<!-- 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-EYspeed
This model is a fine-tuned version of openai/whisper-small on the aihub Y dialogue dataset. It achieves the following results on the evaluation set:
- Loss: 0.2135
- Cer: 4.8893
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 |
|---|---|---|---|---|
| 0.3381 | 0.3 | 100 | 0.2442 | 6.2244 |
| 0.3013 | 0.59 | 200 | 0.2311 | 6.2481 |
| 0.2723 | 0.89 | 300 | 0.2248 | 6.2007 |
| 0.1081 | 1.19 | 400 | 0.2230 | 6.0108 |
| 0.1041 | 1.48 | 500 | 0.2195 | 5.8684 |
| 0.0897 | 1.78 | 600 | 0.2099 | 5.3047 |
| 0.0341 | 2.07 | 700 | 0.2051 | 5.3522 |
| 0.0366 | 2.37 | 800 | 0.2091 | 4.6995 |
| 0.0339 | 2.67 | 900 | 0.2096 | 5.2454 |
| 0.0332 | 2.96 | 1000 | 0.2048 | 5.3284 |
| 0.0128 | 3.26 | 1100 | 0.2150 | 4.9665 |
| 0.0107 | 3.56 | 1200 | 0.2154 | 4.9902 |
| 0.0112 | 3.85 | 1300 | 0.2122 | 5.0199 |
| 0.006 | 4.15 | 1400 | 0.2109 | 6.8415 |
| 0.0044 | 4.44 | 1500 | 0.2118 | 5.5776 |
| 0.0037 | 4.74 | 1600 | 0.2135 | 4.8893 |
Framework versions
- Transformers 4.34.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3
