metadata
library_name: transformers
language:
- en
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- hf-asr-leaderboard
- generated_from_trainer
datasets:
- British_english
metrics:
- wer
model-index:
- name: Whisper tiny Japanese
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: British English
type: British_english
args: 'config: default, split: test'
metrics:
- name: Wer
type: wer
value: 12.709168856313097
Whisper tiny Japanese
This model is a fine-tuned version of openai/whisper-tiny on the British English dataset. It achieves the following results on the evaluation set:
- Loss: 0.3704
- Wer: 12.7092
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: 2e-05
- train_batch_size: 2
- eval_batch_size: 1
- 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: 500
- training_steps: 4000
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.3652 | 0.9901 | 1000 | 0.3394 | 13.2347 |
| 0.1538 | 1.9802 | 2000 | 0.3556 | 13.6357 |
| 0.04 | 2.9703 | 3000 | 0.3733 | 13.0549 |
| 0.002 | 3.9604 | 4000 | 0.3704 | 12.7092 |
Framework versions
- Transformers 4.50.0.dev0
- Pytorch 2.5.1+cu124
- Datasets 3.3.2
- Tokenizers 0.21.0