YAML Metadata
Error:
"model-index[0].name" is not allowed to be empty
This model is a fine-tuned version of DrishtiSharma/wav2vec2-large-xls-r-300m-hi-d3 on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - UR dataset. It achieves the following results on the evaluation set:
- Loss: 1.5443
- Wer: 0.7030
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.000388
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 750
- num_epochs: 100.0
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
10.7052 | 1.96 | 100 | 3.4683 | 1.0 |
3.2395 | 3.92 | 200 | 3.1489 | 1.0 |
2.9951 | 5.88 | 300 | 2.9823 | 1.0007 |
2.3574 | 7.84 | 400 | 1.2614 | 0.7598 |
1.7287 | 9.8 | 500 | 1.1817 | 0.7421 |
1.6144 | 11.76 | 600 | 1.1315 | 0.7321 |
1.5598 | 13.73 | 700 | 1.2322 | 0.7550 |
1.5418 | 15.69 | 800 | 1.2721 | 0.7819 |
1.4578 | 17.65 | 900 | 1.1710 | 0.7531 |
1.4311 | 19.61 | 1000 | 1.2042 | 0.7491 |
1.3483 | 21.57 | 1100 | 1.1702 | 0.7465 |
1.3078 | 23.53 | 1200 | 1.1963 | 0.7421 |
1.2576 | 25.49 | 1300 | 1.1501 | 0.7280 |
1.2173 | 27.45 | 1400 | 1.2526 | 0.7299 |
1.2217 | 29.41 | 1500 | 1.2479 | 0.7310 |
1.1536 | 31.37 | 1600 | 1.2567 | 0.7432 |
1.0939 | 33.33 | 1700 | 1.2801 | 0.7247 |
1.0745 | 35.29 | 1800 | 1.2340 | 0.7151 |
1.0454 | 37.25 | 1900 | 1.2372 | 0.7151 |
1.0101 | 39.22 | 2000 | 1.2461 | 0.7376 |
0.9833 | 41.18 | 2100 | 1.2553 | 0.7269 |
0.9314 | 43.14 | 2200 | 1.2372 | 0.7015 |
0.9147 | 45.1 | 2300 | 1.3035 | 0.7358 |
0.8758 | 47.06 | 2400 | 1.2598 | 0.7092 |
0.8356 | 49.02 | 2500 | 1.2557 | 0.7144 |
0.8105 | 50.98 | 2600 | 1.2619 | 0.7236 |
0.7947 | 52.94 | 2700 | 1.3994 | 0.7491 |
0.7623 | 54.9 | 2800 | 1.2932 | 0.7133 |
0.7282 | 56.86 | 2900 | 1.2799 | 0.7089 |
0.7108 | 58.82 | 3000 | 1.3615 | 0.7148 |
0.6896 | 60.78 | 3100 | 1.3129 | 0.7041 |
0.6496 | 62.75 | 3200 | 1.4050 | 0.6934 |
0.6075 | 64.71 | 3300 | 1.3571 | 0.7026 |
0.6242 | 66.67 | 3400 | 1.3369 | 0.7063 |
0.5865 | 68.63 | 3500 | 1.4368 | 0.7140 |
0.5721 | 70.59 | 3600 | 1.4224 | 0.7066 |
0.5475 | 72.55 | 3700 | 1.4798 | 0.7118 |
0.5086 | 74.51 | 3800 | 1.5107 | 0.7232 |
0.4958 | 76.47 | 3900 | 1.4849 | 0.7089 |
0.5046 | 78.43 | 4000 | 1.4451 | 0.7114 |
0.4694 | 80.39 | 4100 | 1.4674 | 0.7089 |
0.4386 | 82.35 | 4200 | 1.5245 | 0.7103 |
0.4516 | 84.31 | 4300 | 1.5032 | 0.7103 |
0.4113 | 86.27 | 4400 | 1.5246 | 0.7196 |
0.3972 | 88.24 | 4500 | 1.5318 | 0.7114 |
0.4006 | 90.2 | 4600 | 1.5543 | 0.6982 |
0.4014 | 92.16 | 4700 | 1.5442 | 0.7048 |
0.3672 | 94.12 | 4800 | 1.5542 | 0.7137 |
0.3666 | 96.08 | 4900 | 1.5414 | 0.7018 |
0.3574 | 98.04 | 5000 | 1.5465 | 0.7059 |
0.3428 | 100.0 | 5100 | 1.5443 | 0.7030 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.3
- Tokenizers 0.11.0
- Downloads last month
- 12
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.