metadata
language:
- ko
license: apache-2.0
base_model: openai/whisper-base
tags:
- hf-asr-leaderboard
- generated_from_trainer
model-index:
- name: whisper_finetune
results: []
whisper_finetune
This model is a fine-tuned version of openai/whisper-base on the aihub_20000 dataset. It achieves the following results on the evaluation set:
- Loss: 0.4938
- Cer: 14.1359
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 | Validation Loss | Cer |
|---|---|---|---|---|
| 0.3736 | 0.8 | 500 | 0.4641 | 15.1146 |
| 0.2685 | 1.6 | 1000 | 0.4525 | 14.3150 |
| 0.1929 | 2.4 | 1500 | 0.4424 | 14.0470 |
| 0.1241 | 3.2 | 2000 | 0.4584 | 13.8724 |
| 0.1347 | 4.0 | 2500 | 0.4554 | 13.9110 |
| 0.0944 | 4.8 | 3000 | 0.4693 | 14.0391 |
| 0.0761 | 5.6 | 3500 | 0.4851 | 14.1281 |
| 0.0606 | 6.4 | 4000 | 0.4938 | 14.1359 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.15.0
- Tokenizers 0.15.0