<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
exp15-F01-both
This model is a fine-tuned version of yongjian/wav2vec2-large-a on the None dataset. It achieves the following results on the evaluation set:
- Loss: 2.6149
- Wer: 1.0154
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: 0.0001
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 45.2628 | 0.33 | 500 | 2.9838 | 1.0 |
| 3.1304 | 0.67 | 1000 | 2.8311 | 1.0 |
| 2.9607 | 1.0 | 1500 | 2.6426 | 1.0039 |
| 2.7429 | 1.33 | 2000 | 2.5365 | 1.2046 |
| 2.5496 | 1.66 | 2500 | 2.2169 | 1.3050 |
| 2.3134 | 2.0 | 3000 | 2.0450 | 1.3127 |
| 2.1189 | 2.33 | 3500 | 1.8677 | 1.2780 |
| 2.0075 | 2.66 | 4000 | 1.7450 | 1.2703 |
| 1.9014 | 3.0 | 4500 | 1.8381 | 1.2664 |
| 1.7246 | 3.33 | 5000 | 1.7980 | 1.2510 |
| 1.6783 | 3.66 | 5500 | 1.7269 | 1.2510 |
| 1.589 | 3.99 | 6000 | 1.5640 | 1.2664 |
| 1.4085 | 4.33 | 6500 | 1.7296 | 1.2355 |
| 1.4126 | 4.66 | 7000 | 1.5208 | 1.2317 |
| 1.3506 | 4.99 | 7500 | 1.6253 | 1.2317 |
| 1.2276 | 5.33 | 8000 | 1.6222 | 1.2239 |
| 1.1842 | 5.66 | 8500 | 1.4836 | 1.1969 |
| 1.1445 | 5.99 | 9000 | 1.5313 | 1.2046 |
| 1.0254 | 6.32 | 9500 | 1.9130 | 1.2046 |
| 1.0214 | 6.66 | 10000 | 1.8944 | 1.2085 |
| 0.9677 | 6.99 | 10500 | 1.9039 | 1.1853 |
| 0.8822 | 7.32 | 11000 | 1.7036 | 1.1892 |
| 0.8824 | 7.66 | 11500 | 1.6062 | 1.1815 |
| 0.8695 | 7.99 | 12000 | 1.7019 | 1.1853 |
| 0.7536 | 8.32 | 12500 | 1.9117 | 1.1737 |
| 0.775 | 8.66 | 13000 | 1.8778 | 1.1815 |
| 0.7409 | 8.99 | 13500 | 1.7534 | 1.1776 |
| 0.7035 | 9.32 | 14000 | 1.9860 | 1.1853 |
| 0.6905 | 9.65 | 14500 | 1.9141 | 1.1892 |
| 0.6536 | 9.99 | 15000 | 1.7848 | 1.1737 |
| 0.6237 | 10.32 | 15500 | 2.0624 | 1.1544 |
| 0.5986 | 10.65 | 16000 | 1.9958 | 1.1544 |
| 0.5838 | 10.99 | 16500 | 1.8005 | 1.1622 |
| 0.5231 | 11.32 | 17000 | 1.5967 | 1.1351 |
| 0.5452 | 11.65 | 17500 | 1.8145 | 1.1274 |
| 0.5446 | 11.98 | 18000 | 2.0214 | 1.1429 |
| 0.4727 | 12.32 | 18500 | 1.8989 | 1.1313 |
| 0.4908 | 12.65 | 19000 | 1.7152 | 1.1467 |
| 0.483 | 12.98 | 19500 | 1.7354 | 1.1429 |
| 0.4455 | 13.32 | 20000 | 1.9493 | 1.1506 |
| 0.4456 | 13.65 | 20500 | 2.0869 | 1.1197 |
| 0.4306 | 13.98 | 21000 | 1.9248 | 1.1236 |
| 0.3827 | 14.31 | 21500 | 1.9245 | 1.1274 |
