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: 4.6263
- Wer: 0.8568
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: 3e-05
- 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 |
|---|---|---|---|---|
| 5.9849 | 1.68 | 1500 | 5.9623 | 1.1028 |
| 5.1696 | 3.36 | 3000 | 5.5504 | 1.6345 |
| 4.1412 | 5.04 | 4500 | 5.3853 | 1.3565 |
| 2.7226 | 6.73 | 6000 | 5.3072 | 0.9908 |
| 3.2607 | 8.41 | 7500 | 5.4121 | 1.2854 |
| 2.4017 | 10.09 | 9000 | 5.1094 | 1.0303 |
| 1.7361 | 11.77 | 10500 | 4.8928 | 0.9506 |
| 2.0638 | 13.45 | 12000 | 4.8352 | 0.9127 |
| 1.2832 | 15.13 | 13500 | 4.7271 | 0.9103 |
| 1.0439 | 16.82 | 15000 | 4.5980 | 0.8720 |
| 0.4112 | 18.5 | 16500 | 4.6263 | 0.8568 |
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu113
- Datasets 1.18.3
- Tokenizers 0.11.0