whisper-large-telephone-v1
This model is a fine-tuned version of openai/whisper-large-v3 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.4392
- Wer: 52.2874
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: 3e-05
- train_batch_size: 16
- eval_batch_size: 16
- 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_steps: 1250
- num_epochs: 12
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.0345 | 0.1416 | 500 | 1.4197 | 109.5730 |
| 0.8638 | 0.2832 | 1000 | 1.2463 | 80.6058 |
| 0.8131 | 0.4248 | 1500 | 1.1927 | 89.7440 |
| 0.7539 | 0.5664 | 2000 | 1.1132 | 86.8466 |
| 0.7429 | 0.7080 | 2500 | 1.0692 | 78.1367 |
| 0.7524 | 0.8496 | 3000 | 1.0429 | 67.9277 |
| 0.6429 | 0.9912 | 3500 | 0.9951 | 79.6665 |
| 0.5777 | 1.1328 | 4000 | 0.9834 | 70.1924 |
| 0.5681 | 1.2744 | 4500 | 0.9818 | 63.9629 |
| 0.509 | 1.4160 | 5000 | 0.9620 | 60.5350 |
| 0.5353 | 1.5576 | 5500 | 0.9432 | 65.0628 |
| 0.4851 | 1.6992 | 6000 | 0.9316 | 58.2249 |
| 0.501 | 1.8408 | 6500 | 0.9281 | 67.6438 |
| 0.5861 | 1.9824 | 7000 | 0.9029 | 60.6275 |
| 0.3368 | 2.1240 | 7500 | 0.9508 | 59.1756 |
| 0.3633 | 2.2656 | 8000 | 0.9263 | 55.8807 |
| 0.3257 | 2.4073 | 8500 | 0.9449 | 57.2272 |
| 0.361 | 2.5489 | 9000 | 0.9355 | 61.9156 |
| 0.337 | 2.6905 | 9500 | 0.9372 | 58.0384 |
| 0.3408 | 2.8321 | 10000 | 0.9153 | 56.0965 |
| 0.3362 | 2.9737 | 10500 | 0.9213 | 57.5663 |
| 0.196 | 3.1153 | 11000 | 0.9820 | 57.2597 |
| 0.1815 | 3.2569 | 11500 | 1.0006 | 55.0728 |
| 0.178 | 3.3985 | 12000 | 1.0060 | 57.1818 |
| 0.2 | 3.5401 | 12500 | 0.9999 | 54.7241 |
| 0.1896 | 3.6817 | 13000 | 0.9970 | 54.0346 |
| 0.1881 | 3.8233 | 13500 | 0.9912 | 56.3187 |
| 0.2182 | 3.9649 | 14000 | 0.9785 | 55.4233 |
| 0.0855 | 4.1065 | 14500 | 1.0754 | 54.8181 |
| 0.1033 | 4.2481 | 15000 | 1.0664 | 55.2594 |
| 0.0969 | 4.3897 | 15500 | 1.0600 | 55.6406 |
| 0.0985 | 4.5313 | 16000 | 1.0799 | 57.9232 |
| 0.094 | 4.6729 | 16500 | 1.1025 | 56.3755 |
| 0.1026 | 4.8145 | 17000 | 1.0939 | 54.8814 |
| 0.1029 | 4.9561 | 17500 | 1.0987 | 54.8765 |
| 0.0444 | 5.0977 | 18000 | 1.1591 | 54.4142 |
| 0.047 | 5.2393 | 18500 | 1.1655 | 57.5711 |
| 0.0519 | 5.3809 | 19000 | 1.1609 | 55.7461 |
| 0.0479 | 5.5225 | 19500 | 1.1737 | 54.6381 |
| 0.046 | 5.6641 | 20000 | 1.1773 | 55.1053 |
| 0.0448 | 5.8057 | 20500 | 1.1837 | 54.6413 |
| 0.0422 | 5.9473 | 21000 | 1.1810 | 55.3535 |
| 0.0247 | 6.0889 | 21500 | 1.2235 | 54.7095 |
| 0.0267 | 6.2305 | 22000 | 1.2358 | 54.3655 |
| 0.0222 | 6.3721 | 22500 | 1.2313 | 55.2367 |
| 0.0281 | 6.5137 | 23000 | 1.2455 | 54.3380 |
| 0.0231 | 6.6553 | 23500 | 1.2549 | 53.6096 |
| 0.0278 | 6.7969 | 24000 | 1.2543 | 53.9308 |
| 0.0276 | 6.9385 | 24500 | 1.2500 | 53.9389 |
| 0.0121 | 7.0801 | 25000 | 1.2829 | 53.9535 |
| 0.0164 | 7.2218 | 25500 | 1.2841 | 53.6031 |
| 0.0164 | 7.3634 | 26000 | 1.2826 | 55.3000 |
| 0.0139 | 7.5050 | 26500 | 1.2754 | 53.8237 |
| 0.0141 | 7.6466 | 27000 | 1.3143 | 54.2244 |
| 0.0157 | 7.7882 | 27500 | 1.2910 | 53.2981 |
| 0.0155 | 7.9298 | 28000 | 1.2914 | 54.4580 |
| 0.0083 | 8.0714 | 28500 | 1.3115 | 53.1066 |
| 0.0082 | 8.2130 | 29000 | 1.3224 | 53.2072 |
| 0.0088 | 8.3546 | 29500 | 1.3260 | 54.2455 |
| 0.008 | 8.4962 | 30000 | 1.3389 | 52.7579 |
| 0.0065 | 8.6378 | 30500 | 1.3449 | 54.0897 |
| 0.0086 | 8.7794 | 31000 | 1.3381 | 54.0719 |
| 0.0111 | 8.9210 | 31500 | 1.3331 | 53.9843 |
| 0.005 | 9.0626 | 32000 | 1.3531 | 53.3403 |
| 0.0033 | 9.2042 | 32500 | 1.3539 | 53.9535 |
| 0.0041 | 9.3458 | 33000 | 1.3654 | 52.9055 |
| 0.0027 | 9.4874 | 33500 | 1.3526 | 53.0628 |
| 0.0038 | 9.6290 | 34000 | 1.3802 | 52.8730 |
| 0.0054 | 9.7706 | 34500 | 1.3724 | 53.2235 |
| 0.003 | 9.9122 | 35000 | 1.3734 | 53.4635 |
| 0.0024 | 10.0538 | 35500 | 1.3829 | 52.5421 |
| 0.002 | 10.1954 | 36000 | 1.3788 | 52.9347 |
| 0.0034 | 10.3370 | 36500 | 1.3749 | 52.8146 |
| 0.0013 | 10.4786 | 37000 | 1.3951 | 52.4269 |
| 0.0014 | 10.6202 | 37500 | 1.3997 | 52.6865 |
| 0.0015 | 10.7618 | 38000 | 1.3900 | 52.6589 |
| 0.0013 | 10.9034 | 38500 | 1.4116 | 52.4513 |
| 0.0014 | 11.0450 | 39000 | 1.4101 | 52.2014 |
| 0.0032 | 11.1866 | 39500 | 1.4156 | 52.5113 |
| 0.0005 | 11.3282 | 40000 | 1.4236 | 52.3182 |
| 0.0003 | 11.4698 | 40500 | 1.4343 | 52.2890 |
| 0.0002 | 11.6114 | 41000 | 1.4376 | 52.2063 |
| 0.0003 | 11.7530 | 41500 | 1.4428 | 52.1998 |
| 0.0002 | 11.8946 | 42000 | 1.4392 | 52.2874 |
Framework versions
- Transformers 4.54.0.dev0
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.2