metadata
language:
- ko
license: apache-2.0
base_model: openai/whisper-small
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- Yettiesoft/voice_medical_origin_small_vector
model-index:
- name: jazzhong1_medical_whisper_small
results: []
jazzhong1_medical_whisper_small
This model is a fine-tuned version of openai/whisper-small on the voice_medical_origin_small_vector dataset. It achieves the following results on the evaluation set:
- Loss: 0.2648
- Cer: 10.1284
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: 500
- training_steps: 4000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.2886 | 0.4342 | 1000 | 0.3111 | 11.4993 |
| 0.3306 | 0.8684 | 2000 | 0.2815 | 10.8260 |
| 0.215 | 1.3026 | 3000 | 0.2702 | 10.0223 |
| 0.2068 | 1.7369 | 4000 | 0.2648 | 10.1284 |
Framework versions
- Transformers 4.42.0.dev0
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1