<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
Whisper Small ko-Yfreq-E - syp1229
This model is a fine-tuned version of openai/whisper-small on the aihub Y E dialogue dataset. It achieves the following results on the evaluation set:
- Loss: 0.2157
- Cer: 0.0491
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: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.2455 | 0.3 | 100 | 0.2528 | 0.0663 |
| 0.2591 | 0.59 | 200 | 0.2452 | 0.0646 |
| 0.1702 | 0.89 | 300 | 0.2298 | 0.0628 |
| 0.0738 | 1.19 | 400 | 0.2136 | 0.0923 |
| 0.0957 | 1.48 | 500 | 0.2263 | 0.0618 |
| 0.0729 | 1.78 | 600 | 0.2139 | 0.0565 |
| 0.0242 | 2.07 | 700 | 0.2073 | 0.0520 |
| 0.028 | 2.37 | 800 | 0.2063 | 0.0482 |
| 0.0351 | 2.67 | 900 | 0.2162 | 0.0506 |
| 0.0239 | 2.96 | 1000 | 0.2075 | 0.0513 |
| 0.0088 | 3.26 | 1100 | 0.2194 | 0.0495 |
| 0.0079 | 3.56 | 1200 | 0.2187 | 0.0508 |
| 0.0072 | 3.85 | 1300 | 0.2217 | 0.0510 |
| 0.0046 | 4.15 | 1400 | 0.2164 | 0.0488 |
| 0.0038 | 4.44 | 1500 | 0.2149 | 0.0490 |
| 0.003 | 4.74 | 1600 | 0.2157 | 0.0491 |
Framework versions
- Transformers 4.34.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3
