": 0.2909, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.0192, | |
| "grad_norm": 13.90609073638916, | |
| "learning_rate": 9.847871485943776e-06, | |
| "loss": 0.2839, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 17.10437774658203, | |
| "learning_rate": 9.83983935742972e-06, | |
| "loss": 0.3, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.0208, | |
| "grad_norm": 19.14161491394043, | |
| "learning_rate": 9.831807228915664e-06, | |
| "loss": 0.2813, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.0216, | |
| "grad_norm": 13.615035057067871, | |
| "learning_rate": 9.823775100401608e-06, | |
| "loss": 0.3083, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.0224, | |
| "grad_norm": 15.303385734558105, | |
| "learning_rate": 9.81574297188755e-06, | |
| "loss": 0.2885, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.0232, | |
| "grad_norm": 13.770358085632324, | |
| "learning_rate": 9.807710843373494e-06, | |
| "loss": 0.2998, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.024, | |
| "grad_norm": 8.218709945678711, | |
| "learning_rate": 9.799678714859438e-06, | |
| "loss": 0.3097, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.0248, | |
| "grad_norm": 11.543137550354004, | |
| "learning_rate": 9.791646586345382e-06, | |
| "loss": 0.3023, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.0256, | |
| "grad_norm": 17.413169860839844, | |
| "learning_rate": 9.783614457831326e-06, | |
| "loss": 0.2775, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.0264, | |
| "grad_norm": 11.18759536743164, | |
| "learning_rate": 9.77558232931727e-06, | |
| "loss": 0.2624, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.0272, | |
| "grad_norm": 9.198080062866211, | |
| "learning_rate": 9.767550200803213e-06, | |
| "loss": 0.2866, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.028, | |
| "grad_norm": 11.68255615234375, | |
| "learning_rate": 9.759518072289157e-06, | |
| "loss": 0.2834, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.0288, | |
| "grad_norm": 5.951743125915527, | |
| "learning_rate": 9.751485943775101e-06, | |
| "loss": 0.2766, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.0296, | |
| "grad_norm": 11.82551383972168, | |
| "learning_rate": 9.743453815261045e-06, | |
| "loss": 0.2837, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.0304, | |
| "grad_norm": 12.415241241455078, | |
| "learning_rate": 9.735421686746989e-06, | |
| "loss": 0.2834, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.0312, | |
| "grad_norm": 11.2643404006958, | |
| "learning_rate": 9.727389558232933e-06, | |
| "loss": 0.2929, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.032, | |
| "grad_norm": 10.987563133239746, | |
| "learning_rate": 9.719357429718877e-06, | |
| "loss": 0.2566, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.032, | |
| "eval_test1_cer": 0.14171147957278363, | |
| "eval_test1_cer_norm": 0.10705693087137129, | |
| "eval_test1_loss": 0.265484482049942, | |
| "eval_test1_runtime": 1239.3947, | |
| "eval_test1_samples_per_second": 2.017, | |
| "eval_test1_steps_per_second": 0.504, | |
| "eval_test1_wer": 0.30916353246450334, | |
| "eval_test1_wer_norm": 0.24156170548525674, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.032, | |
| "eval_test2_cer": 0.26862377272557586, | |
| "eval_test2_cer_norm": 0.208736829872947, | |
| "eval_test2_loss": 0.44017764925956726, | |
| "eval_test2_runtime": 1372.2097, | |
| "eval_test2_samples_per_second": 1.822, | |
| "eval_test2_steps_per_second": 0.455, | |
| "eval_test2_wer": 0.47839894712748915, | |
| "eval_test2_wer_norm": 0.4037073114829246, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.0328, | |
| "grad_norm": 10.9562406539917, | |
| "learning_rate": 9.711325301204821e-06, | |
| "loss": 0.3019, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.0336, | |
| "grad_norm": 5.758121013641357, | |
| "learning_rate": 9.703293172690765e-06, | |
| "loss": 0.2978, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.0344, | |
| "grad_norm": 10.012574195861816, | |
| "learning_rate": 9.695261044176709e-06, | |
| "loss": 0.295, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.0352, | |
| "grad_norm": 8.023552894592285, | |
| "learning_rate": 9.687228915662651e-06, | |
| "loss": 0.2642, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.036, | |
| "grad_norm": 7.264355182647705, | |
| "learning_rate": 9.679196787148595e-06, | |
| "loss": 0.2856, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.0368, | |
| "grad_norm": 17.713090896606445, | |
| "learning_rate": 9.671164658634539e-06, | |
| "loss": 0.2838, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.0376, | |
| "grad_norm": 6.670265197753906, | |
| "learning_rate": 9.663132530120483e-06, | |
| "loss": 0.2555, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.0384, | |
| "grad_norm": 8.260579109191895, | |
| "learning_rate": 9.655100401606427e-06, | |
| "loss": 0.2664, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.0392, | |
| "grad_norm": 6.691250801086426, | |
| "learning_rate": 9.64706827309237e-06, | |
| "loss": 0.2731, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 9.206613540649414, | |
| "learning_rate": 9.639036144578314e-06, | |
| "loss": 0.2425, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.0408, | |
| "grad_norm": 8.662139892578125, | |
| "learning_rate": 9.631004016064258e-06, | |
| "loss": 0.2774, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.0416, | |
| "grad_norm": 7.598929405212402, | |
| "learning_rate": 9.622971887550202e-06, | |
| "loss": 0.2874, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.0424, | |
| "grad_norm": 6.341280937194824, | |
| "learning_rate": 9.614939759036145e-06, | |
| "loss": 0.2905, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.0432, | |
| "grad_norm": 9.12799072265625, | |
| "learning_rate": 9.606907630522088e-06, | |
| "loss": 0.2913, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.044, | |
| "grad_norm": 6.347501754760742, | |
| "learning_rate": 9.598875502008032e-06, | |
| "loss": 0.2809, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.0448, | |
| "grad_norm": 8.827099800109863, | |
| "learning_rate": 9.590843373493976e-06, | |
| "loss": 0.2826, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.0456, | |
| "grad_norm": 12.478800773620605, | |
| "learning_rate": 9.58281124497992e-06, | |
| "loss": 0.2742, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.0464, | |
| "grad_norm": 8.185443878173828, | |
| "learning_rate": 9.574779116465864e-06, | |
| "loss": 0.2874, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.0472, | |
| "grad_norm": 29.066831588745117, | |
| "learning_rate": 9.566746987951808e-06, | |
| "loss": 0.2753, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.048, | |
| "grad_norm": 8.861905097961426, | |
| "learning_rate": 9.558714859437752e-06, | |
| "loss": 0.2649, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.0488, | |
| "grad_norm": 11.275808334350586, | |
| "learning_rate": 9.550682730923696e-06, | |
| "loss": 0.2835, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.0496, | |
| "grad_norm": 7.473984241485596, | |
| "learning_rate": 9.542650602409638e-06, | |
| "loss": 0.2571, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.0504, | |
| "grad_norm": 6.271098613739014, | |
| "learning_rate": 9.534618473895582e-06, | |
| "loss": 0.2563, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.0512, | |
| "grad_norm": 7.11587381362915, | |
| "learning_rate": 9.526586345381526e-06, | |
| "loss": 0.2642, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.052, | |
| "grad_norm": 6.847942352294922, | |
| "learning_rate": 9.51855421686747e-06, | |
| "loss": 0.2651, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.0528, | |
| "grad_norm": 9.094855308532715, | |
| "learning_rate": 9.510522088353414e-06, | |
| "loss": 0.283, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.0536, | |
| "grad_norm": 7.255010604858398, | |
| "learning_rate": 9.502489959839358e-06, | |
| "loss": 0.2583, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.0544, | |
| "grad_norm": 10.873835563659668, | |
| "learning_rate": 9.494457831325302e-06, | |
| "loss": 0.2707, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.0552, | |
| "grad_norm": 9.488873481750488, | |
| "learning_rate": 9.486425702811246e-06, | |
| "loss": 0.2759, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.056, | |
| "grad_norm": 8.395374298095703, | |
| "learning_rate": 9.47839357429719e-06, | |
| "loss": 0.2754, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.0568, | |
| "grad_norm": 8.9957857131958, | |
| "learning_rate": 9.470361445783133e-06, | |
| "loss": 0.2672, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.0576, | |
| "grad_norm": 12.339411735534668, | |
| "learning_rate": 9.462329317269077e-06, | |
| "loss": 0.2651, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.0584, | |
| "grad_norm": 6.282168388366699, | |
| "learning_rate": 9.454297188755021e-06, | |
| "loss": 0.2746, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.0592, | |
| "grad_norm": 11.324309349060059, | |
| "learning_rate": 9.446265060240965e-06, | |
| "loss": 0.2659, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 19.300565719604492, | |
| "learning_rate": 9.438232931726909e-06, | |
| "loss": 0.2594, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.0608, | |
| "grad_norm": 9.297099113464355, | |
| "learning_rate": 9.430200803212853e-06, | |
| "loss": 0.2841, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.0616, | |
| "grad_norm": 8.913684844970703, | |
| "learning_rate": 9.422168674698797e-06, | |
| "loss": 0.2468, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.0624, | |
| "grad_norm": 9.521671295166016, | |
| "learning_rate": 9.414136546184741e-06, | |
| "loss": 0.2612, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.0632, | |
| "grad_norm": 4.213898181915283, | |
| "learning_rate": 9.406104417670685e-06, | |
| "loss": 0.2692, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.064, | |
| "grad_norm": 12.735309600830078, | |
| "learning_rate": 9.398072289156627e-06, | |
| "loss": 0.2556, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.064, | |
| "eval_test1_cer": 0.10348046904175069, | |
| "eval_test1_cer_norm": 0.07815208778543473, | |
| "eval_test1_loss": 0.24871498346328735, | |
| "eval_test1_runtime": 1203.5945, | |
| "eval_test1_samples_per_second": 2.077, | |
| "eval_test1_steps_per_second": 0.519, | |
| "eval_test1_wer": 0.2385200734715298, | |
| "eval_test1_wer_norm": 0.17586720827610391, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.064, | |
| "eval_test2_cer": 0.17687684324485758, | |
| "eval_test2_cer_norm": 0.13818465292841647, | |
| "eval_test2_loss": 0.41706421971321106, | |
| "eval_test2_runtime": 1259.4129, | |
| "eval_test2_samples_per_second": 1.985, | |
| "eval_test2_steps_per_second": 0.496, | |
| "eval_test2_wer": 0.3423552300297551, | |
| "eval_test2_wer_norm": 0.27343570937428374, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.0648, | |
| "grad_norm": 13.473925590515137, | |
| "learning_rate": 9.390040160642571e-06, | |
| "loss": 0.282, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.0656, | |
| "grad_norm": 10.660150527954102, | |
| "learning_rate": 9.382008032128515e-06, | |
| "loss": 0.2741, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.0664, | |
| "grad_norm": 6.230668544769287, | |
| "learning_rate": 9.373975903614459e-06, | |
| "loss": 0.2686, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.0672, | |
| "grad_norm": 5.246466159820557, | |
| "learning_rate": 9.365943775100403e-06, | |
| "loss": 0.2382, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.068, | |
| "grad_norm": 5.55143928527832, | |
| "learning_rate": 9.357911646586347e-06, | |
| "loss": 0.2386, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.0688, | |
| "grad_norm": 7.775327682495117, | |
| "learning_rate": 9.34987951807229e-06, | |
| "loss": 0.2767, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.0696, | |
| "grad_norm": 9.929217338562012, | |
| "learning_rate": 9.341847389558234e-06, | |
| "loss": 0.241, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.0704, | |
| "grad_norm": 6.488868236541748, | |
| "learning_rate": 9.333815261044178e-06, | |
| "loss": 0.2761, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.0712, | |
| "grad_norm": 10.308218955993652, | |
| "learning_rate": 9.325783132530122e-06, | |
| "loss": 0.2927, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.072, | |
| "grad_norm": 16.63512420654297, | |
| "learning_rate": 9.317751004016065e-06, | |
| "loss": 0.262, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.0728, | |
| "grad_norm": 8.194585800170898, | |
| "learning_rate": 9.309718875502008e-06, | |
| "loss": 0.2535, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.0736, | |
| "grad_norm": 10.36143970489502, | |
| "learning_rate": 9.301686746987952e-06, | |
| "loss": 0.2692, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.0744, | |
| "grad_norm": 8.524481773376465, | |
| "learning_rate": 9.293654618473896e-06, | |
| "loss": 0.2785, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.0752, | |
| "grad_norm": 5.517594337463379, | |
| "learning_rate": 9.28562248995984e-06, | |
| "loss": 0.26, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.076, | |
| "grad_norm": 8.915616989135742, | |
| "learning_rate": 9.277670682730925e-06, | |
| "loss": 0.2491, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.0768, | |
| "grad_norm": 10.76728343963623, | |
| "learning_rate": 9.269638554216868e-06, | |
| "loss": 0.2721, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.0776, | |
| "grad_norm": 9.26211929321289, | |
| "learning_rate": 9.261606425702812e-06, | |
| "loss": 0.2614, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.0784, | |
| "grad_norm": 7.043396472930908, | |
| "learning_rate": 9.253574297188756e-06, | |
| "loss": 0.2445, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.0792, | |
| "grad_norm": 6.651730537414551, | |
| "learning_rate": 9.2455421686747e-06, | |
| "loss": 0.2872, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 19.39574432373047, | |
| "learning_rate": 9.237510040160642e-06, | |
| "loss": 0.2755, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.0808, | |
| "grad_norm": 7.920570373535156, | |
| "learning_rate": 9.229477911646586e-06, | |
| "loss": 0.2596, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.0816, | |
| "grad_norm": 8.022053718566895, | |
| "learning_rate": 9.22144578313253e-06, | |
| "loss": 0.257, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.0824, | |
| "grad_norm": 11.654303550720215, | |
| "learning_rate": 9.213413654618474e-06, | |
| "loss": 0.257, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.0832, | |
| "grad_norm": 8.477298736572266, | |
| "learning_rate": 9.205381526104418e-06, | |
| "loss": 0.2405, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.084, | |
| "grad_norm": 6.778027534484863, | |
| "learning_rate": 9.197349397590362e-06, | |
| "loss": 0.2688, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.0848, | |
| "grad_norm": 3.938582420349121, | |
| "learning_rate": 9.189317269076306e-06, | |
| "loss": 0.2512, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.0856, | |
| "grad_norm": 7.830332279205322, | |
| "learning_rate": 9.18128514056225e-06, | |
