metadata
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
model-index:
- name: Whisper-medium-Arabic-phoneme-v3
results: []
Whisper-medium-Arabic-phoneme-v3
This model is a fine-tuned version of openai/whisper-medium on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2355
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: 0.001
- train_batch_size: 6
- eval_batch_size: 6
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 0.0973 | 1.0 | 546 | 0.2571 |
| 0.0419 | 2.0 | 1092 | 0.1942 |
| 0.0257 | 3.0 | 1638 | 0.2018 |
| 0.0334 | 4.0 | 2184 | 0.2066 |
| 0.0313 | 5.0 | 2730 | 0.2133 |
| 0.0169 | 6.0 | 3276 | 0.2063 |
| 0.0116 | 7.0 | 3822 | 0.2086 |
| 0.0042 | 8.0 | 4368 | 0.2185 |
| 0.0031 | 9.0 | 4914 | 0.2284 |
| 0.0019 | 10.0 | 5460 | 0.2355 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0