metadata
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- lyhourt/clean_3
metrics:
- wer
model-index:
- name: whisper-small-clean_3-400
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: lyhourt/clean_3
type: lyhourt/clean_3
metrics:
- name: Wer
type: wer
value: 4.053271569195136
whisper-small-clean_3-400
This model is a fine-tuned version of openai/whisper-small on the lyhourt/clean_3 dataset. It achieves the following results on the evaluation set:
- Loss: 0.0233
- Wer: 4.0533
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: 64
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- training_steps: 400
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0726 | 0.25 | 100 | 0.0732 | 11.2913 |
| 0.0477 | 0.5 | 200 | 0.0527 | 7.8170 |
| 0.0025 | 1.1425 | 300 | 0.0243 | 4.3428 |
| 0.0011 | 1.3925 | 400 | 0.0233 | 4.0533 |
Framework versions
- Transformers 4.41.0
- Pytorch 2.3.0+cu121
- Datasets 2.19.1
- Tokenizers 0.19.1