|
--- |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- wer |
|
model-index: |
|
- name: Model_ALL_Wav2Vec2 |
|
results: [] |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# Model_ALL_Wav2Vec2 |
|
|
|
This model was trained from scratch on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.7779 |
|
- Wer: 0.1975 |
|
- Cer: 0.0813 |
|
|
|
## 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: 0.0003 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- total_train_batch_size: 32 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 500 |
|
- num_epochs: 30 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer | |
|
|:-------------:|:-----:|:-----:|:---------------:|:------:|:------:| |
|
| 0.8385 | 0.67 | 400 | 0.5656 | 0.3049 | 0.1100 | |
|
| 0.3291 | 1.34 | 800 | 0.5395 | 0.3184 | 0.1128 | |
|
| 0.258 | 2.01 | 1200 | 0.4904 | 0.2770 | 0.1030 | |
|
| 0.217 | 2.68 | 1600 | 0.4673 | 0.2814 | 0.1073 | |
|
| 0.1956 | 3.35 | 2000 | 0.5108 | 0.2697 | 0.1021 | |
|
| 0.1872 | 4.02 | 2400 | 0.5531 | 0.2735 | 0.1050 | |
|
| 0.168 | 4.69 | 2800 | 0.5113 | 0.2536 | 0.0967 | |
|
| 0.1476 | 5.36 | 3200 | 0.6744 | 0.2420 | 0.0941 | |
|
| 0.1531 | 6.04 | 3600 | 0.6433 | 0.2492 | 0.0962 | |
|
| 0.1271 | 6.71 | 4000 | 0.5360 | 0.2392 | 0.0928 | |
|
| 0.1362 | 7.38 | 4400 | 0.5451 | 0.2458 | 0.0958 | |
|
| 0.1169 | 8.05 | 4800 | 0.6710 | 0.2470 | 0.0965 | |
|
| 0.117 | 8.72 | 5200 | 0.5291 | 0.2480 | 0.0990 | |
|
| 0.1146 | 9.39 | 5600 | 0.6168 | 0.2372 | 0.0927 | |
|
| 0.1028 | 10.06 | 6000 | 0.5437 | 0.2294 | 0.0914 | |
|
| 0.0918 | 10.73 | 6400 | 0.6350 | 0.2392 | 0.0947 | |
|
| 0.1037 | 11.4 | 6800 | 0.6351 | 0.2346 | 0.0920 | |
|
| 0.0926 | 12.07 | 7200 | 0.6677 | 0.2316 | 0.0924 | |
|
| 0.0861 | 12.74 | 7600 | 0.5842 | 0.2301 | 0.0934 | |
|
| 0.0791 | 13.41 | 8000 | 0.5862 | 0.2286 | 0.0916 | |
|
| 0.08 | 14.08 | 8400 | 0.6183 | 0.2227 | 0.0900 | |
|
| 0.0707 | 14.75 | 8800 | 0.5985 | 0.2351 | 0.0955 | |
|
| 0.0719 | 15.42 | 9200 | 0.6327 | 0.2200 | 0.0897 | |
|
| 0.0674 | 16.09 | 9600 | 0.6184 | 0.2193 | 0.0889 | |
|
| 0.0612 | 16.76 | 10000 | 0.5501 | 0.2224 | 0.0912 | |
|
| 0.0607 | 17.44 | 10400 | 0.5404 | 0.2233 | 0.0916 | |
|
| 0.0612 | 18.11 | 10800 | 0.6111 | 0.2193 | 0.0889 | |
|
| 0.0542 | 18.78 | 11200 | 0.6610 | 0.2196 | 0.0893 | |
|
| 0.0517 | 19.45 | 11600 | 0.6083 | 0.2199 | 0.0905 | |
|
| 0.0478 | 20.12 | 12000 | 0.6500 | 0.2130 | 0.0874 | |
|
| 0.0464 | 20.79 | 12400 | 0.6671 | 0.2144 | 0.0863 | |
|
| 0.0395 | 21.46 | 12800 | 0.7239 | 0.2113 | 0.0864 | |
|
| 0.0391 | 22.13 | 13200 | 0.7791 | 0.2084 | 0.0851 | |
|
| 0.0362 | 22.8 | 13600 | 0.6682 | 0.2083 | 0.0855 | |
|
| 0.0396 | 23.47 | 14000 | 0.6608 | 0.2065 | 0.0848 | |
|
| 0.0346 | 24.14 | 14400 | 0.7438 | 0.2065 | 0.0856 | |
|
| 0.0368 | 24.81 | 14800 | 0.7382 | 0.2066 | 0.0842 | |
|
| 0.0273 | 25.48 | 15200 | 0.7486 | 0.2020 | 0.0841 | |
|
| 0.0286 | 26.15 | 15600 | 0.7566 | 0.2029 | 0.0838 | |
|
| 0.0268 | 26.82 | 16000 | 0.7680 | 0.2015 | 0.0828 | |
|
| 0.0248 | 27.49 | 16400 | 0.7499 | 0.1994 | 0.0813 | |
|
| 0.0253 | 28.16 | 16800 | 0.7511 | 0.1998 | 0.0820 | |
|
| 0.0228 | 28.83 | 17200 | 0.7686 | 0.1985 | 0.0820 | |
|
| 0.0212 | 29.51 | 17600 | 0.7779 | 0.1975 | 0.0813 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.31.0 |
|
- Pytorch 2.0.1+cu117 |
|
- Datasets 1.18.3 |
|
- Tokenizers 0.13.3 |
|
|
