Whisper Tiny finetuned on PolyAI Minds14 English US
This model is a fine-tuned version of openai/whisper-tiny on the Speech Transcription in English from e-banking domain. dataset. It achieves the following results on the evaluation set:
- Loss: 0.8668
- Wer Ortho: 0.4009
- Wer: 0.3823
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: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 50
- training_steps: 400
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|---|---|---|---|---|---|
| 0.3501 | 3.57 | 100 | 0.7134 | 0.4568 | 0.4212 |
| 0.044 | 7.14 | 200 | 0.7639 | 0.4096 | 0.3746 |
| 0.0048 | 10.71 | 300 | 0.8265 | 0.4109 | 0.3854 |
| 0.0021 | 14.29 | 400 | 0.8668 | 0.4009 | 0.3823 |
Framework versions
- Transformers 4.36.0
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.15.0
- Downloads last month
- 8
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for Shamik/whisper-tiny-polyAI-minds14
Base model
openai/whisper-tinyDataset used to train Shamik/whisper-tiny-polyAI-minds14
Evaluation results
- Wer on Speech Transcription in English from e-banking domain.self-reported0.382