nick_asr_COMBO
This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.4313
- Wer: 0.6723
- Cer: 0.2408
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: 5e-05
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|---|---|---|---|---|---|
| 0.0857 | 1.0 | 687 | 1.3883 | 0.7082 | 0.2576 |
| 0.0627 | 2.0 | 1374 | 1.4099 | 0.7076 | 0.2561 |
| 0.0697 | 3.0 | 2061 | 1.3864 | 0.6906 | 0.2486 |
| 0.0575 | 4.0 | 2748 | 1.4356 | 0.6906 | 0.2455 |
| 0.0552 | 5.0 | 3435 | 1.4061 | 0.6778 | 0.2440 |
| 0.0631 | 6.0 | 4122 | 1.4541 | 0.6839 | 0.2444 |
| 0.0418 | 7.0 | 4809 | 1.4258 | 0.6930 | 0.2465 |
| 0.0407 | 8.0 | 5496 | 1.4193 | 0.6809 | 0.2451 |
| 0.0487 | 9.0 | 6183 | 1.4261 | 0.6778 | 0.2424 |
| 0.0371 | 10.0 | 6870 | 1.4313 | 0.6723 | 0.2408 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.10.0+cu111
- Datasets 2.1.0
- Tokenizers 0.12.1