whisper-small-ko-Yfreq_SA
This model is a fine-tuned version of openai/whisper-small on the aihub adult speed changed dataset. It achieves the following results on the evaluation set:
- Loss: 0.2582
- Cer: 6.8433
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: 1e-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: 2
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Cer |
---|---|---|---|---|
0.314 | 0.13 | 100 | 0.2988 | 7.3426 |
0.2117 | 0.26 | 200 | 0.2714 | 7.1664 |
0.1585 | 0.39 | 300 | 0.2759 | 7.8654 |
0.1633 | 0.52 | 400 | 0.2599 | 7.3015 |
0.1411 | 0.64 | 500 | 0.2663 | 7.2192 |
0.1323 | 0.77 | 600 | 0.2674 | 7.1370 |
0.1354 | 0.9 | 700 | 0.2632 | 7.1781 |
0.082 | 1.03 | 800 | 0.2599 | 7.0841 |
0.0647 | 1.16 | 900 | 0.2590 | 6.9196 |
0.078 | 1.29 | 1000 | 0.2601 | 6.8961 |
0.0671 | 1.42 | 1100 | 0.2593 | 6.8198 |
0.0716 | 1.55 | 1200 | 0.2576 | 6.8609 |
0.0699 | 1.68 | 1300 | 0.2581 | 6.7493 |
0.0682 | 1.81 | 1400 | 0.2574 | 6.8785 |
0.0713 | 1.93 | 1500 | 0.2582 | 6.8433 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 2.1.0+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0
- Downloads last month
- 4