metadata
library_name: transformers
language:
- en
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- MLCommons/peoples_speech
metrics:
- wer
model-index:
- name: Fine Tune Whisper on People Speech
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: Peoples Speech
type: MLCommons/peoples_speech
args: 'config: English, split: test'
metrics:
- name: Wer
type: wer
value: 14.784595300261097
Fine Tune Whisper on People Speech
This model is a fine-tuned version of openai/whisper-small on the Peoples Speech dataset. It achieves the following results on the evaluation set:
- Loss: 0.5068
- Wer: 14.7846
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 with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0071 | 2.0 | 100 | 0.5080 | 14.2950 |
| 0.006 | 4.0 | 200 | 0.4859 | 14.1645 |
| 0.0012 | 6.0 | 300 | 0.4997 | 14.3603 |
| 0.0002 | 8.0 | 400 | 0.5017 | 14.4582 |
| 0.0005 | 10.0 | 500 | 0.5068 | 14.7846 |
Framework versions
- Transformers 4.52.0
- Pytorch 2.9.0+cu126
- Datasets 4.4.1
- Tokenizers 0.21.4