wav2vec2-base-timit-demo-colab
This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4806
- Wer: 0.2866
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
- gradient_accumulation_steps: 2
- 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: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 5.7021 | 2.2026 | 500 | 2.5190 | 1.0 |
| 0.8944 | 4.4053 | 1000 | 0.5051 | 0.4244 |
| 0.4386 | 6.6079 | 1500 | 0.4115 | 0.3509 |
| 0.3094 | 8.8106 | 2000 | 0.4210 | 0.3329 |
| 0.2297 | 11.0132 | 2500 | 0.4137 | 0.3211 |
| 0.1723 | 13.2159 | 3000 | 0.4483 | 0.3150 |
| 0.1329 | 15.4185 | 3500 | 0.4413 | 0.3099 |
| 0.1061 | 17.6211 | 4000 | 0.4707 | 0.3045 |
| 0.0934 | 19.8238 | 4500 | 0.4784 | 0.3015 |
| 0.0823 | 22.0264 | 5000 | 0.4966 | 0.2984 |
| 0.0664 | 24.2291 | 5500 | 0.4878 | 0.2950 |
| 0.0572 | 26.4317 | 6000 | 0.4820 | 0.2883 |
| 0.0495 | 28.6344 | 6500 | 0.4876 | 0.2868 |
| 0.0536 | 30.0 | 6810 | 0.4806 | 0.2866 |
Framework versions
- Transformers 5.5.4
- Pytorch 2.11.0+cu130
- Datasets 4.8.4
- Tokenizers 0.22.2