metadata
base_model: openai/whisper-small
datasets:
- fruk19/N_asr
language:
- th
license: apache-2.0
metrics:
- wer
tags:
- generated_from_trainer
model-index:
- name: North_asri
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: aicookcook
type: fruk19/N_asr
config: default
split: None
args: 'config: th'
metrics:
- type: wer
value: 5.772624833690841
name: Wer
North_asri
This model is a fine-tuned version of openai/whisper-small on the aicookcook dataset. It achieves the following results on the evaluation set:
- Loss: 0.0764
- Wer: 5.7726
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: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0486 | 2.0 | 6000 | 0.0722 | 9.8591 |
| 0.0125 | 4.0 | 12000 | 0.0682 | 6.9130 |
| 0.0038 | 6.0 | 18000 | 0.0722 | 6.3537 |
| 0.0019 | 8.0 | 24000 | 0.0752 | 5.9627 |
| 0.0001 | 10.0 | 30000 | 0.0764 | 5.7726 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.0.1+cu117
- Datasets 2.20.0
- Tokenizers 0.19.1