<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
stt-v11-medium
This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3701
- Wer: 0.2216
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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 271
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.8041 | 14.28 | 500 | 0.3662 | 0.4315 |
| 0.3702 | 28.56 | 1000 | 0.3102 | 0.2966 |
| 0.1978 | 42.85 | 1500 | 0.3378 | 0.2794 |
| 0.1467 | 57.14 | 2000 | 0.3201 | 0.2808 |
| 0.1144 | 71.42 | 2500 | 0.3646 | 0.2698 |
| 0.0969 | 85.7 | 3000 | 0.3234 | 0.2657 |
| 0.0832 | 99.99 | 3500 | 0.3744 | 0.2712 |
| 0.0732 | 114.28 | 4000 | 0.3217 | 0.2602 |
| 0.0635 | 128.56 | 4500 | 0.3419 | 0.2491 |
| 0.0561 | 142.85 | 5000 | 0.3628 | 0.2560 |
| 0.0491 | 157.14 | 5500 | 0.3458 | 0.2436 |
| 0.0439 | 171.42 | 6000 | 0.3615 | 0.2519 |
| 0.0397 | 185.7 | 6500 | 0.3610 | 0.2519 |
| 0.0352 | 199.99 | 7000 | 0.3514 | 0.2374 |
| 0.0314 | 214.28 | 7500 | 0.3469 | 0.2450 |
| 0.0272 | 228.56 | 8000 | 0.3615 | 0.2271 |
| 0.0247 | 242.85 | 8500 | 0.3614 | 0.2292 |
| 0.022 | 257.14 | 9000 | 0.3701 | 0.2216 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.1+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2
