hubert-large-ls960-ft-timit
This model is a fine-tuned version of facebook/hubert-large-ls960-ft on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1074
- Wer: 0.1708
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
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- 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 |
|---|---|---|---|---|
| 6.2278 | 4.0 | 500 | 2.6282 | 0.9999 |
| 0.9389 | 8.0 | 1000 | 0.1154 | 0.2096 |
| 0.2005 | 12.0 | 1500 | 0.0951 | 0.1732 |
| 0.1985 | 16.0 | 2000 | 0.0974 | 0.1759 |
| 0.124 | 20.0 | 2500 | 0.0951 | 0.1728 |
| 0.0797 | 24.0 | 3000 | 0.1064 | 0.1713 |
| 0.1047 | 28.0 | 3500 | 0.1074 | 0.1708 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.10.0+cu111
- Datasets 1.18.3
- Tokenizers 0.10.3