metadata
language:
- ko
license: apache-2.0
tags:
- hf-asr-leaderboard
- generated_from_trainer
base_model: openai/whisper-medium
datasets:
- Marcusxx/cheonanAddresses
model-index:
- name: cheonanAddresses_torchmodel_model
results: []
cheonanAddresses_torchmodel_model
This model is a fine-tuned version of openai/whisper-medium on the Marcusxx/cheonanAddresses dataset. It achieves the following results on the evaluation set:
- Loss: 0.0531
- Cer: 14.2607
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: 100
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.0765 | 0.3101 | 1000 | 0.0767 | 27.7467 |
| 0.0698 | 0.6202 | 2000 | 0.0717 | 2.6100 |
| 0.0624 | 0.9302 | 3000 | 0.0667 | 3.7811 |
| 0.0597 | 1.2403 | 4000 | 0.0629 | 2.1665 |
| 0.0598 | 1.5504 | 5000 | 0.0602 | 2.0783 |
| 0.0557 | 1.8605 | 6000 | 0.0586 | 5.0203 |
| 0.0512 | 2.1705 | 7000 | 0.0566 | 16.9546 |
| 0.0441 | 2.4806 | 8000 | 0.0558 | 13.6386 |
| 0.0533 | 2.7907 | 9000 | 0.0537 | 13.8057 |
| 0.0376 | 3.1008 | 10000 | 0.0531 | 14.2607 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.2.2+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1