<!-- 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. -->
combined-MTL9
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: 1.3413
- Wer: 0.8603
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 |
|---|---|---|---|---|
| 76.4918 | 0.35 | 500 | 3.4171 | 1.0 |
| 2.9927 | 0.69 | 1000 | 2.4743 | 1.0667 |
| 2.2033 | 1.04 | 1500 | 1.6693 | 1.25 |
| 1.6165 | 1.39 | 2000 | 1.5341 | 1.1808 |
| 1.4208 | 1.74 | 2500 | 1.3148 | 1.1179 |
| 1.2858 | 2.08 | 3000 | 1.2272 | 1.0872 |
| 1.1317 | 2.43 | 3500 | 1.0865 | 1.0731 |
| 1.0668 | 2.78 | 4000 | 1.0798 | 1.0474 |
| 1.0429 | 3.12 | 4500 | 1.4627 | 1.0936 |
| 0.9615 | 3.47 | 5000 | 1.2540 | 1.0090 |
| 0.975 | 3.82 | 5500 | 0.9936 | 0.9679 |
| 0.8517 | 4.17 | 6000 | 1.1039 | 1.0282 |
| 0.8281 | 4.51 | 6500 | 1.0609 | 0.9897 |
| 0.8413 | 4.86 | 7000 | 0.9513 | 0.9397 |
| 0.7618 | 5.21 | 7500 | 1.1656 | 0.9718 |
| 0.7173 | 5.56 | 8000 | 1.1974 | 0.9603 |
| 0.7449 | 5.9 | 8500 | 1.0144 | 0.9731 |
| 0.6762 | 6.25 | 9000 | 1.1774 | 0.9231 |
| 0.6749 | 6.6 | 9500 | 1.1823 | 0.9205 |
| 0.6776 | 6.94 | 10000 | 0.9167 | 0.9244 |
| 0.5937 | 7.29 | 10500 | 1.3344 | 0.9769 |
| 0.6488 | 7.64 | 11000 | 1.0245 | 0.9692 |
| 0.6116 | 7.99 | 11500 | 0.9444 | 0.9141 |
| 0.5497 | 8.33 | 12000 | 0.9499 | 0.9692 |
| 0.5937 | 8.68 | 12500 | 1.1087 | 0.9231 |
| 0.5268 | 9.03 | 13000 | 1.3408 | 0.9269 |
| 0.5078 | 9.38 | 13500 | 1.1737 | 0.9038 |
| 0.497 | 9.72 | 14000 | 0.9963 | 0.8987 |
| 0.5231 | 10.07 | 14500 | 1.3247 | 0.9590 |
| 0.4651 | 10.42 | 15000 | 1.1988 | 0.9308 |
| 0.481 | 10.76 | 15500 | 1.0034 | 0.9308 |
| 0.481 | 11.11 | 16000 | 1.0040 | 0.8782 |
| 0.4751 | 11.46 | 16500 | 0.8824 | 0.8538 |
| 0.4554 | 11.81 | 17000 | 0.9741 | 0.8821 |
| 0.426 | 12.15 | 17500 | 0.8552 | 0.8615 |
| 0.4186 | 12.5 | 18000 | 1.0646 | 0.8833 |
| 0.4154 | 12.85 | 18500 | 0.9618 | 0.8936 |
| 0.5115 | 13.19 | 19000 | 1.0312 | 0.8910 |
| 0.3564 | 13.54 | 19500 | 1.0686 | 0.8769 |
| 0.3927 | 13.89 | 20000 | 1.2533 | 0.9103 |
| 0.3628 | 14.24 | 20500 | 1.2945 | 0.8872 |
| 0.3808 | 14.58 | 21000 | 1.0195 | 0.8538 |
| 0.3981 | 14.93 | 21500 | 1.0388 | 0.8808 |
| 0.3337 | 15.28 | 22000 | 1.0464 | 0.8923 |
| 0.3092 | 15.62 | 22500 | 1.0843 | 0.8705 |
| 0.378 | 15.97 | 23000 | 1.0880 | 0.8859 |
| 0.3231 | 16.32 | 23500 | 0.9205 | 0.8782 |
| 0.3588 | 16.67 | 24000 | 1.0064 | 0.8962 |
| 0.3048 | 17.01 | 24500 | 0.9130 | 0.8705 |
| 0.3 | 17.36 | 25000 | 1.0100 | 0.9077 |
| 0.3045 | 17.71 | 25500 | 1.0559 | 0.9077 |
| 0.3024 | 18.06 | 26000 | 1.1225 | 0.9026 |
| 0.2614 | 18.4 | 26500 | 1.0911 | 0.8897 |
| 0.2755 | 18.75 | 27000 | 1.0872 | 0.8808 |
| 0.2798 | 19.1 | 27500 | 1.2911 | 0.9154 |
| 0.2455 | 19.44 | 28000 | 1.0646 | 0.8821 |
| 0.2524 | 19.79 | 28500 | 1.3356 | 0.9154 |
| 0.2435 | 20.14 | 29000 | 1.1257 | 0.8641 |
| 0.2458 | 20.49 | 29500 | 1.2221 | 0.8667 |
| 0.2216 | 20.83 | 30000 | 1.1364 | 0.8769 |
| 0.234 | 21.18 | 30500 | 1.2094 | 0.8808 |
| 0.233 | 21.53 | 31000 | 1.1604 | 0.8910 |
| 0.2536 | 21.88 | 31500 | 1.0934 | 0.8808 |
| 0.1885 | 22.22 | 32000 | 1.2177 | 0.8718 |
| 0.2186 | 22.57 | 32500 | 1.0539 | 0.8667 |
| 0.1991 | 22.92 | 33000 | 1.2222 | 0.8641 |
| 0.2027 | 23.26 | 33500 | 1.3863 | 0.8577 |
| 0.193 | 23.61 | 34000 | 1.2293 | 0.8705 |
| 0.2054 | 23.96 | 34500 | 1.3398 | 0.8769 |
| 0.2197 | 24.31 | 35000 | 1.3138 | 0.8705 |
| 0.1898 | 24.65 | 35500 | 1.2897 | 0.8679 |
| 0.1933 | 25.0 | 36000 | 1.2666 | 0.8769 |
| 0.1632 | 25.35 | 36500 | 1.2758 | 0.8756 |
| 0.1869 | 25.69 | 37000 | 1.1811 | 0.8603 |
| 0.1731 | 26.04 | 37500 | 1.2511 | 0.8679 |
| 0.1821 | 26.39 | 38000 | 1.3391 | 0.8718 |
| 0.1648 | 26.74 | 38500 | 1.2505 | 0.8628 |
| 0.1909 | 27.08 | 39000 | 1.2984 | 0.85 |
| 0.1902 | 27.43 | 39500 | 1.2261 | 0.8487 |
| 0.1449 | 27.78 | 40000 | 1.2853 | 0.8487 |
| 0.1583 | 28.12 | 40500 | 1.3361 | 0.8628 |
| 0.148 | 28.47 | 41000 | 1.3638 | 0.8654 |
| 0.1648 | 28.82 | 41500 | 1.3380 | 0.8603 |
| 0.1461 | 29.17 | 42000 | 1.3561 | 0.8603 |
| 0.1565 | 29.51 | 42500 | 1.3489 | 0.8615 |
| 0.16 | 29.86 | 43000 | 1.3413 | 0.8603 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1+cu113
- Datasets 1.18.3
- Tokenizers 0.13.2
