metadata
language:
- ko
license: apache-2.0
base_model: openai/whisper-base
tags:
- hf-asr-leaderboard
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper_finetune
results: []
whisper_finetune
This model is a fine-tuned version of openai/whisper-base on the aihub_3 dataset. It achieves the following results on the evaluation set:
- Loss: 0.3587
- Cer: 11.8692
- Wer: 34.6801
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: 32
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Cer | Validation Loss | Wer |
|---|---|---|---|---|---|
| 0.267 | 0.4 | 500 | 11.9783 | 0.3521 | 35.1998 |
| 0.2392 | 0.8 | 1000 | 12.1614 | 0.3495 | 34.9449 |
| 0.171 | 1.2 | 1500 | 12.0633 | 0.3516 | 35.2048 |
| 0.1744 | 1.6 | 2000 | 0.3553 | 12.2091 | 35.0598 |
| 0.1722 | 2.0 | 2500 | 0.3515 | 12.0222 | 34.5426 |
| 0.1192 | 2.4 | 3000 | 0.3594 | 12.2281 | 35.4796 |
| 0.1249 | 2.8 | 3500 | 0.3609 | 12.0137 | 34.8949 |
| 0.0858 | 3.2 | 4000 | 0.3587 | 11.8692 | 34.6801 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.15.0
- Tokenizers 0.15.0