metadata
library_name: transformers
license: apache-2.0
base_model: openai/whisper-base
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-base-en
results: []
whisper-base-en
This model is a fine-tuned version of openai/whisper-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1334
- Wer: 6.9935
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: 1e-06
- train_batch_size: 8
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- training_steps: 3000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.5784 | 0.4 | 100 | 0.3881 | 20.8915 |
| 0.2412 | 0.8 | 200 | 0.2176 | 12.1310 |
| 0.1962 | 1.2 | 300 | 0.1909 | 10.6681 |
| 0.182 | 1.6 | 400 | 0.1782 | 9.7530 |
| 0.1683 | 2.0 | 500 | 0.1697 | 8.9785 |
| 0.1418 | 2.4 | 600 | 0.1639 | 8.9699 |
| 0.1605 | 2.8 | 700 | 0.1590 | 8.4593 |
| 0.13 | 3.2 | 800 | 0.1550 | 7.9774 |
| 0.1353 | 3.6 | 900 | 0.1518 | 7.7623 |
| 0.13 | 4.0 | 1000 | 0.1491 | 7.4897 |
| 0.1288 | 4.4 | 1100 | 0.1467 | 7.4897 |
| 0.12 | 4.8 | 1200 | 0.1448 | 7.4180 |
| 0.1161 | 5.2 | 1300 | 0.1428 | 7.3807 |
| 0.113 | 5.6 | 1400 | 0.1414 | 7.5356 |
| 0.1022 | 6.0 | 1500 | 0.1399 | 6.9505 |
| 0.1029 | 6.4 | 1600 | 0.1390 | 6.9361 |
| 0.0981 | 6.8 | 1700 | 0.1379 | 6.8070 |
| 0.1051 | 7.2 | 1800 | 0.1369 | 6.8357 |
| 0.0927 | 7.6 | 1900 | 0.1362 | 6.8988 |
| 0.0973 | 8.0 | 2000 | 0.1354 | 6.8042 |
| 0.0898 | 8.4 | 2100 | 0.1348 | 6.7497 |
| 0.0929 | 8.8 | 2200 | 0.1342 | 6.7870 |
| 0.0937 | 9.2 | 2300 | 0.1338 | 7.0623 |
| 0.0901 | 9.6 | 2400 | 0.1334 | 6.9935 |
Framework versions
- Transformers 4.46.2
- Pytorch 2.5.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3