metadata
language:
- ko
license: apache-2.0
base_model: openai/whisper-base
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- potatoSeop/chimsuja_dataset
model-index:
- name: whisper-chimsuja
results: []
whisper-chimsuja
This model is a fine-tuned version of openai/whisper-base on the potatoSeop/chimsuja_dataset dataset. It achieves the following results on the evaluation set:
- Loss: 0.3795
- Cer: 12.0513
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
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.913 | 0.4 | 50 | 0.4665 | 14.4249 |
| 0.4203 | 0.79 | 100 | 0.4156 | 18.1852 |
| 0.343 | 1.19 | 150 | 0.3972 | 13.9544 |
| 0.2746 | 1.59 | 200 | 0.3891 | 13.2048 |
| 0.2838 | 1.98 | 250 | 0.3827 | 13.8794 |
| 0.2153 | 2.38 | 300 | 0.3808 | 12.0721 |
| 0.2035 | 2.78 | 350 | 0.3795 | 12.0513 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.16.0
- Tokenizers 0.15.0