metadata
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- generated_from_trainer
model-index:
- name: Whisper-tiny-Arabic-phoneme
results: []
Whisper-tiny-Arabic-phoneme
This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2115
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 |
|---|---|---|---|
| 0.0703 | 1.0 | 546 | 0.2243 |
| 0.0452 | 2.0 | 1092 | 0.2117 |
| 0.0273 | 3.0 | 1638 | 0.1945 |
| 0.0239 | 4.0 | 2184 | 0.2105 |
| 0.0181 | 5.0 | 2730 | 0.1959 |
| 0.0124 | 6.0 | 3276 | 0.2083 |
| 0.0132 | 7.0 | 3822 | 0.2026 |
| 0.0066 | 8.0 | 4368 | 0.2110 |
| 0.0055 | 9.0 | 4914 | 0.2118 |
| 0.0029 | 10.0 | 5460 | 0.2115 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0