| "loss": 0.2595, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.0864, | |
| "grad_norm": 6.613556861877441, | |
| "learning_rate": 9.173253012048194e-06, | |
| "loss": 0.2182, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.0872, | |
| "grad_norm": 7.161635398864746, | |
| "learning_rate": 9.165220883534138e-06, | |
| "loss": 0.2548, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.088, | |
| "grad_norm": 8.175447463989258, | |
| "learning_rate": 9.157188755020082e-06, | |
| "loss": 0.2754, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.0888, | |
| "grad_norm": 8.775857925415039, | |
| "learning_rate": 9.149156626506026e-06, | |
| "loss": 0.2552, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.0896, | |
| "grad_norm": 8.940972328186035, | |
| "learning_rate": 9.14112449799197e-06, | |
| "loss": 0.247, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.0904, | |
| "grad_norm": 7.314133644104004, | |
| "learning_rate": 9.133092369477913e-06, | |
| "loss": 0.2463, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.0912, | |
| "grad_norm": 13.069714546203613, | |
| "learning_rate": 9.125060240963857e-06, | |
| "loss": 0.2368, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.092, | |
| "grad_norm": 3.7738101482391357, | |
| "learning_rate": 9.1170281124498e-06, | |
| "loss": 0.2429, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.0928, | |
| "grad_norm": 4.591954708099365, | |
| "learning_rate": 9.108995983935743e-06, | |
| "loss": 0.2465, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.0936, | |
| "grad_norm": 7.4973225593566895, | |
| "learning_rate": 9.100963855421687e-06, | |
| "loss": 0.2526, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.0944, | |
| "grad_norm": 16.275197982788086, | |
| "learning_rate": 9.092931726907631e-06, | |
| "loss": 0.2469, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.0952, | |
| "grad_norm": 13.245892524719238, | |
| "learning_rate": 9.084899598393575e-06, | |
| "loss": 0.2715, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.096, | |
| "grad_norm": 10.107722282409668, | |
| "learning_rate": 9.076867469879519e-06, | |
| "loss": 0.2288, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.096, | |
| "eval_test1_cer": 0.08446299200836507, | |
| "eval_test1_cer_norm": 0.05986985857324657, | |
| "eval_test1_loss": 0.24487894773483276, | |
| "eval_test1_runtime": 1174.9494, | |
| "eval_test1_samples_per_second": 2.128, | |
| "eval_test1_steps_per_second": 0.532, | |
| "eval_test1_wer": 0.2207936091431237, | |
| "eval_test1_wer_norm": 0.15590753674858995, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.096, | |
| "eval_test2_cer": 0.21867417030649194, | |
| "eval_test2_cer_norm": 0.17033041524635884, | |
| "eval_test2_loss": 0.4188956022262573, | |
| "eval_test2_runtime": 1303.0061, | |
| "eval_test2_samples_per_second": 1.919, | |
| "eval_test2_steps_per_second": 0.48, | |
| "eval_test2_wer": 0.38798924238956284, | |
| "eval_test2_wer_norm": 0.31910382764153106, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.0968, | |
| "grad_norm": 13.199896812438965, | |
| "learning_rate": 9.068835341365463e-06, | |
| "loss": 0.2436, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.0976, | |
| "grad_norm": 8.258294105529785, | |
| "learning_rate": 9.060803212851407e-06, | |
| "loss": 0.2677, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.0984, | |
| "grad_norm": 6.465874671936035, | |
| "learning_rate": 9.052771084337351e-06, | |
| "loss": 0.2659, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.0992, | |
| "grad_norm": 8.216276168823242, | |
| "learning_rate": 9.044738955823293e-06, | |
| "loss": 0.2553, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 7.199169158935547, | |
| "learning_rate": 9.036706827309237e-06, | |
| "loss": 0.2244, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.1008, | |
| "grad_norm": 5.689333438873291, | |
| "learning_rate": 9.028674698795181e-06, | |
| "loss": 0.2366, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.1016, | |
| "grad_norm": 9.028944969177246, | |
| "learning_rate": 9.020642570281125e-06, | |
| "loss": 0.2152, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.1024, | |
| "grad_norm": 6.4437127113342285, | |
| "learning_rate": 9.012610441767069e-06, | |
| "loss": 0.2549, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.1032, | |
| "grad_norm": 9.835594177246094, | |
| "learning_rate": 9.004578313253013e-06, | |
| "loss": 0.2383, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.104, | |
| "grad_norm": 7.06815767288208, | |
| "learning_rate": 8.996546184738957e-06, | |
| "loss": 0.2381, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.1048, | |
| "grad_norm": 5.8962297439575195, | |
| "learning_rate": 8.9885140562249e-06, | |
| "loss": 0.251, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.1056, | |
| "grad_norm": 8.3475980758667, | |
| "learning_rate": 8.980481927710844e-06, | |
| "loss": 0.2315, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.1064, | |
| "grad_norm": 9.03033447265625, | |
| "learning_rate": 8.972449799196787e-06, | |
| "loss": 0.2229, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.1072, | |
| "grad_norm": 2.628983497619629, | |
| "learning_rate": 8.96441767068273e-06, | |
| "loss": 0.2563, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.108, | |
| "grad_norm": 12.840631484985352, | |
| "learning_rate": 8.956385542168675e-06, | |
| "loss": 0.2499, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.1088, | |
| "grad_norm": 5.249873161315918, | |
| "learning_rate": 8.94843373493976e-06, | |
| "loss": 0.2406, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.1096, | |
| "grad_norm": 7.155534267425537, | |
| "learning_rate": 8.940401606425704e-06, | |
| "loss": 0.2345, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.1104, | |
| "grad_norm": 6.4776787757873535, | |
| "learning_rate": 8.932369477911648e-06, | |
| "loss": 0.2504, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.1112, | |
| "grad_norm": 13.455100059509277, | |
| "learning_rate": 8.924337349397592e-06, | |
| "loss": 0.2557, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.112, | |
| "grad_norm": 7.236751079559326, | |
| "learning_rate": 8.916305220883535e-06, | |
| "loss": 0.2355, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.1128, | |
| "grad_norm": 2.6910972595214844, | |
| "learning_rate": 8.908273092369478e-06, | |
| "loss": 0.2406, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.1136, | |
| "grad_norm": 10.298126220703125, | |
| "learning_rate": 8.900240963855422e-06, | |
| "loss": 0.2418, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.1144, | |
| "grad_norm": 9.934380531311035, | |
| "learning_rate": 8.892208835341366e-06, | |
| "loss": 0.2457, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.1152, | |
| "grad_norm": 5.105091571807861, | |
| "learning_rate": 8.88417670682731e-06, | |
| "loss": 0.2278, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.116, | |
| "grad_norm": 8.721418380737305, | |
| "learning_rate": 8.876144578313254e-06, | |
| "loss": 0.2355, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.1168, | |
| "grad_norm": 10.942416191101074, | |
| "learning_rate": 8.868112449799198e-06, | |
| "loss": 0.2311, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.1176, | |
| "grad_norm": 5.4364705085754395, | |
| "learning_rate": 8.860080321285142e-06, | |
| "loss": 0.26, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.1184, | |
| "grad_norm": 10.078380584716797, | |
| "learning_rate": 8.852048192771086e-06, | |
| "loss": 0.2596, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.1192, | |
| "grad_norm": 9.630817413330078, | |
| "learning_rate": 8.844016064257028e-06, | |
| "loss": 0.2182, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 9.832001686096191, | |
| "learning_rate": 8.835983935742972e-06, | |
| "loss": 0.2581, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.1208, | |
| "grad_norm": 8.944684028625488, | |
| "learning_rate": 8.827951807228916e-06, | |
| "loss": 0.2388, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.1216, | |
| "grad_norm": 8.616948127746582, | |
| "learning_rate": 8.81991967871486e-06, | |
| "loss": 0.2433, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.1224, | |
| "grad_norm": 7.957048416137695, | |
| "learning_rate": 8.811887550200804e-06, | |
| "loss": 0.2338, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.1232, | |
| "grad_norm": 6.853145122528076, | |
| "learning_rate": 8.803855421686748e-06, | |
| "loss": 0.2328, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.124, | |
| "grad_norm": 5.673152923583984, | |
| "learning_rate": 8.795823293172692e-06, | |
| "loss": 0.2492, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.1248, | |
| "grad_norm": 9.295065879821777, | |
| "learning_rate": 8.787871485943776e-06, | |
| "loss": 0.2342, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.1256, | |
| "grad_norm": 7.0689215660095215, | |
| "learning_rate": 8.77983935742972e-06, | |
| "loss": 0.2481, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.1264, | |
| "grad_norm": 3.174997329711914, | |
| "learning_rate": 8.771807228915664e-06, | |
| "loss": 0.2384, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.1272, | |
| "grad_norm": 10.203186988830566, | |
| "learning_rate": 8.763775100401608e-06, | |
| "loss": 0.2393, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.128, | |
| "grad_norm": 7.220344066619873, | |
| "learning_rate": 8.755742971887552e-06, | |
| "loss": 0.2696, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.128, | |
| "eval_test1_cer": 0.061454365523937564, | |
| "eval_test1_cer_norm": 0.04506927269671285, | |
| "eval_test1_loss": 0.22977162897586823, | |
| "eval_test1_runtime": 1134.9922, | |
| "eval_test1_samples_per_second": 2.203, | |
| "eval_test1_steps_per_second": 0.551, | |
| "eval_test1_wer": 0.17863494562523688, | |
| "eval_test1_wer_norm": 0.11663111137087583, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.128, | |
| "eval_test2_cer": 0.16853791018031133, | |
| "eval_test2_cer_norm": 0.13256313913851875, | |
| "eval_test2_loss": 0.4023756980895996, | |
| "eval_test2_runtime": 1236.1872, | |
| "eval_test2_samples_per_second": 2.022, | |
| "eval_test2_steps_per_second": 0.506, | |
| "eval_test2_wer": 0.32633325703822386, | |
| "eval_test2_wer_norm": 0.2594545037818015, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.1288, | |
| "grad_norm": 6.053903102874756, | |
| "learning_rate": 8.747710843373496e-06, | |
| "loss": 0.2549, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.1296, | |
| "grad_norm": 7.160613536834717, | |
| "learning_rate": 8.73967871485944e-06, | |
| "loss": 0.247, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.1304, | |
| "grad_norm": 9.369489669799805, | |
| "learning_rate": 8.731726907630524e-06, | |
| "loss": 0.2391, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.1312, | |
| "grad_norm": 7.7761549949646, | |
| "learning_rate": 8.723694779116466e-06, | |
| "loss": 0.2477, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.132, | |
| "grad_norm": 15.849722862243652, | |
| "learning_rate": 8.71566265060241e-06, | |
| "loss": 0.2477, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.1328, | |
| "grad_norm": 4.059557914733887, | |
| "learning_rate": 8.707630522088354e-06, | |
| "loss": 0.2358, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.1336, | |
| "grad_norm": 7.216188907623291, | |
| "learning_rate": 8.699598393574298e-06, | |
| "loss": 0.2663, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 0.1344, | |
| "grad_norm": 5.579075336456299, | |
| "learning_rate": 8.691566265060242e-06, | |
| "loss": 0.2388, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.1352, | |
| "grad_norm": 7.436819553375244, | |
| "learning_rate": 8.683534136546186e-06, | |
| "loss": 0.2262, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 0.136, | |
| "grad_norm": 9.426555633544922, | |
| "learning_rate": 8.67550200803213e-06, | |
| "loss": 0.2334, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.1368, | |
| "grad_norm": 11.434511184692383, | |
| "learning_rate": 8.667469879518073e-06, | |
| "loss": 0.2536, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 0.1376, | |
| "grad_norm": 9.391510963439941, | |
| "learning_rate": 8.659437751004017e-06, | |
| "loss": 0.2311, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.1384, | |
| "grad_norm": 14.07253646850586, | |
| "learning_rate": 8.65140562248996e-06, | |
| "loss": 0.2272, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 0.1392, | |
| "grad_norm": 10.38249683380127, | |
| "learning_rate": 8.643373493975904e-06, | |
| "loss": 0.2602, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 11.22042465209961, | |
| "learning_rate": 8.635341365461847e-06, | |
| "loss": 0.2664, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.1408, | |
| "grad_norm": 8.036049842834473, | |
| "learning_rate": 8.627309236947791e-06, | |
| "loss": 0.2423, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.1416, | |
| "grad_norm": 8.139640808105469, | |
| "learning_rate": 8.619277108433735e-06, | |
| "loss": 0.2336, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 0.1424, | |
| "grad_norm": 10.12194538116455, | |
| "learning_rate": 8.61124497991968e-06, | |
| "loss": 0.2329, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.1432, | |
| "grad_norm": 8.494665145874023, | |
| "learning_rate": 8.603212851405623e-06, | |
| "loss": 0.2411, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 0.144, | |
| "grad_norm": 7.564216136932373, | |
| "learning_rate": 8.595180722891567e-06, | |
| "loss": 0.2477, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.1448, | |
| "grad_norm": 6.718033313751221, | |
| "learning_rate": 8.587148594377511e-06, | |
| "loss": 0.243, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 0.1456, | |
| "grad_norm": 4.666207313537598, | |
| "learning_rate": 8.579116465863455e-06, | |
| "loss": 0.2492, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.1464, | |
| "grad_norm": 9.410490036010742, | |
| "learning_rate": 8.571084337349397e-06, | |
| "loss": 0.2265, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 0.1472, | |
| "grad_norm": 6.1468119621276855, | |
| "learning_rate": 8.563052208835341e-06, | |
| "loss": 0.2409, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.148, | |
| "grad_norm": 9.485304832458496, | |
| "learning_rate": 8.555020080321285e-06, | |
| "loss": 0.2584, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.1488, | |
| "grad_norm": 4.741575717926025, | |
| "learning_rate": 8.546987951807229e-06, | |
| "loss": 0.2309, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.1496, | |
| "grad_norm": 7.345742225646973, | |
| "learning_rate": 8.538955823293173e-06, | |
| "loss": 0.233, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 0.1504, | |
| "grad_norm": 17.133358001708984, | |
| "learning_rate": 8.530923694779117e-06, | |
| "loss": 0.2335, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.1512, | |
| "grad_norm": 7.573137283325195, | |
| "learning_rate": 8.52289156626506e-06, | |
| "loss": 0.2346, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 0.152, | |
| "grad_norm": 11.430822372436523, | |
| "learning_rate": 8.514859437751005e-06, | |
| "loss": 0.2435, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.1528, | |
| "grad_norm": 7.896162986755371, | |
| "learning_rate": 8.506827309236948e-06, | |
