base_sami_22kft_ft_pseudo_widv_ep60_tde
This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.0285
- Wer: 0.0215
- Cer: 0.0044
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: 0.0005
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.25
- num_epochs: 60.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.1722 | 1.0 | 972 | 0.0286 | 0.0213 | 0.0043 |
| 0.1455 | 2.0 | 1944 | 0.0484 | 0.0384 | 0.0077 |
| 0.1681 | 3.0 | 2916 | 0.0656 | 0.0715 | 0.0145 |
| 0.192 | 4.0 | 3888 | 0.1178 | 0.1105 | 0.0234 |
| 0.2199 | 5.0 | 4860 | 0.1491 | 0.1904 | 0.0426 |
| 0.2408 | 6.0 | 5832 | 0.1710 | 0.2054 | 0.0504 |
| 0.2761 | 7.0 | 6804 | 0.2405 | 0.2398 | 0.0607 |
| 0.3111 | 8.0 | 7776 | 0.2560 | 0.3894 | 0.0969 |
| 0.3474 | 9.0 | 8748 | 0.3227 | 0.4196 | 0.1234 |
| 0.3899 | 10.0 | 9720 | 0.3253 | 0.4059 | 0.1138 |
| 0.4153 | 11.0 | 10692 | 0.3555 | 0.4881 | 0.1411 |
| 0.4633 | 12.0 | 11664 | 0.4051 | 0.5148 | 0.1554 |
| 0.5276 | 13.0 | 12636 | 0.4871 | 0.5114 | 0.1601 |
| 0.5434 | 14.0 | 13608 | 0.5272 | 0.6465 | 0.2077 |
| 0.555 | 15.0 | 14580 | 0.5462 | 0.6470 | 0.2203 |
| 0.5475 | 16.0 | 15552 | 0.4817 | 0.5892 | 0.1855 |
| 0.5442 | 17.0 | 16524 | 0.4538 | 0.5851 | 0.1835 |
| 0.8933 | 18.0 | 17496 | 0.4262 | 0.5167 | 0.1676 |
| 0.5099 | 19.0 | 18468 | 0.3809 | 0.4778 | 0.1485 |
| 0.5529 | 20.0 | 19440 | 0.3897 | 0.4954 | 0.1511 |
| 0.4446 | 21.0 | 20412 | 0.3875 | 0.4507 | 0.1365 |
| 0.4376 | 22.0 | 21384 | 0.3487 | 0.4592 | 0.1403 |
| 0.4044 | 23.0 | 22356 | 0.3340 | 0.4146 | 0.1234 |
| 0.4067 | 24.0 | 23328 | 0.3254 | 0.4069 | 0.1223 |
| 0.3696 | 25.0 | 24300 | 0.3158 | 0.3938 | 0.1161 |
| 0.3417 | 26.0 | 25272 | 0.3046 | 0.3981 | 0.1148 |
| 0.3381 | 27.0 | 26244 | 0.2911 | 0.3753 | 0.1105 |
| 0.3187 | 28.0 | 27216 | 0.3069 | 0.3865 | 0.1190 |
| 0.3016 | 29.0 | 28188 | 0.2523 | 0.3525 | 0.1006 |
| 0.2837 | 30.0 | 29160 | 0.2433 | 0.3294 | 0.0959 |
| 0.2622 | 31.0 | 30132 | 0.2359 | 0.3172 | 0.0908 |
| 0.2584 | 32.0 | 31104 | 0.2495 | 0.3269 | 0.0963 |
| 0.239 | 33.0 | 32076 | 0.2421 | 0.3188 | 0.0922 |
| 0.2275 | 34.0 | 33048 | 0.2253 | 0.3124 | 0.0917 |
| 0.2216 | 35.0 | 34020 | 0.2188 | 0.2876 | 0.0828 |
| 0.2078 | 36.0 | 34992 | 0.2263 | 0.2967 | 0.0828 |
| 0.1993 | 37.0 | 35964 | 0.2169 | 0.2873 | 0.0824 |
| 0.1846 | 38.0 | 36936 | 0.2136 | 0.2735 | 0.0778 |
| 0.1797 | 39.0 | 37908 | 0.2095 | 0.2625 | 0.0762 |
| 0.182 | 40.0 | 38880 | 0.2157 | 0.2581 | 0.0742 |
| 0.1633 | 41.0 | 39852 | 0.1872 | 0.2526 | 0.0726 |
| 0.1674 | 42.0 | 40824 | 0.1886 | 0.2471 | 0.0717 |
| 0.1598 | 43.0 | 41796 | 0.2110 | 0.2573 | 0.0729 |
| 0.1485 | 44.0 | 42768 | 0.2113 | 0.2413 | 0.0698 |
| 0.143 | 45.0 | 43740 | 0.1945 | 0.2335 | 0.0689 |
| 0.1297 | 46.0 | 44712 | 0.1945 | 0.2222 | 0.0640 |
| 0.1243 | 47.0 | 45684 | 0.1878 | 0.2217 | 0.0628 |
| 0.1153 | 48.0 | 46656 | 0.1935 | 0.2165 | 0.0614 |
| 0.1115 | 49.0 | 47628 | 0.1976 | 0.2115 | 0.0604 |
| 0.1012 | 50.0 | 48600 | 0.1816 | 0.2102 | 0.0601 |
| 0.1002 | 51.0 | 49572 | 0.1782 | 0.2041 | 0.0582 |
| 0.0917 | 52.0 | 50544 | 0.2061 | 0.2015 | 0.0588 |
| 0.086 | 53.0 | 51516 | 0.1777 | 0.1966 | 0.0579 |
| 0.0769 | 54.0 | 52488 | 0.1872 | 0.1889 | 0.0550 |
| 0.0765 | 55.0 | 53460 | 0.1923 | 0.1880 | 0.0539 |
| 0.0673 | 56.0 | 54432 | 0.1879 | 0.1820 | 0.0535 |
| 0.0646 | 57.0 | 55404 | 0.2011 | 0.1820 | 0.0534 |
| 0.0636 | 58.0 | 56376 | 0.1952 | 0.1800 | 0.0528 |
| 0.0618 | 59.0 | 57348 | 0.2021 | 0.1775 | 0.0520 |
| 0.0557 | 60.0 | 58320 | 0.1996 | 0.1745 | 0.0512 |
Framework versions
- Transformers 4.48.3
- Pytorch 2.5.1
- Datasets 3.2.0
- Tokenizers 0.21.0
- Downloads last month
- 5