facebook_large_CV_bn3
This model is a fine-tuned version of Sameen53/facebook_large_CV_bn on the common_voice dataset. It achieves the following results on the evaluation set:
- Loss: 0.2308
- Wer: 0.2379
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: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- 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: 200
- num_epochs: 6
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| No log | 0.87 | 1000 | 0.2473 | 0.2524 |
| 0.2308 | 1.73 | 2000 | 0.2073 | 0.2450 |
| 0.261 | 2.6 | 3000 | 0.2036 | 0.2345 |
| 0.2498 | 3.47 | 4000 | 0.1916 | 0.2311 |
| 0.2433 | 4.33 | 5000 | 0.1869 | 0.2344 |
| 0.2588 | 5.2 | 6000 | 0.2308 | 0.2379 |
Framework versions
- Transformers 4.20.1
- Pytorch 1.11.0
- Datasets 2.1.0
- Tokenizers 0.12.1