metadata
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: whisper-large-v2-japanese-24h
results: []
whisper-large-v2-japanese-24h
This model is a fine-tuned version of openai/whisper-large-v2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.4200
- Wer: 0.7449
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: 50
- 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: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0111 | 7.63 | 1000 | 0.3210 | 0.7888 |
| 0.0007 | 15.27 | 2000 | 0.3585 | 0.7478 |
| 0.0003 | 22.9 | 3000 | 0.3937 | 0.7432 |
| 0.0002 | 30.53 | 4000 | 0.4123 | 0.7443 |
| 0.0002 | 38.17 | 5000 | 0.4200 | 0.7449 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.1
- Datasets 2.8.1.dev0
- Tokenizers 0.13.2