This model was trained from scratch on the librispeech_asr dataset. It achieves the following results on the evaluation set:
- Loss: 1.8529
- Wer: 0.9977
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: 4
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 20.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.5506 | 2.8 | 2500 | 4.4928 | 1.8772 |
| 0.5145 | 5.61 | 5000 | 1.8942 | 1.1063 |
| 0.2736 | 8.41 | 7500 | 1.6550 | 1.0372 |
| 0.0807 | 11.21 | 10000 | 1.7601 | 1.0004 |
| 0.0439 | 14.01 | 12500 | 1.8014 | 1.0022 |
| 0.043 | 16.82 | 15000 | 1.8534 | 1.0097 |
| 0.0434 | 19.62 | 17500 | 1.8529 | 0.9977 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu113
- Datasets 1.18.3
- Tokenizers 0.11.0