jeju_stt
This model is a fine-tuned version of openai/whisper-base on the jeju_audio dataset. It achieves the following results on the evaluation set:
- Loss: 0.3820
- Cer: 12.0409
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 |
|---|---|---|---|---|
| 0.3689 | 2.0 | 1000 | 0.3853 | 13.4054 |
| 0.1884 | 4.0 | 2000 | 0.3488 | 11.9817 |
| 0.1059 | 6.0 | 3000 | 0.3607 | 11.9350 |
| 0.0634 | 8.0 | 4000 | 0.3820 | 12.0409 |
Framework versions
- Transformers 4.42.0.dev0
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.19.1
- Downloads last month
- 3
Model tree for rlaorrn/jeju_stt_v2
Base model
openai/whisper-base