IbratDO/whisper-small-linkdata-7K

Whisper Small UZ - Link Data

This model is a fine-tuned version of openai/whisper-small on the UzbekVoice 7K Row dataset. It achieves the following results on the evaluation set:

• Loss: 0.5898
• Wer: 39.6424

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: Use OptimizerNames.ADAMW_TORCH_FUSED 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
0.2091	2.5381	1000	0.4622	44.1343
0.0232	5.0761	2000	0.4993	41.2124
0.004	7.6142	3000	0.5522	39.5988
0.001	10.1523	4000	0.5789	39.2499
0.0009	12.6904	5000	0.5898	39.6424

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu126
• Datasets 4.1.1
• Tokenizers 0.22.1