whisper-tiny-akan
This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.1467
- Wer: 45.0456
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: 0.0001
- train_batch_size: 32
- 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: 2000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.5414 | 10.0 | 250 | 0.7596 | 55.6922 |
| 0.09 | 20.0 | 500 | 0.9048 | 52.3362 |
| 0.0275 | 30.0 | 750 | 1.0300 | 49.0091 |
| 0.0114 | 40.0 | 1000 | 1.0959 | 47.4902 |
| 0.004 | 50.0 | 1250 | 1.1247 | 45.3783 |
| 0.0017 | 60.0 | 1500 | 1.1412 | 45.3493 |
| 0.0009 | 70.0 | 1750 | 1.1437 | 44.6261 |
| 0.0007 | 80.0 | 2000 | 1.1467 | 45.0456 |
Framework versions
- Transformers 4.44.2
- Pytorch 2.4.0+cu121
- Datasets 3.0.0
- Tokenizers 0.19.1
- Downloads last month
- 14
Inference API (serverless) is not available, repository is disabled.
Model tree for devkyle/whisper-3000ms-v2
Base model
openai/whisper-tiny
Finetuned
this model