metadata
language:
- ko
license: apache-2.0
tags:
- hf-asr-leaderboard
- generated_from_trainer
base_model: openai/whisper-large
datasets:
- Marcusxx/gwanju
model-index:
- name: gwanju_large_model
results: []
gwanju_large_model
This model is a fine-tuned version of openai/whisper-large on the Marcusxx/gwanju dataset. It achieves the following results on the evaluation set:
- Loss: 0.3547
- Cer: 395.4097
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: 16
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 250
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.4409 | 0.2964 | 1000 | 0.4335 | 106.5804 |
| 0.3824 | 0.5928 | 2000 | 0.3930 | 442.5527 |
| 0.3757 | 0.8892 | 3000 | 0.3727 | 455.4447 |
| 0.2271 | 1.1855 | 4000 | 0.3712 | 463.1818 |
| 0.2528 | 1.4819 | 5000 | 0.3600 | 468.4532 |
| 0.2068 | 1.7783 | 6000 | 0.3523 | 468.7220 |
| 0.1221 | 2.0747 | 7000 | 0.3592 | 480.3038 |
| 0.1157 | 2.3711 | 8000 | 0.3614 | 377.9116 |
| 0.121 | 2.6675 | 9000 | 0.3579 | 401.4220 |
| 0.1046 | 2.9638 | 10000 | 0.3547 | 395.4097 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.2.2+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1