base-asr-with-tpbs-cv3
This model is a fine-tuned version of Botnoi/base-asr-with-lm on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.5934
- Wer: 0.2629
- Cer: 0.0756
- Clean Cer: 0.0696
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: 32
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer | Clean Cer |
|---|---|---|---|---|---|---|
| 0.1269 | 26.67 | 400 | 0.5346 | 0.2723 | 0.0793 | 0.0726 |
| 0.0436 | 53.33 | 800 | 0.5695 | 0.2647 | 0.0767 | 0.0705 |
| 0.0287 | 80.0 | 1200 | 0.5934 | 0.2629 | 0.0756 | 0.0696 |
Framework versions
- Transformers 4.27.4
- Pytorch 1.13.1+cu117
- Datasets 2.11.0
- Tokenizers 0.13.2