Yorkinjon/whisper-small-uzbek-ynv2

Whisper Small uz - Yorkerdev

This model is a fine-tuned version of openai/whisper-small on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3621
• Wer: 34.6135

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: 2
• seed: 42
• gradient_accumulation_steps: 4
• 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: 6000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6038	0.2640	1000	0.5630	48.7719
0.4917	0.5279	2000	0.4511	40.7366
0.4377	0.7919	3000	0.4073	37.9496
0.3151	1.0557	4000	0.3867	38.4776
0.2944	1.3197	5000	0.3679	35.5133
0.275	1.5836	6000	0.3621	34.6135

Framework versions

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