metadata
library_name: transformers
language:
- uz
license: apache-2.0
base_model: jamshidahmadov/whisper-uz-v2
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: Whisper base uz V3 - Jamshid Ahmadov
results: []
Whisper base uz V3 - Jamshid Ahmadov
This model is a fine-tuned version of jamshidahmadov/whisper-uz-v2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1568
- Wer: 16.6718
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: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Use 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: 2500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.2166 | 0.5333 | 500 | 0.2338 | 22.8910 |
| 0.1257 | 1.0661 | 1000 | 0.1892 | 18.7192 |
| 0.1153 | 1.5995 | 1500 | 0.1673 | 17.1644 |
| 0.0672 | 2.1323 | 2000 | 0.1603 | 17.1644 |
| 0.0663 | 2.6656 | 2500 | 0.1568 | 16.6718 |
Framework versions
- Transformers 4.47.0
- Pytorch 2.4.0
- Datasets 3.2.0
- Tokenizers 0.21.0