| "loss": 0.239, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 0.1536, | |
| "grad_norm": 7.543678283691406, | |
| "learning_rate": 8.498795180722892e-06, | |
| "loss": 0.2509, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.1544, | |
| "grad_norm": 11.438505172729492, | |
| "learning_rate": 8.490763052208836e-06, | |
| "loss": 0.2244, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 0.1552, | |
| "grad_norm": 5.8480730056762695, | |
| "learning_rate": 8.48281124497992e-06, | |
| "loss": 0.2428, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.156, | |
| "grad_norm": 6.436691761016846, | |
| "learning_rate": 8.474779116465865e-06, | |
| "loss": 0.2367, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.1568, | |
| "grad_norm": 7.197071075439453, | |
| "learning_rate": 8.466746987951808e-06, | |
| "loss": 0.2162, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.1576, | |
| "grad_norm": 4.8729634284973145, | |
| "learning_rate": 8.458714859437752e-06, | |
| "loss": 0.2231, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 0.1584, | |
| "grad_norm": 7.567530155181885, | |
| "learning_rate": 8.450682730923695e-06, | |
| "loss": 0.2208, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.1592, | |
| "grad_norm": 6.356396675109863, | |
| "learning_rate": 8.442650602409639e-06, | |
| "loss": 0.2217, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 11.299875259399414, | |
| "learning_rate": 8.434618473895582e-06, | |
| "loss": 0.2376, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_test1_cer": 0.0942891552767197, | |
| "eval_test1_cer_norm": 0.06996422311330948, | |
| "eval_test1_loss": 0.22547538578510284, | |
| "eval_test1_runtime": 1189.7817, | |
| "eval_test1_samples_per_second": 2.101, | |
| "eval_test1_steps_per_second": 0.525, | |
| "eval_test1_wer": 0.22478789469080732, | |
| "eval_test1_wer_norm": 0.16227826646015373, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_test2_cer": 0.17272370926195543, | |
| "eval_test2_cer_norm": 0.13756488224356989, | |
| "eval_test2_loss": 0.3925027847290039, | |
| "eval_test2_runtime": 1241.7968, | |
| "eval_test2_samples_per_second": 2.013, | |
| "eval_test2_steps_per_second": 0.503, | |
| "eval_test2_wer": 0.3331712062256809, | |
| "eval_test2_wer_norm": 0.26449690579876234, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.1608, | |
| "grad_norm": 3.2471466064453125, | |
| "learning_rate": 8.426586345381526e-06, | |
| "loss": 0.2182, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 0.1616, | |
| "grad_norm": 8.284675598144531, | |
| "learning_rate": 8.41855421686747e-06, | |
| "loss": 0.2382, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 0.1624, | |
| "grad_norm": 5.969610214233398, | |
| "learning_rate": 8.410522088353414e-06, | |
| "loss": 0.2372, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 0.1632, | |
| "grad_norm": 12.841833114624023, | |
| "learning_rate": 8.402489959839358e-06, | |
| "loss": 0.2389, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 0.164, | |
| "grad_norm": 6.656698703765869, | |
| "learning_rate": 8.394457831325302e-06, | |
| "loss": 0.2368, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.1648, | |
| "grad_norm": 5.6654276847839355, | |
| "learning_rate": 8.386425702811246e-06, | |
| "loss": 0.2122, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 0.1656, | |
| "grad_norm": 4.837275505065918, | |
| "learning_rate": 8.378393574297188e-06, | |
| "loss": 0.2088, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 0.1664, | |
| "grad_norm": 7.151382923126221, | |
| "learning_rate": 8.370361445783132e-06, | |
| "loss": 0.2328, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.1672, | |
| "grad_norm": 10.351961135864258, | |
| "learning_rate": 8.362329317269076e-06, | |
| "loss": 0.2217, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 0.168, | |
| "grad_norm": 4.848665714263916, | |
| "learning_rate": 8.35429718875502e-06, | |
| "loss": 0.2372, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.1688, | |
| "grad_norm": 4.431462287902832, | |
| "learning_rate": 8.346265060240964e-06, | |
| "loss": 0.2193, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 0.1696, | |
| "grad_norm": 4.42501163482666, | |
| "learning_rate": 8.338232931726908e-06, | |
| "loss": 0.2198, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 0.1704, | |
| "grad_norm": 11.573799133300781, | |
| "learning_rate": 8.330200803212852e-06, | |
| "loss": 0.2454, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 0.1712, | |
| "grad_norm": 5.41486120223999, | |
| "learning_rate": 8.322168674698796e-06, | |
| "loss": 0.2328, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 0.172, | |
| "grad_norm": 5.6592488288879395, | |
| "learning_rate": 8.31413654618474e-06, | |
| "loss": 0.2388, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.1728, | |
| "grad_norm": 5.244436264038086, | |
| "learning_rate": 8.306104417670683e-06, | |
| "loss": 0.2384, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 0.1736, | |
| "grad_norm": 4.57738733291626, | |
| "learning_rate": 8.298072289156627e-06, | |
| "loss": 0.2227, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 0.1744, | |
| "grad_norm": 11.43252944946289, | |
| "learning_rate": 8.290040160642571e-06, | |
| "loss": 0.2265, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 0.1752, | |
| "grad_norm": 4.118782997131348, | |
| "learning_rate": 8.282008032128515e-06, | |
| "loss": 0.2211, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 0.176, | |
| "grad_norm": 7.751514434814453, | |
| "learning_rate": 8.273975903614459e-06, | |
| "loss": 0.2359, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.1768, | |
| "grad_norm": 11.36629867553711, | |
| "learning_rate": 8.265943775100403e-06, | |
| "loss": 0.234, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 0.1776, | |
| "grad_norm": 4.669189453125, | |
| "learning_rate": 8.257911646586347e-06, | |
| "loss": 0.222, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 0.1784, | |
| "grad_norm": 8.595556259155273, | |
| "learning_rate": 8.249879518072291e-06, | |
| "loss": 0.2188, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 0.1792, | |
| "grad_norm": 9.70352840423584, | |
| "learning_rate": 8.241847389558235e-06, | |
| "loss": 0.2289, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 6.345508575439453, | |
| "learning_rate": 8.233815261044177e-06, | |
| "loss": 0.222, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.1808, | |
| "grad_norm": 4.9803643226623535, | |
| "learning_rate": 8.225783132530121e-06, | |
| "loss": 0.2338, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 0.1816, | |
| "grad_norm": 9.151689529418945, | |
| "learning_rate": 8.217751004016065e-06, | |
| "loss": 0.2475, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 0.1824, | |
| "grad_norm": 7.338719844818115, | |
| "learning_rate": 8.209718875502009e-06, | |
| "loss": 0.2194, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 0.1832, | |
| "grad_norm": 15.189725875854492, | |
| "learning_rate": 8.201686746987953e-06, | |
| "loss": 0.2406, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 0.184, | |
| "grad_norm": 6.638333320617676, | |
| "learning_rate": 8.193654618473897e-06, | |
| "loss": 0.2365, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.1848, | |
| "grad_norm": 11.742992401123047, | |
| "learning_rate": 8.18562248995984e-06, | |
| "loss": 0.2187, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 0.1856, | |
| "grad_norm": 13.45174789428711, | |
| "learning_rate": 8.177590361445784e-06, | |
| "loss": 0.215, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 0.1864, | |
| "grad_norm": 4.851787090301514, | |
| "learning_rate": 8.169558232931728e-06, | |
| "loss": 0.2547, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 0.1872, | |
| "grad_norm": 15.86487865447998, | |
| "learning_rate": 8.161526104417672e-06, | |
| "loss": 0.205, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 0.188, | |
| "grad_norm": 7.100121974945068, | |
| "learning_rate": 8.153574297188755e-06, | |
| "loss": 0.2496, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.1888, | |
| "grad_norm": 4.416376113891602, | |
| "learning_rate": 8.145542168674699e-06, | |
| "loss": 0.2205, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 0.1896, | |
| "grad_norm": 8.244140625, | |
| "learning_rate": 8.137510040160643e-06, | |
| "loss": 0.216, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 0.1904, | |
| "grad_norm": 7.894223690032959, | |
| "learning_rate": 8.129477911646587e-06, | |
| "loss": 0.2482, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 0.1912, | |
| "grad_norm": 7.517197608947754, | |
| "learning_rate": 8.121526104417673e-06, | |
| "loss": 0.2349, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 0.192, | |
| "grad_norm": 18.533597946166992, | |
| "learning_rate": 8.113493975903615e-06, | |
| "loss": 0.2205, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.192, | |
| "eval_test1_cer": 0.08556464261707372, | |
| "eval_test1_cer_norm": 0.06268398684178933, | |
| "eval_test1_loss": 0.22241076827049255, | |
| "eval_test1_runtime": 1196.1164, | |
| "eval_test1_samples_per_second": 2.09, | |
| "eval_test1_steps_per_second": 0.523, | |
| "eval_test1_wer": 0.21058923000670574, | |
| "eval_test1_wer_norm": 0.14939068938309127, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.192, | |
| "eval_test2_cer": 0.13958729980960913, | |
| "eval_test2_cer_norm": 0.1119799736597459, | |
| "eval_test2_loss": 0.3912878632545471, | |
| "eval_test2_runtime": 1198.2977, | |
| "eval_test2_samples_per_second": 2.086, | |
| "eval_test2_steps_per_second": 0.522, | |
| "eval_test2_wer": 0.28147173266193637, | |
| "eval_test2_wer_norm": 0.21352853541141417, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.1928, | |
| "grad_norm": 7.209935665130615, | |
| "learning_rate": 8.1055421686747e-06, | |
| "loss": 0.2215, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 0.1936, | |
| "grad_norm": 8.64651107788086, | |
| "learning_rate": 8.097590361445784e-06, | |
| "loss": 0.2163, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 0.1944, | |
| "grad_norm": 11.35953426361084, | |
| "learning_rate": 8.089558232931727e-06, | |
| "loss": 0.2356, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 0.1952, | |
| "grad_norm": 10.773683547973633, | |
| "learning_rate": 8.081526104417671e-06, | |
| "loss": 0.2359, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 0.196, | |
| "grad_norm": 7.809534072875977, | |
| "learning_rate": 8.073493975903615e-06, | |
| "loss": 0.2224, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.1968, | |
| "grad_norm": 6.901273727416992, | |
| "learning_rate": 8.06546184738956e-06, | |
| "loss": 0.2193, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 0.1976, | |
| "grad_norm": 10.301507949829102, | |
| "learning_rate": 8.057429718875503e-06, | |
| "loss": 0.238, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 0.1984, | |
| "grad_norm": 6.89518404006958, | |
| "learning_rate": 8.049397590361447e-06, | |
| "loss": 0.2358, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 0.1992, | |
| "grad_norm": 3.8990767002105713, | |
| "learning_rate": 8.041365461847391e-06, | |
| "loss": 0.2192, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 6.363491058349609, | |
| "learning_rate": 8.033333333333335e-06, | |
| "loss": 0.2334, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.2008, | |
| "grad_norm": 21.063215255737305, | |
| "learning_rate": 8.025301204819279e-06, | |
| "loss": 0.236, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 0.2016, | |
| "grad_norm": 7.973589897155762, | |
| "learning_rate": 8.017269076305221e-06, | |
| "loss": 0.237, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 0.2024, | |
| "grad_norm": 8.546441078186035, | |
| "learning_rate": 8.009236947791165e-06, | |
| "loss": 0.2591, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 0.2032, | |
| "grad_norm": 14.01245403289795, | |
| "learning_rate": 8.001204819277109e-06, | |
| "loss": 0.2364, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 0.204, | |
| "grad_norm": 3.486008405685425, | |
| "learning_rate": 7.993172690763053e-06, | |
| "loss": 0.2349, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.2048, | |
| "grad_norm": 8.98222827911377, | |
| "learning_rate": 7.985140562248997e-06, | |
| "loss": 0.2138, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.2056, | |
| "grad_norm": 10.626484870910645, | |
| "learning_rate": 7.97710843373494e-06, | |
| "loss": 0.2229, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 0.2064, | |
| "grad_norm": 6.444272041320801, | |
| "learning_rate": 7.969076305220885e-06, | |
| "loss": 0.2145, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 0.2072, | |
| "grad_norm": 6.722916603088379, | |
| "learning_rate": 7.961044176706828e-06, | |
| "loss": 0.2189, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 0.208, | |
| "grad_norm": 10.45226001739502, | |
| "learning_rate": 7.953012048192772e-06, | |
| "loss": 0.214, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.2088, | |
| "grad_norm": 7.136618614196777, | |
| "learning_rate": 7.944979919678716e-06, | |
| "loss": 0.2147, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 0.2096, | |
| "grad_norm": 5.5284953117370605, | |
| "learning_rate": 7.937028112449799e-06, | |
| "loss": 0.2236, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 0.2104, | |
| "grad_norm": 7.089993000030518, | |
| "learning_rate": 7.928995983935743e-06, | |
| "loss": 0.2083, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 0.2112, | |
| "grad_norm": 9.043608665466309, | |
| "learning_rate": 7.920963855421687e-06, | |
| "loss": 0.2478, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 0.212, | |
| "grad_norm": 5.048727989196777, | |
| "learning_rate": 7.91293172690763e-06, | |
| "loss": 0.2063, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.2128, | |
| "grad_norm": 9.47496509552002, | |
| "learning_rate": 7.904899598393575e-06, | |
| "loss": 0.2401, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 0.2136, | |
| "grad_norm": 5.263045787811279, | |
| "learning_rate": 7.896867469879519e-06, | |
| "loss": 0.2373, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 0.2144, | |
| "grad_norm": 10.800714492797852, | |
| "learning_rate": 7.888835341365462e-06, | |
| "loss": 0.2123, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 0.2152, | |
| "grad_norm": 5.966376304626465, | |
| "learning_rate": 7.880803212851406e-06, | |
| "loss": 0.2335, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 0.216, | |
| "grad_norm": 4.061577320098877, | |
| "learning_rate": 7.87277108433735e-06, | |
| "loss": 0.2183, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.2168, | |
| "grad_norm": 10.104215621948242, | |
| "learning_rate": 7.864738955823293e-06, | |
| "loss": 0.1969, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 0.2176, | |
| "grad_norm": 6.445654392242432, | |
| "learning_rate": 7.856706827309236e-06, | |
| "loss": 0.2144, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 0.2184, | |
| "grad_norm": 9.509758949279785, | |
| "learning_rate": 7.84867469879518e-06, | |
| "loss": 0.2022, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 0.2192, | |
| "grad_norm": 11.04575252532959, | |
| "learning_rate": 7.840642570281124e-06, | |
| "loss": 0.2153, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 10.30285358428955, | |
