metadata
license: mit
base_model: distil-whisper/distil-small.en
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: distill-whisper-jargon
results: []
distill-whisper-jargon
This model is a fine-tuned version of distil-whisper/distil-small.en on the btemirov/fin-terms dataset. It achieves the following results on the evaluation set:
- Loss: 4.4314
- Wer: 78.4173
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: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 200
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.8199 | 22.22 | 100 | 3.5227 | 79.3525 |
| 2.3504 | 44.44 | 200 | 3.7073 | 77.3022 |
| 1.4612 | 66.67 | 300 | 4.1042 | 78.6691 |
| 0.9713 | 88.89 | 400 | 4.3164 | 77.7698 |
| 0.7453 | 111.11 | 500 | 4.4314 | 78.4173 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.0+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0