metadata
license: apache-2.0
base_model: openai/whisper-medium
tags:
- generated_from_trainer
metrics:
- bleu
- wer
model-index:
- name: Whisper Medium GA-EN Speech Translation
results: []
Whisper Medium GA-EN Speech Translation
This model is a fine-tuned version of openai/whisper-medium on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 3.6073
- Bleu: 0.22
- Chrf: 12.93
- Wer: 104.3224
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: 0.0003
- 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: 0.03
- training_steps: 1000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Chrf | Wer |
|---|---|---|---|---|---|---|
| 4.698 | 0.11 | 100 | 4.6156 | 0.0 | 1.58 | 1852.0486 |
| 4.2397 | 0.22 | 200 | 4.0051 | 0.06 | 4.23 | 206.9338 |
| 3.8768 | 0.32 | 300 | 3.7864 | 0.18 | 5.21 | 97.9739 |
| 3.8755 | 0.43 | 400 | 3.6008 | 0.19 | 5.28 | 96.0378 |
| 3.529 | 0.54 | 500 | 3.5774 | 0.29 | 6.96 | 108.0144 |
| 3.32 | 0.65 | 600 | 3.5197 | 0.23 | 8.97 | 99.3697 |
| 3.3073 | 0.76 | 700 | 3.5056 | 0.49 | 9.43 | 104.0973 |
| 3.1893 | 0.86 | 800 | 3.5861 | 0.32 | 12.86 | 114.3629 |
| 3.0739 | 0.97 | 900 | 3.5003 | 0.6 | 12.1 | 101.3057 |
| 2.7718 | 1.08 | 1000 | 3.6073 | 0.22 | 12.93 | 104.3224 |
Framework versions
- Transformers 4.39.3
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2