| "learning_rate": 7.832610441767068e-06, | |
| "loss": 0.2269, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.2208, | |
| "grad_norm": 9.692445755004883, | |
| "learning_rate": 7.824578313253012e-06, | |
| "loss": 0.2372, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 0.2216, | |
| "grad_norm": 7.303224563598633, | |
| "learning_rate": 7.816546184738956e-06, | |
| "loss": 0.2229, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 0.2224, | |
| "grad_norm": 7.364916801452637, | |
| "learning_rate": 7.8085140562249e-06, | |
| "loss": 0.2312, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 0.2232, | |
| "grad_norm": 8.000985145568848, | |
| "learning_rate": 7.800481927710844e-06, | |
| "loss": 0.2146, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 0.224, | |
| "grad_norm": 7.1691131591796875, | |
| "learning_rate": 7.792449799196788e-06, | |
| "loss": 0.2328, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.224, | |
| "eval_test1_cer": 0.06472664127268653, | |
| "eval_test1_cer_norm": 0.04759046269839364, | |
| "eval_test1_loss": 0.21469315886497498, | |
| "eval_test1_runtime": 1210.6322, | |
| "eval_test1_samples_per_second": 2.065, | |
| "eval_test1_steps_per_second": 0.516, | |
| "eval_test1_wer": 0.18274585264876528, | |
| "eval_test1_wer_norm": 0.12037172331161052, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.224, | |
| "eval_test2_cer": 0.17353100384514877, | |
| "eval_test2_cer_norm": 0.13871242640223116, | |
| "eval_test2_loss": 0.37395521998405457, | |
| "eval_test2_runtime": 1313.1792, | |
| "eval_test2_samples_per_second": 1.904, | |
| "eval_test2_steps_per_second": 0.476, | |
| "eval_test2_wer": 0.35056649118791483, | |
| "eval_test2_wer_norm": 0.2835205134082054, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.2248, | |
| "grad_norm": 5.209496021270752, | |
| "learning_rate": 7.784417670682732e-06, | |
| "loss": 0.2212, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 0.2256, | |
| "grad_norm": 6.241885185241699, | |
| "learning_rate": 7.776465863453816e-06, | |
| "loss": 0.2175, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 0.2264, | |
| "grad_norm": 6.768528461456299, | |
| "learning_rate": 7.76843373493976e-06, | |
| "loss": 0.2111, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 0.2272, | |
| "grad_norm": 16.620553970336914, | |
| "learning_rate": 7.760401606425704e-06, | |
| "loss": 0.2201, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 0.228, | |
| "grad_norm": 5.785097122192383, | |
| "learning_rate": 7.752369477911648e-06, | |
| "loss": 0.2195, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.2288, | |
| "grad_norm": 7.3884663581848145, | |
| "learning_rate": 7.744337349397592e-06, | |
| "loss": 0.2107, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 0.2296, | |
| "grad_norm": 6.965432643890381, | |
| "learning_rate": 7.736305220883534e-06, | |
| "loss": 0.247, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 0.2304, | |
| "grad_norm": 7.899501323699951, | |
| "learning_rate": 7.728273092369478e-06, | |
| "loss": 0.2315, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 0.2312, | |
| "grad_norm": 8.336009979248047, | |
| "learning_rate": 7.720240963855422e-06, | |
| "loss": 0.2262, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 0.232, | |
| "grad_norm": 7.90683126449585, | |
| "learning_rate": 7.712208835341366e-06, | |
| "loss": 0.2322, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.2328, | |
| "grad_norm": 6.586302757263184, | |
| "learning_rate": 7.70417670682731e-06, | |
| "loss": 0.2341, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 0.2336, | |
| "grad_norm": 6.594479084014893, | |
| "learning_rate": 7.696144578313254e-06, | |
| "loss": 0.2226, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 0.2344, | |
| "grad_norm": 7.850062370300293, | |
| "learning_rate": 7.688112449799197e-06, | |
| "loss": 0.2094, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 0.2352, | |
| "grad_norm": 12.644471168518066, | |
| "learning_rate": 7.680080321285141e-06, | |
| "loss": 0.226, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 0.236, | |
| "grad_norm": 7.289064884185791, | |
| "learning_rate": 7.672048192771085e-06, | |
| "loss": 0.2214, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.2368, | |
| "grad_norm": 10.635286331176758, | |
| "learning_rate": 7.664016064257028e-06, | |
| "loss": 0.2142, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 0.2376, | |
| "grad_norm": 5.214193344116211, | |
| "learning_rate": 7.655983935742971e-06, | |
| "loss": 0.2218, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 0.2384, | |
| "grad_norm": 6.800753593444824, | |
| "learning_rate": 7.647951807228915e-06, | |
| "loss": 0.2179, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 0.2392, | |
| "grad_norm": 8.79254150390625, | |
| "learning_rate": 7.63991967871486e-06, | |
| "loss": 0.2066, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 7.172235488891602, | |
| "learning_rate": 7.631887550200803e-06, | |
| "loss": 0.2207, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.2408, | |
| "grad_norm": 5.066342830657959, | |
| "learning_rate": 7.623855421686748e-06, | |
| "loss": 0.2034, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 0.2416, | |
| "grad_norm": 5.975930690765381, | |
| "learning_rate": 7.615903614457832e-06, | |
| "loss": 0.2136, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 0.2424, | |
| "grad_norm": 5.854097366333008, | |
| "learning_rate": 7.607871485943775e-06, | |
| "loss": 0.2174, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 0.2432, | |
| "grad_norm": 6.811086177825928, | |
| "learning_rate": 7.599839357429719e-06, | |
| "loss": 0.2232, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 0.244, | |
| "grad_norm": 7.538053035736084, | |
| "learning_rate": 7.591807228915663e-06, | |
| "loss": 0.2003, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.2448, | |
| "grad_norm": 5.63380241394043, | |
| "learning_rate": 7.583775100401607e-06, | |
| "loss": 0.22, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 0.2456, | |
| "grad_norm": 7.308541297912598, | |
| "learning_rate": 7.575742971887551e-06, | |
| "loss": 0.194, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 0.2464, | |
| "grad_norm": 12.48105239868164, | |
| "learning_rate": 7.567710843373495e-06, | |
| "loss": 0.2254, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 0.2472, | |
| "grad_norm": 6.443123817443848, | |
| "learning_rate": 7.559678714859439e-06, | |
| "loss": 0.2157, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 0.248, | |
| "grad_norm": 5.197587013244629, | |
| "learning_rate": 7.551646586345383e-06, | |
| "loss": 0.2095, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.2488, | |
| "grad_norm": 5.799818515777588, | |
| "learning_rate": 7.543614457831327e-06, | |
| "loss": 0.2432, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 0.2496, | |
| "grad_norm": 6.902918338775635, | |
| "learning_rate": 7.535582329317269e-06, | |
| "loss": 0.2106, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 0.2504, | |
| "grad_norm": 12.562540054321289, | |
| "learning_rate": 7.527550200803213e-06, | |
| "loss": 0.2241, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 0.2512, | |
| "grad_norm": 6.103118419647217, | |
| "learning_rate": 7.519518072289157e-06, | |
| "loss": 0.2141, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 0.252, | |
| "grad_norm": 3.2933740615844727, | |
| "learning_rate": 7.511485943775101e-06, | |
| "loss": 0.2079, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.2528, | |
| "grad_norm": 4.528769493103027, | |
| "learning_rate": 7.503453815261045e-06, | |
| "loss": 0.2072, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 0.2536, | |
| "grad_norm": 5.217092990875244, | |
| "learning_rate": 7.4954216867469886e-06, | |
| "loss": 0.2165, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 0.2544, | |
| "grad_norm": 6.123150825500488, | |
| "learning_rate": 7.4873895582329325e-06, | |
| "loss": 0.1907, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 0.2552, | |
| "grad_norm": 4.618309020996094, | |
| "learning_rate": 7.479357429718876e-06, | |
| "loss": 0.2214, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 0.256, | |
| "grad_norm": 8.602248191833496, | |
| "learning_rate": 7.47132530120482e-06, | |
| "loss": 0.2075, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.256, | |
| "eval_test1_cer": 0.06202853088356113, | |
| "eval_test1_cer_norm": 0.046020121497346744, | |
| "eval_test1_loss": 0.21270166337490082, | |
| "eval_test1_runtime": 1209.2653, | |
| "eval_test1_samples_per_second": 2.067, | |
| "eval_test1_steps_per_second": 0.517, | |
| "eval_test1_wer": 0.18376629056240706, | |
| "eval_test1_wer_norm": 0.12262193518220871, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.256, | |
| "eval_test2_cer": 0.12037555530667861, | |
| "eval_test2_cer_norm": 0.09361442516268981, | |
| "eval_test2_loss": 0.3685109317302704, | |
| "eval_test2_runtime": 1222.5262, | |
| "eval_test2_samples_per_second": 2.045, | |
| "eval_test2_steps_per_second": 0.511, | |
| "eval_test2_wer": 0.2525177386129549, | |
| "eval_test2_wer_norm": 0.18381847352738942, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.2568, | |
| "grad_norm": 4.757772922515869, | |
| "learning_rate": 7.463293172690763e-06, | |
| "loss": 0.2147, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 0.2576, | |
| "grad_norm": 2.9504878520965576, | |
| "learning_rate": 7.455261044176707e-06, | |
| "loss": 0.225, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 0.2584, | |
| "grad_norm": 6.651636600494385, | |
| "learning_rate": 7.447228915662651e-06, | |
| "loss": 0.2233, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 0.2592, | |
| "grad_norm": 6.254730224609375, | |
| "learning_rate": 7.4392771084337355e-06, | |
| "loss": 0.2067, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 8.979650497436523, | |
| "learning_rate": 7.4312449799196795e-06, | |
| "loss": 0.2281, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.2608, | |
| "grad_norm": 4.179977893829346, | |
| "learning_rate": 7.423212851405623e-06, | |
| "loss": 0.1986, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 0.2616, | |
| "grad_norm": 5.462186336517334, | |
| "learning_rate": 7.415180722891567e-06, | |
| "loss": 0.2104, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 0.2624, | |
| "grad_norm": 7.806402683258057, | |
| "learning_rate": 7.40714859437751e-06, | |
| "loss": 0.2153, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 0.2632, | |
| "grad_norm": 5.667836666107178, | |
| "learning_rate": 7.399116465863454e-06, | |
| "loss": 0.1883, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 0.264, | |
| "grad_norm": 8.295492172241211, | |
| "learning_rate": 7.391084337349398e-06, | |
| "loss": 0.2139, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.2648, | |
| "grad_norm": 10.960628509521484, | |
| "learning_rate": 7.383052208835342e-06, | |
| "loss": 0.2415, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 0.2656, | |
| "grad_norm": 8.189926147460938, | |
| "learning_rate": 7.375020080321286e-06, | |
| "loss": 0.2168, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 0.2664, | |
| "grad_norm": 8.110308647155762, | |
| "learning_rate": 7.36698795180723e-06, | |
| "loss": 0.2169, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 0.2672, | |
| "grad_norm": 7.833503723144531, | |
| "learning_rate": 7.358955823293174e-06, | |
| "loss": 0.2131, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 0.268, | |
| "grad_norm": 4.76417350769043, | |
| "learning_rate": 7.350923694779117e-06, | |
| "loss": 0.2093, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.2688, | |
| "grad_norm": 6.401252269744873, | |
| "learning_rate": 7.342891566265061e-06, | |
| "loss": 0.2128, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 0.2696, | |
| "grad_norm": 4.773887634277344, | |
| "learning_rate": 7.334859437751004e-06, | |
| "loss": 0.2042, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 0.2704, | |
| "grad_norm": 3.329989433288574, | |
| "learning_rate": 7.326827309236948e-06, | |
| "loss": 0.1916, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 0.2712, | |
| "grad_norm": 5.542182445526123, | |
| "learning_rate": 7.318795180722892e-06, | |
| "loss": 0.2039, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 0.272, | |
| "grad_norm": 10.034430503845215, | |
| "learning_rate": 7.310763052208836e-06, | |
| "loss": 0.207, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.2728, | |
| "grad_norm": 6.576022624969482, | |
| "learning_rate": 7.30273092369478e-06, | |
| "loss": 0.2136, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 0.2736, | |
| "grad_norm": 5.687163829803467, | |
| "learning_rate": 7.2946987951807236e-06, | |
| "loss": 0.1901, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 0.2744, | |
| "grad_norm": 7.63690710067749, | |
| "learning_rate": 7.2866666666666675e-06, | |
| "loss": 0.2094, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 0.2752, | |
| "grad_norm": 17.66728401184082, | |
| "learning_rate": 7.278714859437752e-06, | |
| "loss": 0.2149, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 0.276, | |
| "grad_norm": 6.838101863861084, | |
| "learning_rate": 7.270682730923695e-06, | |
| "loss": 0.2181, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.2768, | |
| "grad_norm": 9.850558280944824, | |
| "learning_rate": 7.262650602409639e-06, | |
| "loss": 0.1968, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 0.2776, | |
| "grad_norm": 5.048129558563232, | |
| "learning_rate": 7.254618473895583e-06, | |
| "loss": 0.2264, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 0.2784, | |
| "grad_norm": 14.177302360534668, | |
| "learning_rate": 7.246586345381527e-06, | |
| "loss": 0.2079, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 0.2792, | |
| "grad_norm": 6.149969577789307, | |
| "learning_rate": 7.2385542168674706e-06, | |
| "loss": 0.2301, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 5.608641624450684, | |
| "learning_rate": 7.2305220883534145e-06, | |
| "loss": 0.2208, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.2808, | |
| "grad_norm": 6.874965667724609, | |
| "learning_rate": 7.222489959839358e-06, | |
| "loss": 0.208, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 0.2816, | |
| "grad_norm": 5.777594089508057, | |
| "learning_rate": 7.214457831325302e-06, | |
| "loss": 0.1997, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 0.2824, | |
| "grad_norm": 2.708245277404785, | |
| "learning_rate": 7.206425702811246e-06, | |
| "loss": 0.2111, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 0.2832, | |
| "grad_norm": 3.1390602588653564, | |
| "learning_rate": 7.1983935742971885e-06, | |
| "loss": 0.2094, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 0.284, | |
| "grad_norm": 4.028870105743408, | |
| "learning_rate": 7.190361445783132e-06, | |
| "loss": 0.2246, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.2848, | |
| "grad_norm": 15.425012588500977, | |
| "learning_rate": 7.182329317269076e-06, | |
| "loss": 0.1956, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 0.2856, | |
| "grad_norm": 5.734979152679443, | |
| "learning_rate": 7.17429718875502e-06, | |
| "loss": 0.2125, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 0.2864, | |
| "grad_norm": 5.9506916999816895, | |
