Yorkinjon/whisper-small-uzbek-ynv4

Whisper small uz v4 - YorkinjonDev

This model is a fine-tuned version of Yorkinjon/whisper-small-uzbek-ynv3 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0709
• Wer: 6.6564

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• 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: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0397	1.1429	1000	0.0785	8.5583
0.015	2.2857	2000	0.0697	7.3160
0.0069	3.4286	3000	0.0661	6.4264
0.0022	4.5714	4000	0.0700	6.9172
0.0011	5.7143	5000	0.0709	6.6564

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu118
• Datasets 3.3.2
• Tokenizers 0.21.0