las_asr-scr_w2v2-base_001
This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.5063
- Per: 0.1385
- Pcc: 0.7152
- Ctc Loss: 0.4413
- Mse Loss: 1.0747
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: 4
- eval_batch_size: 1
- seed: 1111
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 742
- training_steps: 7420
Training results
| Training Loss | Epoch | Step | Validation Loss | Per | Pcc | Ctc Loss | Mse Loss |
|---|---|---|---|---|---|---|---|
| 13.8511 | 1.0 | 742 | 4.6856 | 0.9897 | 0.6425 | 3.8063 | 1.0726 |
| 4.3974 | 2.0 | 1484 | 3.9341 | 0.9897 | 0.6978 | 3.3804 | 0.9671 |
| 2.3337 | 3.0 | 2226 | 1.9487 | 0.2101 | 0.7032 | 0.7264 | 1.1612 |
| 1.349 | 4.0 | 2968 | 1.9665 | 0.1870 | 0.7096 | 0.5798 | 1.2719 |
| 0.9566 | 5.0 | 3710 | 2.3409 | 0.1694 | 0.7104 | 0.5209 | 1.5863 |
| 0.618 | 6.0 | 4452 | 1.9887 | 0.1549 | 0.7153 | 0.4887 | 1.3187 |
| 0.3086 | 7.0 | 5194 | 1.4061 | 0.1520 | 0.7152 | 0.4632 | 0.9542 |
| 0.0495 | 8.0 | 5936 | 1.5966 | 0.1487 | 0.7123 | 0.4548 | 1.1008 |
| -0.1576 | 9.0 | 6678 | 1.5246 | 0.1443 | 0.7182 | 0.4423 | 1.0794 |
| -0.2832 | 10.0 | 7420 | 1.5063 | 0.1428 | 0.7167 | 0.4413 | 1.0747 |
Framework versions
- Transformers 4.38.1
- Pytorch 2.0.1
- Datasets 2.16.1
- Tokenizers 0.15.2
- Downloads last month
- 2
Model tree for excalibur12/las_asr-scr_w2v2-base_001
Base model
facebook/wav2vec2-base