| "learning_rate": 7.166265060240964e-06, | |
| "loss": 0.205, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 0.2872, | |
| "grad_norm": 6.654294013977051, | |
| "learning_rate": 7.158232931726908e-06, | |
| "loss": 0.2183, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 0.288, | |
| "grad_norm": 4.6381754875183105, | |
| "learning_rate": 7.150200803212852e-06, | |
| "loss": 0.2108, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.288, | |
| "eval_test1_cer": 0.05035850324893569, | |
| "eval_test1_cer_norm": 0.034854851489903235, | |
| "eval_test1_loss": 0.2031903713941574, | |
| "eval_test1_runtime": 1180.1384, | |
| "eval_test1_samples_per_second": 2.118, | |
| "eval_test1_steps_per_second": 0.53, | |
| "eval_test1_wer": 0.15638939910784572, | |
| "eval_test1_wer_norm": 0.09538560448873433, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.288, | |
| "eval_test2_cer": 0.15337197147870235, | |
| "eval_test2_cer_norm": 0.11982878834831112, | |
| "eval_test2_loss": 0.36121314764022827, | |
| "eval_test2_runtime": 1277.8248, | |
| "eval_test2_samples_per_second": 1.956, | |
| "eval_test2_steps_per_second": 0.489, | |
| "eval_test2_wer": 0.2875944151979858, | |
| "eval_test2_wer_norm": 0.22060508824203529, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.2888, | |
| "grad_norm": 6.482578277587891, | |
| "learning_rate": 7.142168674698796e-06, | |
| "loss": 0.2319, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 0.2896, | |
| "grad_norm": 6.14495849609375, | |
| "learning_rate": 7.13413654618474e-06, | |
| "loss": 0.2008, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 0.2904, | |
| "grad_norm": 7.641204357147217, | |
| "learning_rate": 7.126104417670683e-06, | |
| "loss": 0.2154, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 0.2912, | |
| "grad_norm": 10.709319114685059, | |
| "learning_rate": 7.118072289156627e-06, | |
| "loss": 0.1996, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 0.292, | |
| "grad_norm": 9.637004852294922, | |
| "learning_rate": 7.110120481927711e-06, | |
| "loss": 0.214, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.2928, | |
| "grad_norm": 2.7376303672790527, | |
| "learning_rate": 7.102088353413655e-06, | |
| "loss": 0.2064, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 0.2936, | |
| "grad_norm": 5.24441385269165, | |
| "learning_rate": 7.094056224899599e-06, | |
| "loss": 0.2105, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 0.2944, | |
| "grad_norm": 2.739924907684326, | |
| "learning_rate": 7.086024096385543e-06, | |
| "loss": 0.231, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 0.2952, | |
| "grad_norm": 10.428742408752441, | |
| "learning_rate": 7.077991967871487e-06, | |
| "loss": 0.1984, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 0.296, | |
| "grad_norm": 7.877098560333252, | |
| "learning_rate": 7.06995983935743e-06, | |
| "loss": 0.2054, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.2968, | |
| "grad_norm": 4.027400493621826, | |
| "learning_rate": 7.061927710843374e-06, | |
| "loss": 0.2088, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 0.2976, | |
| "grad_norm": 2.8548176288604736, | |
| "learning_rate": 7.053895582329318e-06, | |
| "loss": 0.2126, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 0.2984, | |
| "grad_norm": 8.31721019744873, | |
| "learning_rate": 7.045863453815262e-06, | |
| "loss": 0.2093, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 0.2992, | |
| "grad_norm": 10.732782363891602, | |
| "learning_rate": 7.0378313253012056e-06, | |
| "loss": 0.2148, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 6.584483623504639, | |
| "learning_rate": 7.0297991967871495e-06, | |
| "loss": 0.215, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.3008, | |
| "grad_norm": 5.9521684646606445, | |
| "learning_rate": 7.021767068273093e-06, | |
| "loss": 0.1889, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 0.3016, | |
| "grad_norm": 4.7626166343688965, | |
| "learning_rate": 7.013734939759037e-06, | |
| "loss": 0.2022, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 0.3024, | |
| "grad_norm": 4.184934616088867, | |
| "learning_rate": 7.005702811244981e-06, | |
| "loss": 0.2237, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 0.3032, | |
| "grad_norm": 5.2891998291015625, | |
| "learning_rate": 6.9976706827309235e-06, | |
| "loss": 0.1974, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 0.304, | |
| "grad_norm": 12.890066146850586, | |
| "learning_rate": 6.989638554216867e-06, | |
| "loss": 0.2317, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.3048, | |
| "grad_norm": 13.514941215515137, | |
| "learning_rate": 6.981606425702811e-06, | |
| "loss": 0.2175, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 0.3056, | |
| "grad_norm": 5.781461238861084, | |
| "learning_rate": 6.973574297188755e-06, | |
| "loss": 0.1852, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 0.3064, | |
| "grad_norm": 8.979798316955566, | |
| "learning_rate": 6.965542168674699e-06, | |
| "loss": 0.2156, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 0.3072, | |
| "grad_norm": 6.392210006713867, | |
| "learning_rate": 6.957510040160643e-06, | |
| "loss": 0.1815, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.308, | |
| "grad_norm": 7.205411434173584, | |
| "learning_rate": 6.949558232931728e-06, | |
| "loss": 0.2074, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.3088, | |
| "grad_norm": 6.6332621574401855, | |
| "learning_rate": 6.9415261044176704e-06, | |
| "loss": 0.2037, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 0.3096, | |
| "grad_norm": 6.583282470703125, | |
| "learning_rate": 6.933493975903614e-06, | |
| "loss": 0.2161, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 0.3104, | |
| "grad_norm": 5.363083362579346, | |
| "learning_rate": 6.925461847389558e-06, | |
| "loss": 0.1842, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 0.3112, | |
| "grad_norm": 11.732734680175781, | |
| "learning_rate": 6.917429718875502e-06, | |
| "loss": 0.1955, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 0.312, | |
| "grad_norm": 4.4274444580078125, | |
| "learning_rate": 6.909397590361446e-06, | |
| "loss": 0.2065, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.3128, | |
| "grad_norm": 8.530516624450684, | |
| "learning_rate": 6.90136546184739e-06, | |
| "loss": 0.2015, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 0.3136, | |
| "grad_norm": 4.273055076599121, | |
| "learning_rate": 6.893333333333334e-06, | |
| "loss": 0.1997, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 0.3144, | |
| "grad_norm": 5.483185291290283, | |
| "learning_rate": 6.885301204819278e-06, | |
| "loss": 0.1987, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 0.3152, | |
| "grad_norm": 8.107812881469727, | |
| "learning_rate": 6.877269076305222e-06, | |
| "loss": 0.2146, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 0.316, | |
| "grad_norm": 11.944705963134766, | |
| "learning_rate": 6.869236947791165e-06, | |
| "loss": 0.2243, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.3168, | |
| "grad_norm": 6.820212364196777, | |
| "learning_rate": 6.861204819277109e-06, | |
| "loss": 0.2106, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 0.3176, | |
| "grad_norm": 16.74178123474121, | |
| "learning_rate": 6.853172690763053e-06, | |
| "loss": 0.208, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 0.3184, | |
| "grad_norm": 8.6283597946167, | |
| "learning_rate": 6.845140562248997e-06, | |
| "loss": 0.191, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 0.3192, | |
| "grad_norm": 6.8946661949157715, | |
| "learning_rate": 6.8371084337349406e-06, | |
| "loss": 0.2055, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 6.649729251861572, | |
| "learning_rate": 6.8290763052208845e-06, | |
| "loss": 0.2005, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_test1_cer": 0.04994771827619688, | |
| "eval_test1_cer_norm": 0.03569524815713017, | |
| "eval_test1_loss": 0.20380495488643646, | |
| "eval_test1_runtime": 1179.5149, | |
| "eval_test1_samples_per_second": 2.12, | |
| "eval_test1_steps_per_second": 0.53, | |
| "eval_test1_wer": 0.16318260007580396, | |
| "eval_test1_wer_norm": 0.10292527543177767, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_test2_cer": 0.10917142643819763, | |
| "eval_test2_cer_norm": 0.08240529129222188, | |
| "eval_test2_loss": 0.36028367280960083, | |
| "eval_test2_runtime": 1209.014, | |
| "eval_test2_samples_per_second": 2.068, | |
| "eval_test2_steps_per_second": 0.517, | |
| "eval_test2_wer": 0.24101625085831999, | |
| "eval_test2_wer_norm": 0.17456451982580792, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.3208, | |
| "grad_norm": 6.898298740386963, | |
| "learning_rate": 6.821044176706828e-06, | |
| "loss": 0.1872, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 0.3216, | |
| "grad_norm": 5.272704124450684, | |
| "learning_rate": 6.813012048192772e-06, | |
| "loss": 0.207, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 0.3224, | |
| "grad_norm": 3.6596715450286865, | |
| "learning_rate": 6.804979919678716e-06, | |
| "loss": 0.2143, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 0.3232, | |
| "grad_norm": 3.9344301223754883, | |
| "learning_rate": 6.7969477911646585e-06, | |
| "loss": 0.1844, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 0.324, | |
| "grad_norm": 10.56173324584961, | |
| "learning_rate": 6.788915662650602e-06, | |
| "loss": 0.2096, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.3248, | |
| "grad_norm": 4.7323713302612305, | |
| "learning_rate": 6.780883534136546e-06, | |
| "loss": 0.2012, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 0.3256, | |
| "grad_norm": 13.328306198120117, | |
| "learning_rate": 6.7729317269076315e-06, | |
| "loss": 0.1891, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 0.3264, | |
| "grad_norm": 6.378972053527832, | |
| "learning_rate": 6.764899598393575e-06, | |
| "loss": 0.2113, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 0.3272, | |
| "grad_norm": 2.7503671646118164, | |
| "learning_rate": 6.756867469879519e-06, | |
| "loss": 0.2242, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 0.328, | |
| "grad_norm": 7.0376296043396, | |
| "learning_rate": 6.748835341365463e-06, | |
| "loss": 0.1938, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.3288, | |
| "grad_norm": 7.667542457580566, | |
| "learning_rate": 6.740803212851406e-06, | |
| "loss": 0.2199, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 0.3296, | |
| "grad_norm": 7.959465026855469, | |
| "learning_rate": 6.732771084337349e-06, | |
| "loss": 0.2046, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 0.3304, | |
| "grad_norm": 6.901830196380615, | |
| "learning_rate": 6.724738955823293e-06, | |
| "loss": 0.2257, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 0.3312, | |
| "grad_norm": 3.7971105575561523, | |
| "learning_rate": 6.716706827309237e-06, | |
| "loss": 0.1972, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 0.332, | |
| "grad_norm": 7.321595191955566, | |
| "learning_rate": 6.708674698795181e-06, | |
| "loss": 0.1896, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.3328, | |
| "grad_norm": 4.369442462921143, | |
| "learning_rate": 6.700642570281125e-06, | |
| "loss": 0.2164, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 0.3336, | |
| "grad_norm": 6.851342678070068, | |
| "learning_rate": 6.692610441767069e-06, | |
| "loss": 0.2097, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 0.3344, | |
| "grad_norm": 7.6141228675842285, | |
| "learning_rate": 6.684578313253013e-06, | |
| "loss": 0.2214, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 0.3352, | |
| "grad_norm": 2.4399728775024414, | |
| "learning_rate": 6.676546184738957e-06, | |
| "loss": 0.2117, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 0.336, | |
| "grad_norm": 4.702713966369629, | |
| "learning_rate": 6.668514056224901e-06, | |
| "loss": 0.2122, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.3368, | |
| "grad_norm": 13.61172866821289, | |
| "learning_rate": 6.660481927710844e-06, | |
| "loss": 0.2114, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 0.3376, | |
| "grad_norm": 8.276487350463867, | |
| "learning_rate": 6.652449799196788e-06, | |
| "loss": 0.2041, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 0.3384, | |
| "grad_norm": 7.3196539878845215, | |
| "learning_rate": 6.644417670682732e-06, | |
| "loss": 0.2107, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 0.3392, | |
| "grad_norm": 6.650379657745361, | |
| "learning_rate": 6.6363855421686756e-06, | |
| "loss": 0.2114, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 9.550190925598145, | |
| "learning_rate": 6.628353413654619e-06, | |
| "loss": 0.1889, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.3408, | |
| "grad_norm": 4.141082286834717, | |
| "learning_rate": 6.6203212851405626e-06, | |
| "loss": 0.2006, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 0.3416, | |
| "grad_norm": 8.406717300415039, | |
| "learning_rate": 6.6122891566265065e-06, | |
| "loss": 0.2016, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 0.3424, | |
| "grad_norm": 11.38437557220459, | |
| "learning_rate": 6.604337349397591e-06, | |
| "loss": 0.2128, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 0.3432, | |
| "grad_norm": 4.217543125152588, | |
| "learning_rate": 6.596305220883535e-06, | |
| "loss": 0.2126, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 0.344, | |
| "grad_norm": 7.4914703369140625, | |
| "learning_rate": 6.588273092369479e-06, | |
| "loss": 0.1981, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.3448, | |
| "grad_norm": 9.509758949279785, | |
| "learning_rate": 6.580240963855422e-06, | |
| "loss": 0.2041, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 0.3456, | |
| "grad_norm": 6.551538467407227, | |
| "learning_rate": 6.572208835341366e-06, | |
| "loss": 0.2226, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 0.3464, | |
| "grad_norm": 6.811914443969727, | |
| "learning_rate": 6.5641767068273095e-06, | |
| "loss": 0.2015, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 0.3472, | |
| "grad_norm": 4.834348201751709, | |
| "learning_rate": 6.5561445783132535e-06, | |
| "loss": 0.2036, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 0.348, | |
| "grad_norm": 10.421252250671387, | |
| "learning_rate": 6.548112449799197e-06, | |
| "loss": 0.2204, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.3488, | |
| "grad_norm": 7.1061882972717285, | |
| "learning_rate": 6.540080321285141e-06, | |
| "loss": 0.1971, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 0.3496, | |
| "grad_norm": 3.9507927894592285, | |
| "learning_rate": 6.532048192771084e-06, | |
| "loss": 0.217, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 0.3504, | |
| "grad_norm": 6.255137920379639, | |
| "learning_rate": 6.524016064257028e-06, | |
| "loss": 0.199, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 0.3512, | |
| "grad_norm": 6.308269500732422, | |
| "learning_rate": 6.515983935742972e-06, | |
| "loss": 0.2024, | |
| "step": 43900 | |
| }, | |
| { | |
| "epoch": 0.352, | |
| "grad_norm": 2.7685184478759766, | |
| "learning_rate": 6.507951807228916e-06, | |
| "loss": 0.2076, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.352, | |
| "eval_test1_cer": 0.05585275225931735, | |
