metadata
license: apache-2.0
base_model: openai/whisper-small
tags:
- generated_from_trainer
datasets:
- verba_lex_voice
metrics:
- wer
model-index:
- name: verbalex-zh
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: verba_lex_voice
type: verba_lex_voice
config: zh
split: test
args: zh
metrics:
- name: Wer
type: wer
value: 4.537114261884904
verbalex-zh
This model is a fine-tuned version of openai/whisper-small on the verba_lex_voice dataset. It achieves the following results on the evaluation set:
- Loss: 0.1166
- Wer: 4.5371
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: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0031 | 5.0505 | 1000 | 0.1021 | 4.9124 |
| 0.0002 | 10.1010 | 2000 | 0.1103 | 4.6956 |
| 0.0001 | 15.1515 | 3000 | 0.1134 | 4.5705 |
| 0.0001 | 20.2020 | 4000 | 0.1158 | 4.5788 |
| 0.0001 | 25.2525 | 5000 | 0.1166 | 4.5371 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.1.2
- Datasets 2.16.0
- Tokenizers 0.19.1