<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
whisper-base-zh-20230715-1 - au2a
This model is a fine-tuned version of openai/whisper-base on the some hakka audio dataset. It achieves the following results on the evaluation set:
- Loss: 0.5128
- Cer: 65.2716
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: 5e-06
- train_batch_size: 64
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 10000
Training results
| Training Loss | Epoch | Step | Validation Loss | Cer |
|---|---|---|---|---|
| 0.2461 | 2.59 | 1000 | 0.5164 | 34.5962 |
| 0.0686 | 5.17 | 2000 | 0.4523 | 35.0268 |
| 0.0187 | 7.76 | 3000 | 0.4622 | 48.4098 |
| 0.0064 | 10.35 | 4000 | 0.4741 | 62.4008 |
| 0.0037 | 12.94 | 5000 | 0.4820 | 56.8256 |
| 0.0023 | 15.52 | 6000 | 0.4922 | 63.3452 |
| 0.0016 | 18.11 | 7000 | 0.4992 | 60.8597 |
| 0.0012 | 20.7 | 8000 | 0.5073 | 59.6472 |
| 0.0009 | 23.29 | 9000 | 0.5108 | 64.7465 |
| 0.0009 | 25.87 | 10000 | 0.5128 | 65.2716 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
