metadata
library_name: transformers
language:
- fr
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: 52Hz Small Fr - IMT Atlantique X 52 Hertz
results: []
52Hz Small Fr - IMT Atlantique X 52 Hertz
This model is a fine-tuned version of openai/whisper-small on the Transcriptions IMTx52Hz v3 dataset. It achieves the following results on the evaluation set:
- Loss: 0.4085
- Wer: 17.6471
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: 2e-05
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 1.461 | 1.0 | 22 | 0.8945 | 41.0131 |
| 0.6454 | 2.0 | 44 | 0.5209 | 25.4902 |
| 0.3068 | 3.0 | 66 | 0.4752 | 21.0784 |
| 0.2361 | 4.0 | 88 | 0.4253 | 19.9346 |
| 0.1448 | 5.0 | 110 | 0.4112 | 19.7712 |
| 0.1367 | 6.0 | 132 | 0.4118 | 17.9739 |
| 0.0945 | 7.0 | 154 | 0.4220 | 19.6078 |
| 0.0575 | 8.0 | 176 | 0.4065 | 16.9935 |
| 0.0421 | 9.0 | 198 | 0.4356 | 17.6471 |
| 0.0414 | 10.0 | 220 | 0.4186 | 17.8105 |
| 0.027 | 11.0 | 242 | 0.4105 | 17.3203 |
| 0.0249 | 12.0 | 264 | 0.4037 | 16.9935 |
| 0.0314 | 13.0 | 286 | 0.4077 | 17.3203 |
| 0.0214 | 14.0 | 308 | 0.4085 | 17.6471 |
| 0.0148 | 15.0 | 330 | 0.4085 | 17.6471 |
Framework versions
- Transformers 4.57.3
- Pytorch 2.9.1+cu130
- Datasets 4.4.2
- Tokenizers 0.22.2