| "eval_test1_cer_norm": 0.04021898336014599, | |
| "eval_test1_loss": 0.19787286221981049, | |
| "eval_test1_runtime": 1201.2365, | |
| "eval_test1_samples_per_second": 2.081, | |
| "eval_test1_steps_per_second": 0.52, | |
| "eval_test1_wer": 0.17079215137467565, | |
| "eval_test1_wer_norm": 0.10894532277389754, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.352, | |
| "eval_test2_cer": 0.11820565946167917, | |
| "eval_test2_cer_norm": 0.0912660753021382, | |
| "eval_test2_loss": 0.3588273823261261, | |
| "eval_test2_runtime": 1231.6351, | |
| "eval_test2_samples_per_second": 2.03, | |
| "eval_test2_steps_per_second": 0.507, | |
| "eval_test2_wer": 0.252431906614786, | |
| "eval_test2_wer_norm": 0.18519367407746964, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.3528, | |
| "grad_norm": 10.823402404785156, | |
| "learning_rate": 6.49991967871486e-06, | |
| "loss": 0.1976, | |
| "step": 44100 | |
| }, | |
| { | |
| "epoch": 0.3536, | |
| "grad_norm": 3.3830673694610596, | |
| "learning_rate": 6.491887550200804e-06, | |
| "loss": 0.1952, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 0.3544, | |
| "grad_norm": 8.530442237854004, | |
| "learning_rate": 6.483855421686748e-06, | |
| "loss": 0.2122, | |
| "step": 44300 | |
| }, | |
| { | |
| "epoch": 0.3552, | |
| "grad_norm": 4.178443908691406, | |
| "learning_rate": 6.475823293172692e-06, | |
| "loss": 0.2014, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 0.356, | |
| "grad_norm": 13.146326065063477, | |
| "learning_rate": 6.467791164658636e-06, | |
| "loss": 0.179, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.3568, | |
| "grad_norm": 9.41238021850586, | |
| "learning_rate": 6.459759036144578e-06, | |
| "loss": 0.2135, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 0.3576, | |
| "grad_norm": 4.255426406860352, | |
| "learning_rate": 6.451726907630522e-06, | |
| "loss": 0.1957, | |
| "step": 44700 | |
| }, | |
| { | |
| "epoch": 0.3584, | |
| "grad_norm": 6.541229248046875, | |
| "learning_rate": 6.443775100401607e-06, | |
| "loss": 0.2, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 0.3592, | |
| "grad_norm": 4.5669636726379395, | |
| "learning_rate": 6.435742971887551e-06, | |
| "loss": 0.187, | |
| "step": 44900 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 6.742759704589844, | |
| "learning_rate": 6.427710843373495e-06, | |
| "loss": 0.1961, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.3608, | |
| "grad_norm": 6.676318168640137, | |
| "learning_rate": 6.419678714859439e-06, | |
| "loss": 0.2023, | |
| "step": 45100 | |
| }, | |
| { | |
| "epoch": 0.3616, | |
| "grad_norm": 5.580842018127441, | |
| "learning_rate": 6.411646586345383e-06, | |
| "loss": 0.2257, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 0.3624, | |
| "grad_norm": 12.327486038208008, | |
| "learning_rate": 6.403614457831325e-06, | |
| "loss": 0.1922, | |
| "step": 45300 | |
| }, | |
| { | |
| "epoch": 0.3632, | |
| "grad_norm": 3.540501117706299, | |
| "learning_rate": 6.395582329317269e-06, | |
| "loss": 0.1914, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 0.364, | |
| "grad_norm": 5.7018351554870605, | |
| "learning_rate": 6.387550200803213e-06, | |
| "loss": 0.1921, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.3648, | |
| "grad_norm": 5.11850118637085, | |
| "learning_rate": 6.379518072289157e-06, | |
| "loss": 0.1847, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 0.3656, | |
| "grad_norm": 9.415154457092285, | |
| "learning_rate": 6.371485943775101e-06, | |
| "loss": 0.1992, | |
| "step": 45700 | |
| }, | |
| { | |
| "epoch": 0.3664, | |
| "grad_norm": 5.810830116271973, | |
| "learning_rate": 6.3634538152610445e-06, | |
| "loss": 0.2, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 0.3672, | |
| "grad_norm": 6.580953598022461, | |
| "learning_rate": 6.3554216867469885e-06, | |
| "loss": 0.1853, | |
| "step": 45900 | |
| }, | |
| { | |
| "epoch": 0.368, | |
| "grad_norm": 5.693816184997559, | |
| "learning_rate": 6.347389558232932e-06, | |
| "loss": 0.1935, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.3688, | |
| "grad_norm": 3.0731186866760254, | |
| "learning_rate": 6.339357429718876e-06, | |
| "loss": 0.1779, | |
| "step": 46100 | |
| }, | |
| { | |
| "epoch": 0.3696, | |
| "grad_norm": 7.017910003662109, | |
| "learning_rate": 6.331325301204819e-06, | |
| "loss": 0.189, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 0.3704, | |
| "grad_norm": 13.572850227355957, | |
| "learning_rate": 6.323293172690763e-06, | |
| "loss": 0.1949, | |
| "step": 46300 | |
| }, | |
| { | |
| "epoch": 0.3712, | |
| "grad_norm": 4.851452827453613, | |
| "learning_rate": 6.315261044176707e-06, | |
| "loss": 0.1865, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 0.372, | |
| "grad_norm": 3.416046619415283, | |
| "learning_rate": 6.307228915662651e-06, | |
| "loss": 0.2003, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.3728, | |
| "grad_norm": 7.885288715362549, | |
| "learning_rate": 6.299196787148595e-06, | |
| "loss": 0.2126, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 0.3736, | |
| "grad_norm": 3.072671890258789, | |
| "learning_rate": 6.291164658634539e-06, | |
| "loss": 0.2006, | |
| "step": 46700 | |
| }, | |
| { | |
| "epoch": 0.3744, | |
| "grad_norm": 3.8454980850219727, | |
| "learning_rate": 6.283132530120483e-06, | |
| "loss": 0.1911, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 0.3752, | |
| "grad_norm": 1.3022770881652832, | |
| "learning_rate": 6.275100401606427e-06, | |
| "loss": 0.1904, | |
| "step": 46900 | |
| }, | |
| { | |
| "epoch": 0.376, | |
| "grad_norm": 9.75496768951416, | |
| "learning_rate": 6.267068273092371e-06, | |
| "loss": 0.1902, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.3768, | |
| "grad_norm": 5.822065353393555, | |
| "learning_rate": 6.259116465863454e-06, | |
| "loss": 0.2156, | |
| "step": 47100 | |
| }, | |
| { | |
| "epoch": 0.3776, | |
| "grad_norm": 11.050053596496582, | |
| "learning_rate": 6.251084337349398e-06, | |
| "loss": 0.2068, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 0.3784, | |
| "grad_norm": 11.641470909118652, | |
| "learning_rate": 6.243052208835342e-06, | |
| "loss": 0.1784, | |
| "step": 47300 | |
| }, | |
| { | |
| "epoch": 0.3792, | |
| "grad_norm": 6.159931182861328, | |
| "learning_rate": 6.235020080321286e-06, | |
| "loss": 0.1869, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 5.114069938659668, | |
| "learning_rate": 6.22698795180723e-06, | |
| "loss": 0.1897, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.3808, | |
| "grad_norm": 4.50574254989624, | |
| "learning_rate": 6.218955823293174e-06, | |
| "loss": 0.1936, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 0.3816, | |
| "grad_norm": 8.144811630249023, | |
| "learning_rate": 6.210923694779118e-06, | |
| "loss": 0.2113, | |
| "step": 47700 | |
| }, | |
| { | |
| "epoch": 0.3824, | |
| "grad_norm": 4.306349754333496, | |
| "learning_rate": 6.202891566265062e-06, | |
| "loss": 0.1951, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 0.3832, | |
| "grad_norm": 6.88608455657959, | |
| "learning_rate": 6.194859437751004e-06, | |
| "loss": 0.2107, | |
| "step": 47900 | |
| }, | |
| { | |
| "epoch": 0.384, | |
| "grad_norm": 10.85550594329834, | |
| "learning_rate": 6.186827309236948e-06, | |
| "loss": 0.2117, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.384, | |
| "eval_test1_cer": 0.05253379640002988, | |
| "eval_test1_cer_norm": 0.035930559223953704, | |
| "eval_test1_loss": 0.1944696605205536, | |
| "eval_test1_runtime": 2131.2533, | |
| "eval_test1_samples_per_second": 1.173, | |
| "eval_test1_steps_per_second": 0.293, | |
| "eval_test1_wer": 0.15726406017668154, | |
| "eval_test1_wer_norm": 0.09462579268827259, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.384, | |
| "eval_test2_cer": 0.11918094598125957, | |
| "eval_test2_cer_norm": 0.08796870158041524, | |
| "eval_test2_loss": 0.3547360897064209, | |
| "eval_test2_runtime": 2218.4059, | |
| "eval_test2_samples_per_second": 1.127, | |
| "eval_test2_steps_per_second": 0.282, | |
| "eval_test2_wer": 0.2637045090409705, | |
| "eval_test2_wer_norm": 0.1967396286958515, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.3848, | |
| "grad_norm": 8.570365905761719, | |
| "learning_rate": 6.178795180722892e-06, | |
| "loss": 0.2176, | |
| "step": 48100 | |
| }, | |
| { | |
| "epoch": 0.3856, | |
| "grad_norm": 6.4876604080200195, | |
| "learning_rate": 6.170763052208836e-06, | |
| "loss": 0.2062, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 0.3864, | |
| "grad_norm": 4.121545314788818, | |
| "learning_rate": 6.1627309236947796e-06, | |
| "loss": 0.1855, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 0.3872, | |
| "grad_norm": 8.148366928100586, | |
| "learning_rate": 6.1546987951807235e-06, | |
| "loss": 0.1953, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 0.388, | |
| "grad_norm": 5.750129222869873, | |
| "learning_rate": 6.146666666666667e-06, | |
| "loss": 0.1947, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.3888, | |
| "grad_norm": 7.019141674041748, | |
| "learning_rate": 6.138634538152611e-06, | |
| "loss": 0.1868, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 0.3896, | |
| "grad_norm": 4.4510273933410645, | |
| "learning_rate": 6.130602409638555e-06, | |
| "loss": 0.2102, | |
| "step": 48700 | |
| }, | |
| { | |
| "epoch": 0.3904, | |
| "grad_norm": 7.094069480895996, | |
| "learning_rate": 6.122570281124498e-06, | |
| "loss": 0.1899, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 0.3912, | |
| "grad_norm": 6.22620153427124, | |
| "learning_rate": 6.114538152610442e-06, | |
| "loss": 0.2206, | |
| "step": 48900 | |
| }, | |
| { | |
| "epoch": 0.392, | |
| "grad_norm": 7.25572395324707, | |
| "learning_rate": 6.106506024096386e-06, | |
| "loss": 0.1914, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.3928, | |
| "grad_norm": 5.544373512268066, | |
| "learning_rate": 6.0985542168674705e-06, | |
| "loss": 0.1877, | |
| "step": 49100 | |
| }, | |
| { | |
| "epoch": 1.000792, | |
| "grad_norm": 5.080130577087402, | |
| "learning_rate": 6.090522088353414e-06, | |
| "loss": 0.1799, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 1.001592, | |
| "grad_norm": 2.341817617416382, | |
| "learning_rate": 6.082489959839358e-06, | |
| "loss": 0.1597, | |
| "step": 49300 | |
| }, | |
| { | |
| "epoch": 1.002392, | |
| "grad_norm": 3.7877089977264404, | |
| "learning_rate": 6.074457831325302e-06, | |
| "loss": 0.1463, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 1.003192, | |
| "grad_norm": 2.4026944637298584, | |
| "learning_rate": 6.066425702811245e-06, | |
| "loss": 0.1515, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.003992, | |
| "grad_norm": 4.575283527374268, | |
| "learning_rate": 6.058393574297189e-06, | |
| "loss": 0.1579, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 1.004792, | |
| "grad_norm": 5.175034999847412, | |
| "learning_rate": 6.050361445783133e-06, | |
| "loss": 0.1313, | |
| "step": 49700 | |
| }, | |
| { | |
| "epoch": 1.005592, | |
| "grad_norm": 4.901139259338379, | |
| "learning_rate": 6.042329317269077e-06, | |
| "loss": 0.1501, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 1.006392, | |
| "grad_norm": 3.304086685180664, | |
| "learning_rate": 6.034297188755021e-06, | |
| "loss": 0.1295, | |
| "step": 49900 | |
| }, | |
| { | |
| "epoch": 1.007192, | |
| "grad_norm": 3.6357131004333496, | |
| "learning_rate": 6.026265060240965e-06, | |
| "loss": 0.127, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.007992, | |
| "grad_norm": 23.502784729003906, | |
| "learning_rate": 6.018232931726908e-06, | |
| "loss": 0.1354, | |
| "step": 50100 | |
| }, | |
| { | |
| "epoch": 1.008792, | |
| "grad_norm": 2.340665817260742, | |
| "learning_rate": 6.010200803212852e-06, | |
| "loss": 0.1405, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 1.009592, | |
| "grad_norm": 4.528487205505371, | |
| "learning_rate": 6.002168674698796e-06, | |
| "loss": 0.1299, | |
| "step": 50300 | |
| }, | |
| { | |
| "epoch": 1.010392, | |
| "grad_norm": 3.589071035385132, | |
| "learning_rate": 5.994136546184739e-06, | |
| "loss": 0.1478, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 1.011192, | |
| "grad_norm": 11.359587669372559, | |
| "learning_rate": 5.986104417670683e-06, | |
| "loss": 0.133, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.011992, | |
| "grad_norm": 2.8059144020080566, | |
| "learning_rate": 5.978072289156627e-06, | |
| "loss": 0.1366, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 1.012792, | |
| "grad_norm": 8.648773193359375, | |
| "learning_rate": 5.970040160642571e-06, | |
| "loss": 0.1499, | |
| "step": 50700 | |
| }, | |
| { | |
| "epoch": 1.013592, | |
| "grad_norm": 7.918379306793213, | |
| "learning_rate": 5.9620080321285146e-06, | |
| "loss": 0.1493, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 1.014392, | |
| "grad_norm": 5.396472454071045, | |
| "learning_rate": 5.9539759036144585e-06, | |
| "loss": 0.1433, | |
| "step": 50900 | |
| }, | |
| { | |
| "epoch": 1.015192, | |
| "grad_norm": 6.0125412940979, | |
| "learning_rate": 5.945943775100402e-06, | |
| "loss": 0.1216, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.015992, | |
| "grad_norm": 5.78593111038208, | |
| "learning_rate": 5.937911646586346e-06, | |
| "loss": 0.1524, | |
| "step": 51100 | |
| }, | |
| { | |
| "epoch": 1.016792, | |
| "grad_norm": 2.646674156188965, | |
| "learning_rate": 5.92995983935743e-06, | |
| "loss": 0.1303, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 1.017592, | |
| "grad_norm": 4.267711639404297, | |
| "learning_rate": 5.921927710843374e-06, | |
| "loss": 0.1298, | |
| "step": 51300 | |
| }, | |
| { | |
| "epoch": 1.018392, | |
| "grad_norm": 6.027831554412842, | |
| "learning_rate": 5.913895582329318e-06, | |
| "loss": 0.1361, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 1.019192, | |
| "grad_norm": 7.927816390991211, | |
| "learning_rate": 5.9058634538152615e-06, | |
| "loss": 0.1341, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.019992, | |
| "grad_norm": 9.55642318725586, | |
| "learning_rate": 5.8978313253012055e-06, | |
| "loss": 0.1351, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 1.020792, | |
| "grad_norm": 5.5017409324646, | |
| "learning_rate": 5.889799196787149e-06, | |
| "loss": 0.1325, | |
| "step": 51700 | |
| }, | |
| { | |
| "epoch": 1.021592, | |
| "grad_norm": 5.213791370391846, | |
| "learning_rate": 5.881767068273093e-06, | |
| "loss": 0.1418, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 1.022392, | |
| "grad_norm": 2.497223138809204, | |
| "learning_rate": 5.873734939759037e-06, | |
| "loss": 0.1407, | |
| "step": 51900 | |
| }, | |
| { | |
| "epoch": 1.023192, | |
| "grad_norm": 3.251509428024292, | |
| "learning_rate": 5.8657028112449794e-06, | |
| "loss": 0.133, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.023192, | |
| "eval_test1_cer": 0.04808518186571066, | |
| "eval_test1_cer_norm": 0.032487334021658223, | |
| "eval_test1_loss": 0.19697345793247223, | |
| "eval_test1_runtime": 1190.0899, | |
| "eval_test1_samples_per_second": 2.101, | |
| "eval_test1_steps_per_second": 0.525, | |
| "eval_test1_wer": 0.1493629551881979, | |
| "eval_test1_wer_norm": 0.08699845115286829, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.023192, | |
| "eval_test2_cer": 0.11403385970806734, | |
| "eval_test2_cer_norm": 0.08345115432290053, | |