| 0.4059 | 14.65 | 22000 | 1.9478 | 1.1313 |
| 0.3941 | 14.98 | 22500 | 2.2373 | 1.1197 |
| 0.4094 | 15.31 | 23000 | 2.0268 | 1.1158 |
| 0.3584 | 15.65 | 23500 | 1.9292 | 1.1313 |
| 0.3615 | 15.98 | 24000 | 2.1744 | 1.0965 |
| 0.3564 | 16.31 | 24500 | 2.4167 | 1.0927 |
| 0.3202 | 16.64 | 25000 | 2.6332 | 1.1081 |
| 0.3099 | 16.98 | 25500 | 2.9448 | 1.1004 |
| 0.3126 | 17.31 | 26000 | 2.4662 | 1.0927 |
| 0.3189 | 17.64 | 26500 | 2.3619 | 1.0772 |
| 0.3929 | 17.98 | 27000 | 2.3571 | 1.0618 |
| 0.27 | 18.31 | 27500 | 2.2457 | 1.0734 |
| 0.2664 | 18.64 | 28000 | 2.5133 | 1.0772 |
| 0.2875 | 18.97 | 28500 | 2.2798 | 1.0618 |
| 0.2336 | 19.31 | 29000 | 2.3515 | 1.0347 |
| 0.2597 | 19.64 | 29500 | 2.3072 | 1.0463 |
| 0.2573 | 19.97 | 30000 | 2.1702 | 1.0425 |
| 0.2431 | 20.31 | 30500 | 2.2727 | 1.0618 |
| 0.2362 | 20.64 | 31000 | 2.3082 | 1.0772 |
| 0.2377 | 20.97 | 31500 | 2.5453 | 1.0734 |
| 0.228 | 21.3 | 32000 | 2.6838 | 1.0618 |
| 0.2082 | 21.64 | 32500 | 2.7629 | 1.0695 |
| 0.2041 | 21.97 | 33000 | 2.4433 | 1.0347 |
| 0.2208 | 22.3 | 33500 | 2.2516 | 1.0463 |
| 0.2505 | 22.64 | 34000 | 2.4056 | 1.0541 |
| 0.187 | 22.97 | 34500 | 2.6017 | 1.0347 |
| 0.1987 | 23.3 | 35000 | 2.5061 | 1.0425 |
| 0.1952 | 23.64 | 35500 | 2.4440 | 1.0463 |
| 0.1777 | 23.97 | 36000 | 2.4333 | 1.0463 |
| 0.1981 | 24.3 | 36500 | 2.4327 | 1.0309 |
| 0.1729 | 24.63 | 37000 | 2.4114 | 1.0309 |
| 0.1895 | 24.97 | 37500 | 2.3885 | 1.0347 |
| 0.1766 | 25.3 | 38000 | 2.2978 | 1.0154 |
| 0.1603 | 25.63 | 38500 | 2.3070 | 1.0039 |
| 0.1764 | 25.97 | 39000 | 2.4975 | 1.0154 |
| 0.1502 | 26.3 | 39500 | 2.3422 | 0.9923 |
| 0.1574 | 26.63 | 40000 | 2.5013 | 1.0077 |
| 0.1794 | 26.96 | 40500 | 2.4088 | 1.0039 |
| 0.1481 | 27.3 | 41000 | 2.3456 | 1.0077 |
| 0.1594 | 27.63 | 41500 | 2.4916 | 1.0154 |
| 0.1384 | 27.96 | 42000 | 2.4173 | 1.0077 |
| 0.1649 | 28.3 | 42500 | 2.5922 | 1.0116 |
| 0.145 | 28.63 | 43000 | 2.5461 | 1.0039 |
| 0.1654 | 28.96 | 43500 | 2.5312 | 1.0039 |
| 0.1389 | 29.29 | 44000 | 2.5974 | 1.0077 |
| 0.1592 | 29.63 | 44500 | 2.6050 | 1.0193 |
| 0.1055 | 29.96 | 45000 | 2.6149 | 1.0154 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1+cu113
- Datasets 1.18.3
- Tokenizers 0.13.2
