openai/whisper-medium
This model is a fine-tuned version of openai/whisper-medium on the pphuc25/FrenchMed dataset. It achieves the following results on the evaluation set:
- Loss: 1.8094
- Wer: 45.6012
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: 8
- 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: 100
- num_epochs: 20
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
1.2928 | 1.0 | 215 | 1.2902 | 56.6716 |
0.8083 | 2.0 | 430 | 1.4740 | 75.1466 |
0.4531 | 3.0 | 645 | 1.4275 | 64.5161 |
0.316 | 4.0 | 860 | 1.6528 | 53.3724 |
0.1942 | 5.0 | 1075 | 1.7240 | 61.3636 |
0.1557 | 6.0 | 1290 | 1.6985 | 46.1877 |
0.1254 | 7.0 | 1505 | 1.8613 | 52.6393 |
0.1052 | 8.0 | 1720 | 1.7694 | 50.6598 |
0.0719 | 9.0 | 1935 | 1.7321 | 45.8944 |
0.0606 | 10.0 | 2150 | 1.8430 | 49.7801 |
0.0446 | 11.0 | 2365 | 1.8449 | 49.7801 |
0.0387 | 12.0 | 2580 | 1.8400 | 51.6862 |
0.0305 | 13.0 | 2795 | 1.8258 | 57.1114 |
0.0138 | 14.0 | 3010 | 1.9455 | 50.1466 |
0.0104 | 15.0 | 3225 | 1.7864 | 50.8065 |
0.0117 | 16.0 | 3440 | 1.8213 | 46.3343 |
0.0034 | 17.0 | 3655 | 1.7827 | 44.5748 |
0.0023 | 18.0 | 3870 | 1.7990 | 44.2082 |
0.0007 | 19.0 | 4085 | 1.8095 | 44.4282 |
0.0008 | 20.0 | 4300 | 1.8094 | 45.6012 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.19.1
- Downloads last month
- 8