| "eval_test2_loss": 0.36047160625457764, | |
| "eval_test2_runtime": 1226.7275, | |
| "eval_test2_samples_per_second": 2.038, | |
| "eval_test2_steps_per_second": 0.509, | |
| "eval_test2_wer": 0.24490730144197756, | |
| "eval_test2_wer_norm": 0.17745817098326838, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.023992, | |
| "grad_norm": 7.025960922241211, | |
| "learning_rate": 5.857670682730923e-06, | |
| "loss": 0.1494, | |
| "step": 52100 | |
| }, | |
| { | |
| "epoch": 1.024792, | |
| "grad_norm": 4.487267971038818, | |
| "learning_rate": 5.849638554216867e-06, | |
| "loss": 0.1403, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 1.025592, | |
| "grad_norm": 4.388223648071289, | |
| "learning_rate": 5.841606425702811e-06, | |
| "loss": 0.1172, | |
| "step": 52300 | |
| }, | |
| { | |
| "epoch": 1.026392, | |
| "grad_norm": 6.941164493560791, | |
| "learning_rate": 5.833574297188755e-06, | |
| "loss": 0.1218, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 1.027192, | |
| "grad_norm": 6.419257640838623, | |
| "learning_rate": 5.825542168674699e-06, | |
| "loss": 0.1317, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.027992, | |
| "grad_norm": 6.4062323570251465, | |
| "learning_rate": 5.817510040160643e-06, | |
| "loss": 0.1345, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 1.028792, | |
| "grad_norm": 8.374984741210938, | |
| "learning_rate": 5.809477911646587e-06, | |
| "loss": 0.126, | |
| "step": 52700 | |
| }, | |
| { | |
| "epoch": 1.029592, | |
| "grad_norm": 9.091133117675781, | |
| "learning_rate": 5.801445783132531e-06, | |
| "loss": 0.1247, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 1.030392, | |
| "grad_norm": 6.033024311065674, | |
| "learning_rate": 5.793413654618474e-06, | |
| "loss": 0.1273, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 1.031192, | |
| "grad_norm": 4.625008583068848, | |
| "learning_rate": 5.785381526104418e-06, | |
| "loss": 0.1448, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.031992, | |
| "grad_norm": 3.7176895141601562, | |
| "learning_rate": 5.777349397590362e-06, | |
| "loss": 0.1105, | |
| "step": 53100 | |
| }, | |
| { | |
| "epoch": 1.032792, | |
| "grad_norm": 4.577524185180664, | |
| "learning_rate": 5.769317269076306e-06, | |
| "loss": 0.1568, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 1.033592, | |
| "grad_norm": 3.0853545665740967, | |
| "learning_rate": 5.7612851405622496e-06, | |
| "loss": 0.1369, | |
| "step": 53300 | |
| }, | |
| { | |
| "epoch": 1.034392, | |
| "grad_norm": 5.356490135192871, | |
| "learning_rate": 5.7532530120481935e-06, | |
| "loss": 0.1373, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 1.035192, | |
| "grad_norm": 7.52249813079834, | |
| "learning_rate": 5.745220883534137e-06, | |
| "loss": 0.1256, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.035992, | |
| "grad_norm": 6.246498107910156, | |
| "learning_rate": 5.737188755020081e-06, | |
| "loss": 0.1473, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 1.036792, | |
| "grad_norm": 6.8541693687438965, | |
| "learning_rate": 5.729236947791165e-06, | |
| "loss": 0.1281, | |
| "step": 53700 | |
| }, | |
| { | |
| "epoch": 1.037592, | |
| "grad_norm": 2.0510551929473877, | |
| "learning_rate": 5.721204819277109e-06, | |
| "loss": 0.1243, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 1.038392, | |
| "grad_norm": 6.91987419128418, | |
| "learning_rate": 5.713172690763053e-06, | |
| "loss": 0.1199, | |
| "step": 53900 | |
| }, | |
| { | |
| "epoch": 1.039192, | |
| "grad_norm": 3.939608097076416, | |
| "learning_rate": 5.7051405622489965e-06, | |
| "loss": 0.1334, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.039992, | |
| "grad_norm": 8.798985481262207, | |
| "learning_rate": 5.6971084337349405e-06, | |
| "loss": 0.1119, | |
| "step": 54100 | |
| }, | |
| { | |
| "epoch": 1.040792, | |
| "grad_norm": 6.9459123611450195, | |
| "learning_rate": 5.689076305220884e-06, | |
| "loss": 0.1324, | |
| "step": 54200 | |
| }, | |
| { | |
| "epoch": 1.041592, | |
| "grad_norm": 7.139726161956787, | |
| "learning_rate": 5.681044176706828e-06, | |
| "loss": 0.1342, | |
| "step": 54300 | |
| }, | |
| { | |
| "epoch": 1.042392, | |
| "grad_norm": 1.988398790359497, | |
| "learning_rate": 5.673012048192772e-06, | |
| "loss": 0.123, | |
| "step": 54400 | |
| }, | |
| { | |
| "epoch": 1.043192, | |
| "grad_norm": 1.8150029182434082, | |
| "learning_rate": 5.664979919678716e-06, | |
| "loss": 0.1376, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.043992, | |
| "grad_norm": 8.445022583007812, | |
| "learning_rate": 5.656947791164658e-06, | |
| "loss": 0.1393, | |
| "step": 54600 | |
| }, | |
| { | |
| "epoch": 1.044792, | |
| "grad_norm": 6.788310527801514, | |
| "learning_rate": 5.648915662650602e-06, | |
| "loss": 0.1412, | |
| "step": 54700 | |
| }, | |
| { | |
| "epoch": 1.045592, | |
| "grad_norm": 6.29990816116333, | |
| "learning_rate": 5.640883534136546e-06, | |
| "loss": 0.1312, | |
| "step": 54800 | |
| }, | |
| { | |
| "epoch": 1.046392, | |
| "grad_norm": 1.2310799360275269, | |
| "learning_rate": 5.63285140562249e-06, | |
| "loss": 0.1331, | |
| "step": 54900 | |
| }, | |
| { | |
| "epoch": 1.047192, | |
| "grad_norm": 5.38347053527832, | |
| "learning_rate": 5.624819277108434e-06, | |
| "loss": 0.1295, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.047992, | |
| "grad_norm": 3.4652884006500244, | |
| "learning_rate": 5.616787148594378e-06, | |
| "loss": 0.1215, | |
| "step": 55100 | |
| }, | |
| { | |
| "epoch": 1.048792, | |
| "grad_norm": 7.579988479614258, | |
| "learning_rate": 5.608755020080322e-06, | |
| "loss": 0.1388, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 1.049592, | |
| "grad_norm": 0.8664081692695618, | |
| "learning_rate": 5.600722891566266e-06, | |
| "loss": 0.1199, | |
| "step": 55300 | |
| }, | |
| { | |
| "epoch": 1.050392, | |
| "grad_norm": 6.584039211273193, | |
| "learning_rate": 5.59269076305221e-06, | |
| "loss": 0.123, | |
| "step": 55400 | |
| }, | |
| { | |
| "epoch": 1.051192, | |
| "grad_norm": 4.955043792724609, | |
| "learning_rate": 5.584658634538153e-06, | |
| "loss": 0.1223, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.051992, | |
| "grad_norm": 2.857651710510254, | |
| "learning_rate": 5.576626506024097e-06, | |
| "loss": 0.1334, | |
| "step": 55600 | |
| }, | |
| { | |
| "epoch": 1.052792, | |
| "grad_norm": 6.896506309509277, | |
| "learning_rate": 5.568594377510041e-06, | |
| "loss": 0.1377, | |
| "step": 55700 | |
| }, | |
| { | |
| "epoch": 1.053592, | |
| "grad_norm": 7.342093467712402, | |
| "learning_rate": 5.5605622489959846e-06, | |
| "loss": 0.1272, | |
| "step": 55800 | |
| }, | |
| { | |
| "epoch": 1.054392, | |
| "grad_norm": 4.082319259643555, | |
| "learning_rate": 5.552610441767069e-06, | |
| "loss": 0.1355, | |
| "step": 55900 | |
| }, | |
| { | |
| "epoch": 1.055192, | |
| "grad_norm": 5.364993572235107, | |
| "learning_rate": 5.544578313253013e-06, | |
| "loss": 0.1305, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.055192, | |
| "eval_test1_cer": 0.05301460153857644, | |
| "eval_test1_cer_norm": 0.03525824189017216, | |
| "eval_test1_loss": 0.19662749767303467, | |
| "eval_test1_runtime": 1188.1388, | |
| "eval_test1_samples_per_second": 2.104, | |
| "eval_test1_steps_per_second": 0.526, | |
| "eval_test1_wer": 0.15286159946354122, | |
| "eval_test1_wer_norm": 0.09231713375610041, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.055192, | |
| "eval_test2_cer": 0.1047943032067794, | |
| "eval_test2_cer_norm": 0.08095754570808801, | |
| "eval_test2_loss": 0.36192458868026733, | |
| "eval_test2_runtime": 1219.4681, | |
| "eval_test2_samples_per_second": 2.05, | |
| "eval_test2_steps_per_second": 0.513, | |
| "eval_test2_wer": 0.22871366445410848, | |
| "eval_test2_wer_norm": 0.1619012147604859, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.055992, | |
| "grad_norm": 5.108110427856445, | |
| "learning_rate": 5.536546184738957e-06, | |
| "loss": 0.1304, | |
| "step": 56100 | |
| }, | |
| { | |
| "epoch": 1.056792, | |
| "grad_norm": 1.9956285953521729, | |
| "learning_rate": 5.5285140562249e-06, | |
| "loss": 0.1325, | |
| "step": 56200 | |
| }, | |
| { | |
| "epoch": 1.057592, | |
| "grad_norm": 7.1683220863342285, | |
| "learning_rate": 5.520481927710844e-06, | |
| "loss": 0.1315, | |
| "step": 56300 | |
| }, | |
| { | |
| "epoch": 1.058392, | |
| "grad_norm": 6.5139546394348145, | |
| "learning_rate": 5.512449799196788e-06, | |
| "loss": 0.1287, | |
| "step": 56400 | |
| }, | |
| { | |
| "epoch": 1.059192, | |
| "grad_norm": 6.996273040771484, | |
| "learning_rate": 5.5044176706827315e-06, | |
| "loss": 0.1273, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 1.059992, | |
| "grad_norm": 1.6335968971252441, | |
| "learning_rate": 5.4963855421686755e-06, | |
| "loss": 0.1157, | |
| "step": 56600 | |
| }, | |
| { | |
| "epoch": 1.060792, | |
| "grad_norm": 4.1844892501831055, | |
| "learning_rate": 5.488353413654619e-06, | |
| "loss": 0.1316, | |
| "step": 56700 | |
| }, | |
| { | |
| "epoch": 1.061592, | |
| "grad_norm": 8.634051322937012, | |
| "learning_rate": 5.480321285140563e-06, | |
| "loss": 0.1083, | |
| "step": 56800 | |
| }, | |
| { | |
| "epoch": 1.062392, | |
| "grad_norm": 8.461755752563477, | |
| "learning_rate": 5.472289156626507e-06, | |
| "loss": 0.1307, | |
| "step": 56900 | |
| }, | |
| { | |
| "epoch": 1.063192, | |
| "grad_norm": 5.679012298583984, | |
| "learning_rate": 5.464257028112451e-06, | |
| "loss": 0.133, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.063992, | |
| "grad_norm": 5.279029846191406, | |
| "learning_rate": 5.456224899598393e-06, | |
| "loss": 0.1203, | |
| "step": 57100 | |
| }, | |
| { | |
| "epoch": 1.064792, | |
| "grad_norm": 6.086945533752441, | |
| "learning_rate": 5.448192771084337e-06, | |
| "loss": 0.1371, | |
| "step": 57200 | |
| }, | |
| { | |
| "epoch": 1.065592, | |
| "grad_norm": 4.381275177001953, | |
| "learning_rate": 5.440160642570281e-06, | |
| "loss": 0.134, | |
| "step": 57300 | |
| }, | |
| { | |
| "epoch": 1.066392, | |
| "grad_norm": 3.3568952083587646, | |
| "learning_rate": 5.432128514056225e-06, | |
| "loss": 0.1294, | |
| "step": 57400 | |
| }, | |
| { | |
| "epoch": 1.067192, | |
| "grad_norm": 5.998335838317871, | |
| "learning_rate": 5.424096385542169e-06, | |
| "loss": 0.1074, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 1.067992, | |
| "grad_norm": 5.995476245880127, | |
| "learning_rate": 5.416064257028113e-06, | |
| "loss": 0.1129, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 1.068792, | |
| "grad_norm": 7.281689167022705, | |
| "learning_rate": 5.408112449799197e-06, | |
| "loss": 0.1342, | |
| "step": 57700 | |
| }, | |
| { | |
| "epoch": 1.069592, | |
| "grad_norm": 3.4488370418548584, | |
| "learning_rate": 5.40008032128514e-06, | |
| "loss": 0.1126, | |
| "step": 57800 | |
| }, | |
| { | |
| "epoch": 1.070392, | |
| "grad_norm": 6.784258842468262, | |
| "learning_rate": 5.392048192771084e-06, | |
| "loss": 0.129, | |
| "step": 57900 | |
| }, | |
| { | |
| "epoch": 1.071192, | |
| "grad_norm": 3.7626733779907227, | |
| "learning_rate": 5.384016064257028e-06, | |
| "loss": 0.1323, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.071992, | |
| "grad_norm": 6.320082187652588, | |
| "learning_rate": 5.375983935742972e-06, | |
| "loss": 0.1275, | |
| "step": 58100 | |
| }, | |
| { | |
| "epoch": 1.072792, | |
| "grad_norm": 10.221648216247559, | |
| "learning_rate": 5.367951807228916e-06, | |
| "loss": 0.1229, | |
| "step": 58200 | |
| }, | |
| { | |
| "epoch": 1.073592, | |
| "grad_norm": 3.2260794639587402, | |
| "learning_rate": 5.35991967871486e-06, | |
| "loss": 0.1237, | |
| "step": 58300 | |
| }, | |
| { | |
| "epoch": 1.074392, | |
| "grad_norm": 7.026370525360107, | |
| "learning_rate": 5.351887550200804e-06, | |
| "loss": 0.1351, | |
| "step": 58400 | |
| }, | |
| { | |
| "epoch": 1.075192, | |
| "grad_norm": 3.7613117694854736, | |
| "learning_rate": 5.343855421686748e-06, | |
| "loss": 0.1246, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.075992, | |
| "grad_norm": 19.284780502319336, | |
| "learning_rate": 5.335823293172692e-06, | |
| "loss": 0.1211, | |
| "step": 58600 | |
| }, | |
| { | |
| "epoch": 1.076792, | |
| "grad_norm": 2.5568318367004395, | |
| "learning_rate": 5.327791164658635e-06, | |
| "loss": 0.1386, | |
| "step": 58700 | |
| }, | |
| { | |
| "epoch": 1.077592, | |
| "grad_norm": 8.899717330932617, | |
| "learning_rate": 5.319759036144579e-06, | |
| "loss": 0.1297, | |
| "step": 58800 | |
| }, | |
| { | |
| "epoch": 1.078392, | |
| "grad_norm": 7.388535022735596, | |
| "learning_rate": 5.311726907630523e-06, | |
| "loss": 0.1218, | |
| "step": 58900 | |
| }, | |
| { | |
| "epoch": 1.079192, | |
| "grad_norm": 7.612710952758789, | |
| "learning_rate": 5.3036947791164666e-06, | |
| "loss": 0.1342, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.079992, | |
| "grad_norm": 6.390445232391357, | |
| "learning_rate": 5.29566265060241e-06, | |
| "loss": 0.138, | |
| "step": 59100 | |
| }, | |
| { | |
| "epoch": 1.080792, | |
| "grad_norm": 2.2183034420013428, | |
| "learning_rate": 5.2876305220883535e-06, | |
| "loss": 0.1212, | |
| "step": 59200 | |
| }, | |
| { | |
| "epoch": 1.081592, | |
| "grad_norm": 4.985264778137207, | |
| "learning_rate": 5.2795983935742975e-06, | |
| "loss": 0.1187, | |
| "step": 59300 | |
| }, | |
| { | |
| "epoch": 1.082392, | |
| "grad_norm": 10.64930534362793, | |
| "learning_rate": 5.271566265060241e-06, | |
| "loss": 0.1109, | |
| "step": 59400 | |
| }, | |
| { | |
| "epoch": 1.083192, | |
| "grad_norm": 6.063853740692139, | |
| "learning_rate": 5.263534136546185e-06, | |
| "loss": 0.1136, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.083992, | |
| "grad_norm": 6.947314739227295, | |
| "learning_rate": 5.255502008032128e-06, | |
| "loss": 0.1289, | |
| "step": 59600 | |
| }, | |
| { | |
| "epoch": 1.084792, | |
| "grad_norm": 2.270883560180664, | |
| "learning_rate": 5.247469879518072e-06, | |
| "loss": 0.1298, | |
| "step": 59700 | |
| }, | |
| { | |
| "epoch": 1.0855920000000001, | |
| "grad_norm": 0.7966949939727783, | |
| "learning_rate": 5.239437751004016e-06, | |
| "loss": 0.1272, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 1.086392, | |
| "grad_norm": 2.719560384750366, | |
| "learning_rate": 5.23140562248996e-06, | |
| "loss": 0.0987, | |
| "step": 59900 | |
| }, | |
| { | |
| "epoch": 1.087192, | |
| "grad_norm": 6.565618515014648, | |
| "learning_rate": 5.223373493975904e-06, | |
| "loss": 0.1165, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.087192, | |
| "eval_test1_cer": 0.05625886922100232, | |
| "eval_test1_cer_norm": 0.0387686988258458, | |
| "eval_test1_loss": 0.199602872133255, | |
| "eval_test1_runtime": 1190.6217, | |
| "eval_test1_samples_per_second": 2.1, | |
| "eval_test1_steps_per_second": 0.525, | |
| "eval_test1_wer": 0.1555438934079711, | |
| "eval_test1_wer_norm": 0.09591162804290014, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.087192, | |
| "eval_test2_cer": 0.12617127711203197, | |
| "eval_test2_cer_norm": 0.09629202819956616, | |
| "eval_test2_loss": 0.36100009083747864, | |
| "eval_test2_runtime": 1239.0925, | |
| "eval_test2_samples_per_second": 2.018, | |
| "eval_test2_steps_per_second": 0.504, | |
| "eval_test2_wer": 0.25226024261844815, | |
