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_1 dataset. It achieves the following results on the evaluation set:
- Loss: 0.3707
- Cer: 12.6289
- Wer: 36.7564
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 |
|---|---|---|---|---|---|
| 0.2787 | 0.32 | 500 | 14.5190 | 0.4086 | 39.9745 |
| 0.2996 | 0.64 | 1000 | 13.7403 | 0.3984 | 38.7252 |
| 0.3226 | 0.96 | 1500 | 13.8005 | 0.3772 | 38.4629 |
| 0.2281 | 1.28 | 2000 | 13.0192 | 0.3682 | 37.1511 |
| 0.2242 | 1.6 | 2500 | 12.9577 | 0.3762 | 37.2961 |
| 0.2284 | 1.92 | 3000 | 0.3733 | 12.7289 | 36.4465 |
| 0.1648 | 2.24 | 3500 | 0.3720 | 12.8054 | 36.9687 |
| 0.173 | 2.56 | 4000 | 0.3707 | 12.6289 | 36.7564 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.15.0
- Tokenizers 0.15.0