metadata
tags:
- generated_from_trainer
datasets:
- librispeech_asr
model-index:
- name: ''
results: []
This model was trained from scratch on the librispeech_asr dataset. It achieves the following results on the evaluation set:
- Loss: 0.3123
- Wer: 0.0908
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: 0.0003
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- 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: 500
- num_epochs: 3.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 4.4937 | 0.28 | 500 | 5.2312 | 0.9660 |
| 3.821 | 0.56 | 1000 | 4.5810 | 0.9066 |
| 1.2129 | 0.84 | 1500 | 1.3723 | 0.3928 |
| 0.6575 | 1.12 | 2000 | 0.6645 | 0.1810 |
| 0.489 | 1.4 | 2500 | 0.5523 | 0.1479 |
| 0.3541 | 1.68 | 3000 | 0.4585 | 0.1195 |
| 0.3573 | 1.96 | 3500 | 0.3859 | 0.1066 |
| 0.2437 | 2.24 | 4000 | 0.3747 | 0.1015 |
| 0.1406 | 2.52 | 4500 | 0.3346 | 0.0952 |
| 0.1468 | 2.8 | 5000 | 0.3123 | 0.0908 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu113
- Datasets 1.18.3
- Tokenizers 0.11.0