| "eval_test2_wer_norm": 0.18728512491404997, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.087992, | |
| "grad_norm": 6.482507228851318, | |
| "learning_rate": 5.215341365461848e-06, | |
| "loss": 0.1414, | |
| "step": 60100 | |
| }, | |
| { | |
| "epoch": 1.088792, | |
| "grad_norm": 6.777385234832764, | |
| "learning_rate": 5.207309236947792e-06, | |
| "loss": 0.1258, | |
| "step": 60200 | |
| }, | |
| { | |
| "epoch": 1.0895920000000001, | |
| "grad_norm": 6.040877819061279, | |
| "learning_rate": 5.199277108433736e-06, | |
| "loss": 0.1167, | |
| "step": 60300 | |
| }, | |
| { | |
| "epoch": 1.090392, | |
| "grad_norm": 4.490113258361816, | |
| "learning_rate": 5.19124497991968e-06, | |
| "loss": 0.113, | |
| "step": 60400 | |
| }, | |
| { | |
| "epoch": 1.091192, | |
| "grad_norm": 7.03712797164917, | |
| "learning_rate": 5.183212851405624e-06, | |
| "loss": 0.1084, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.091992, | |
| "grad_norm": 2.478569984436035, | |
| "learning_rate": 5.175180722891566e-06, | |
| "loss": 0.1162, | |
| "step": 60600 | |
| }, | |
| { | |
| "epoch": 1.092792, | |
| "grad_norm": 2.2656235694885254, | |
| "learning_rate": 5.16714859437751e-06, | |
| "loss": 0.1138, | |
| "step": 60700 | |
| }, | |
| { | |
| "epoch": 1.093592, | |
| "grad_norm": 6.424910545349121, | |
| "learning_rate": 5.159116465863454e-06, | |
| "loss": 0.1212, | |
| "step": 60800 | |
| }, | |
| { | |
| "epoch": 1.094392, | |
| "grad_norm": 4.551820278167725, | |
| "learning_rate": 5.151084337349398e-06, | |
| "loss": 0.1148, | |
| "step": 60900 | |
| }, | |
| { | |
| "epoch": 1.095192, | |
| "grad_norm": 4.301316738128662, | |
| "learning_rate": 5.1430522088353416e-06, | |
| "loss": 0.1316, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.095992, | |
| "grad_norm": 5.7936530113220215, | |
| "learning_rate": 5.1350200803212855e-06, | |
| "loss": 0.1036, | |
| "step": 61100 | |
| }, | |
| { | |
| "epoch": 1.096792, | |
| "grad_norm": 7.100841045379639, | |
| "learning_rate": 5.126987951807229e-06, | |
| "loss": 0.1138, | |
| "step": 61200 | |
| }, | |
| { | |
| "epoch": 1.097592, | |
| "grad_norm": 2.908416986465454, | |
| "learning_rate": 5.118955823293173e-06, | |
| "loss": 0.1315, | |
| "step": 61300 | |
| }, | |
| { | |
| "epoch": 1.098392, | |
| "grad_norm": 3.4737966060638428, | |
| "learning_rate": 5.110923694779117e-06, | |
| "loss": 0.1333, | |
| "step": 61400 | |
| }, | |
| { | |
| "epoch": 1.099192, | |
| "grad_norm": 4.395986080169678, | |
| "learning_rate": 5.10289156626506e-06, | |
| "loss": 0.1168, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.099992, | |
| "grad_norm": 5.61540412902832, | |
| "learning_rate": 5.094859437751004e-06, | |
| "loss": 0.1006, | |
| "step": 61600 | |
| }, | |
| { | |
| "epoch": 1.100792, | |
| "grad_norm": 7.517305850982666, | |
| "learning_rate": 5.0869076305220885e-06, | |
| "loss": 0.1156, | |
| "step": 61700 | |
| }, | |
| { | |
| "epoch": 1.101592, | |
| "grad_norm": 3.4721179008483887, | |
| "learning_rate": 5.0788755020080325e-06, | |
| "loss": 0.1029, | |
| "step": 61800 | |
| }, | |
| { | |
| "epoch": 1.102392, | |
| "grad_norm": 8.264998435974121, | |
| "learning_rate": 5.070843373493976e-06, | |
| "loss": 0.1173, | |
| "step": 61900 | |
| }, | |
| { | |
| "epoch": 1.103192, | |
| "grad_norm": 5.457461833953857, | |
| "learning_rate": 5.06281124497992e-06, | |
| "loss": 0.1189, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.103992, | |
| "grad_norm": 5.589449882507324, | |
| "learning_rate": 5.054779116465864e-06, | |
| "loss": 0.1162, | |
| "step": 62100 | |
| }, | |
| { | |
| "epoch": 1.104792, | |
| "grad_norm": 6.30127477645874, | |
| "learning_rate": 5.046746987951807e-06, | |
| "loss": 0.1196, | |
| "step": 62200 | |
| }, | |
| { | |
| "epoch": 1.105592, | |
| "grad_norm": 3.9898123741149902, | |
| "learning_rate": 5.038714859437751e-06, | |
| "loss": 0.1033, | |
| "step": 62300 | |
| }, | |
| { | |
| "epoch": 1.106392, | |
| "grad_norm": 4.696240425109863, | |
| "learning_rate": 5.030682730923695e-06, | |
| "loss": 0.1028, | |
| "step": 62400 | |
| }, | |
| { | |
| "epoch": 1.107192, | |
| "grad_norm": 5.9711384773254395, | |
| "learning_rate": 5.022650602409639e-06, | |
| "loss": 0.1151, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.107992, | |
| "grad_norm": 4.918034076690674, | |
| "learning_rate": 5.014618473895583e-06, | |
| "loss": 0.1182, | |
| "step": 62600 | |
| }, | |
| { | |
| "epoch": 1.108792, | |
| "grad_norm": 2.1706135272979736, | |
| "learning_rate": 5.006586345381527e-06, | |
| "loss": 0.1219, | |
| "step": 62700 | |
| }, | |
| { | |
| "epoch": 1.109592, | |
| "grad_norm": 3.3897037506103516, | |
| "learning_rate": 4.998554216867471e-06, | |
| "loss": 0.109, | |
| "step": 62800 | |
| }, | |
| { | |
| "epoch": 1.110392, | |
| "grad_norm": 5.085102558135986, | |
| "learning_rate": 4.990522088353414e-06, | |
| "loss": 0.1182, | |
| "step": 62900 | |
| }, | |
| { | |
| "epoch": 1.111192, | |
| "grad_norm": 6.255206108093262, | |
| "learning_rate": 4.982489959839358e-06, | |
| "loss": 0.1228, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.111992, | |
| "grad_norm": 5.1899309158325195, | |
| "learning_rate": 4.974457831325302e-06, | |
| "loss": 0.1158, | |
| "step": 63100 | |
| }, | |
| { | |
| "epoch": 1.112792, | |
| "grad_norm": 7.743951320648193, | |
| "learning_rate": 4.966425702811246e-06, | |
| "loss": 0.1247, | |
| "step": 63200 | |
| }, | |
| { | |
| "epoch": 1.113592, | |
| "grad_norm": 4.700534343719482, | |
| "learning_rate": 4.958393574297189e-06, | |
| "loss": 0.1152, | |
| "step": 63300 | |
| }, | |
| { | |
| "epoch": 1.114392, | |
| "grad_norm": 5.209968090057373, | |
| "learning_rate": 4.950361445783133e-06, | |
| "loss": 0.1098, | |
| "step": 63400 | |
| }, | |
| { | |
| "epoch": 1.115192, | |
| "grad_norm": 6.524351119995117, | |
| "learning_rate": 4.9423293172690766e-06, | |
| "loss": 0.1073, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.115992, | |
| "grad_norm": 8.537145614624023, | |
| "learning_rate": 4.9342971887550205e-06, | |
| "loss": 0.1137, | |
| "step": 63600 | |
| }, | |
| { | |
| "epoch": 1.116792, | |
| "grad_norm": 8.046852111816406, | |
| "learning_rate": 4.926265060240964e-06, | |
| "loss": 0.1059, | |
| "step": 63700 | |
| }, | |
| { | |
| "epoch": 1.117592, | |
| "grad_norm": 6.976133346557617, | |
| "learning_rate": 4.9182329317269075e-06, | |
| "loss": 0.1238, | |
| "step": 63800 | |
| }, | |
| { | |
| "epoch": 1.118392, | |
| "grad_norm": 3.892697334289551, | |
| "learning_rate": 4.910200803212851e-06, | |
| "loss": 0.1257, | |
| "step": 63900 | |
| }, | |
| { | |
| "epoch": 1.119192, | |
| "grad_norm": 4.94338321685791, | |
| "learning_rate": 4.902168674698795e-06, | |
| "loss": 0.1004, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.119192, | |
| "eval_test1_cer": 0.046203973411009035, | |
| "eval_test1_cer_norm": 0.031037049487358033, | |
| "eval_test1_loss": 0.20126041769981384, | |
| "eval_test1_runtime": 1185.1944, | |
| "eval_test1_samples_per_second": 2.109, | |
| "eval_test1_steps_per_second": 0.527, | |
| "eval_test1_wer": 0.1458934662818158, | |
| "eval_test1_wer_norm": 0.08451445103597417, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.119192, | |
| "eval_test2_cer": 0.09468212192481427, | |
| "eval_test2_cer_norm": 0.07436764022311744, | |
| "eval_test2_loss": 0.36592334508895874, | |
| "eval_test2_runtime": 1203.0799, | |
| "eval_test2_samples_per_second": 2.078, | |
| "eval_test2_steps_per_second": 0.519, | |
| "eval_test2_wer": 0.21420805676356144, | |
| "eval_test2_wer_norm": 0.14769080907632362, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.119992, | |
| "grad_norm": 7.434136867523193, | |
| "learning_rate": 4.894136546184739e-06, | |
| "loss": 0.1235, | |
| "step": 64100 | |
| }, | |
| { | |
| "epoch": 1.120792, | |
| "grad_norm": 4.429540157318115, | |
| "learning_rate": 4.886104417670683e-06, | |
| "loss": 0.1095, | |
| "step": 64200 | |
| }, | |
| { | |
| "epoch": 1.121592, | |
| "grad_norm": 3.018216133117676, | |
| "learning_rate": 4.878072289156627e-06, | |
| "loss": 0.1165, | |
| "step": 64300 | |
| }, | |
| { | |
| "epoch": 1.122392, | |
| "grad_norm": 3.923384189605713, | |
| "learning_rate": 4.870040160642571e-06, | |
| "loss": 0.1174, | |
| "step": 64400 | |
| }, | |
| { | |
| "epoch": 1.123192, | |
| "grad_norm": 9.313278198242188, | |
| "learning_rate": 4.862008032128515e-06, | |
| "loss": 0.1145, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.123992, | |
| "grad_norm": 4.272242069244385, | |
| "learning_rate": 4.853975903614459e-06, | |
| "loss": 0.1187, | |
| "step": 64600 | |
| }, | |
| { | |
| "epoch": 1.124792, | |
| "grad_norm": 5.922658920288086, | |
| "learning_rate": 4.845943775100402e-06, | |
| "loss": 0.1142, | |
| "step": 64700 | |
| }, | |
| { | |
| "epoch": 1.125592, | |
| "grad_norm": 6.251376628875732, | |
| "learning_rate": 4.837911646586346e-06, | |
| "loss": 0.115, | |
| "step": 64800 | |
| }, | |
| { | |
| "epoch": 1.126392, | |
| "grad_norm": 7.813810348510742, | |
| "learning_rate": 4.82987951807229e-06, | |
| "loss": 0.119, | |
| "step": 64900 | |
| }, | |
| { | |
| "epoch": 1.127192, | |
| "grad_norm": 2.6300604343414307, | |
| "learning_rate": 4.821847389558234e-06, | |
| "loss": 0.1097, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.1279919999999999, | |
| "grad_norm": 3.9435408115386963, | |
| "learning_rate": 4.813815261044177e-06, | |
| "loss": 0.1368, | |
| "step": 65100 | |
| }, | |
| { | |
| "epoch": 1.128792, | |
| "grad_norm": 3.7023379802703857, | |
| "learning_rate": 4.805783132530121e-06, | |
| "loss": 0.1216, | |
| "step": 65200 | |
| }, | |
| { | |
| "epoch": 1.129592, | |
| "grad_norm": 3.5150551795959473, | |
| "learning_rate": 4.797751004016065e-06, | |
| "loss": 0.1188, | |
| "step": 65300 | |
| }, | |
| { | |
| "epoch": 1.130392, | |
| "grad_norm": 8.510544776916504, | |
| "learning_rate": 4.7897188755020085e-06, | |
| "loss": 0.1071, | |
| "step": 65400 | |
| }, | |
| { | |
| "epoch": 1.131192, | |
| "grad_norm": 4.728481292724609, | |
| "learning_rate": 4.7816867469879524e-06, | |
| "loss": 0.1149, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.1319919999999999, | |
| "grad_norm": 2.2337489128112793, | |
| "learning_rate": 4.7736546184738955e-06, | |
| "loss": 0.1155, | |
| "step": 65600 | |
| }, | |
| { | |
| "epoch": 1.132792, | |
| "grad_norm": 8.161867141723633, | |
| "learning_rate": 4.765622489959839e-06, | |
| "loss": 0.115, | |
| "step": 65700 | |
| }, | |
| { | |
| "epoch": 1.133592, | |
| "grad_norm": 5.036279678344727, | |
| "learning_rate": 4.757590361445783e-06, | |
| "loss": 0.1316, | |
| "step": 65800 | |
| }, | |
| { | |
| "epoch": 1.134392, | |
| "grad_norm": 9.957592964172363, | |
| "learning_rate": 4.749558232931727e-06, | |
| "loss": 0.1127, | |
| "step": 65900 | |
| }, | |
| { | |
| "epoch": 1.135192, | |
| "grad_norm": 4.2074809074401855, | |
| "learning_rate": 4.741526104417671e-06, | |
| "loss": 0.1092, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.135992, | |
| "grad_norm": 3.1782147884368896, | |
| "learning_rate": 4.733493975903615e-06, | |
| "loss": 0.111, | |
| "step": 66100 | |
| }, | |
| { | |
| "epoch": 1.136792, | |
| "grad_norm": 3.371274709701538, | |
| "learning_rate": 4.725461847389559e-06, | |
| "loss": 0.1286, | |
| "step": 66200 | |
| }, | |
| { | |
| "epoch": 1.137592, | |
| "grad_norm": 3.8433542251586914, | |
| "learning_rate": 4.717429718875502e-06, | |
| "loss": 0.109, | |
| "step": 66300 | |
| }, | |
| { | |
| "epoch": 1.138392, | |
| "grad_norm": 5.162513256072998, | |
| "learning_rate": 4.709477911646586e-06, | |
| "loss": 0.1066, | |
| "step": 66400 | |
| }, | |
| { | |
| "epoch": 1.139192, | |
| "grad_norm": 3.1563637256622314, | |
| "learning_rate": 4.70144578313253e-06, | |
| "loss": 0.1252, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.139992, | |
| "grad_norm": 10.928020477294922, | |
| "learning_rate": 4.693413654618474e-06, | |
| "loss": 0.1255, | |
| "step": 66600 | |
| }, | |
| { | |
| "epoch": 1.140792, | |
| "grad_norm": 12.533303260803223, | |
| "learning_rate": 4.685381526104418e-06, | |
| "loss": 0.1141, | |
| "step": 66700 | |
| }, | |
| { | |
| "epoch": 1.141592, | |
| "grad_norm": 4.71325159072876, | |
| "learning_rate": 4.677349397590361e-06, | |
| "loss": 0.1074, | |
| "step": 66800 | |
| }, | |
| { | |
| "epoch": 1.142392, | |
| "grad_norm": 4.4170427322387695, | |
| "learning_rate": 4.669317269076305e-06, | |
| "loss": 0.1081, | |
| "step": 66900 | |
| }, | |
| { | |
| "epoch": 1.143192, | |
| "grad_norm": 7.2962965965271, | |
| "learning_rate": 4.661285140562249e-06, | |
| "loss": 0.1115, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.143992, | |
| "grad_norm": 6.532419204711914, | |
| "learning_rate": 4.653253012048193e-06, | |
| "loss": 0.1194, | |
| "step": 67100 | |
| }, | |
| { | |
| "epoch": 1.144792, | |
| "grad_norm": 9.143524169921875, | |
| "learning_rate": 4.645220883534137e-06, | |
| "loss": 0.1148, | |
| "step": 67200 | |
| }, | |
| { | |
| "epoch": 1.145592, | |
| "grad_norm": 2.3634395599365234, | |
| "learning_rate": 4.637188755020081e-06, | |
| "loss": 0.1322, | |
| "step": 67300 | |
| }, | |
| { | |
| "epoch": 1.146392, | |
| "grad_norm": 8.534736633300781, | |
| "learning_rate": 4.629156626506025e-06, | |
| "loss": 0.1033, | |
| "step": 67400 | |
| }, | |
| { | |
| "epoch": 1.147192, | |
| "grad_norm": 7.3824944496154785, | |
| "learning_rate": 4.621124497991969e-06, | |
| "loss": 0.1109, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.147992, | |
| "grad_norm": 3.094473123550415, | |
| "learning_rate": 4.613092369477913e-06, | |
| "loss": 0.1201, | |
| "step": 67600 | |
| }, | |
| { | |
| "epoch": 1.148792, | |
| "grad_norm": 4.593748569488525, | |
| "learning_rate": 4.605060240963856e-06, | |
| "loss": 0.1096, | |
| "step": 67700 | |
| }, | |
| { | |
| "epoch": 1.149592, | |
| "grad_norm": 4.944604396820068, | |
| "learning_rate": 4.5970281124498e-06, | |
| "loss": 0.1128, | |
| "step": 67800 | |
| }, | |
| { | |
| "epoch": 1.150392, | |
| "grad_norm": 6.725574493408203, | |
| "learning_rate": 4.5889959839357435e-06, | |
| "loss": 0.1129, | |
| "step": 67900 | |
| }, | |
| { | |
| "epoch": 1.151192, | |
| "grad_norm": 8.640876770019531, | |
| "learning_rate": 4.5809638554216874e-06, | |
| "loss": 0.0995, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.151192, | |
| "eval_test1_cer": 0.05561935170662484, | |
| "eval_test1_cer_norm": 0.03838931975892621, | |
| "eval_test1_loss": 0.19823457300662994, | |
| "eval_test1_runtime": 1196.2654, | |
| "eval_test1_samples_per_second": 2.09, | |
| "eval_test1_steps_per_second": 0.522, | |
| "eval_test1_wer": 0.15913000379019795, | |
| "eval_test1_wer_norm": 0.09649609865863994, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.151192, | |
| "eval_test2_cer": 0.09447679844700788, | |
| "eval_test2_cer_norm": 0.07438216609854353, | |
| "eval_test2_loss": 0.36025092005729675, | |
| "eval_test2_runtime": 1407.094, | |
| "eval_test2_samples_per_second": 1.777, | |
| "eval_test2_steps_per_second": 0.444, | |
| "eval_test2_wer": 0.21621080338750287, | |
| "eval_test2_wer_norm": 0.14837840935136373, | |
| "step": 68000 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 125000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 4000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.7760453484544e+20, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |