whisper-tiny-minds14-en-us
This model is a fine-tuned version of openai/whisper-tiny on the PolyAI/minds14 dataset. It achieves the following results on the evaluation set:
- Loss: 0.6155
- Wer: 0.3211
- Wer Ortho: 0.3263
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: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 50
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer | Wer Ortho |
|---|---|---|---|---|---|
| 0.2739 | 3.4483 | 100 | 0.4633 | 0.3400 | 0.3523 |
| 0.0319 | 6.8966 | 200 | 0.5263 | 0.3294 | 0.3405 |
| 0.0039 | 10.3448 | 300 | 0.6010 | 0.3288 | 0.3350 |
| 0.0012 | 13.7931 | 400 | 0.6155 | 0.3211 | 0.3263 |
| 0.0009 | 17.2414 | 500 | 0.6408 | 0.3235 | 0.3282 |
Framework versions
- Transformers 4.57.6
- Pytorch 2.8.0+cu128
- Datasets 4.8.4
- Tokenizers 0.22.2
- Downloads last month
- 25
Safetensors
Model size
37.8M params
Tensor type
F32
·
Model tree for VoicesColeby/whisper-tiny-minds14-en-us
Base model
openai/whisper-tinyDataset used to train VoicesColeby/whisper-tiny-minds14-en-us
Evaluation results
- Wer on PolyAI/minds14self-reported0.321