whisper-tiny-uz

This model is a fine-tuned version of openai/whisper-tiny on the islomov/it_youtube_uzbek_speech_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.5242
• Wer Ortho: 53.9320
• Wer: 46.2731

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: 32
• eval_batch_size: 32
• 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: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
1.7493	0.9381	500	0.8690	86.7163	80.0547
1.2903	1.8762	1000	0.6905	75.2657	65.6485
1.0749	2.8143	1500	0.6177	60.1753	51.7438
0.9037	3.7523	2000	0.5750	56.9341	48.7121
0.8147	4.6904	2500	0.5497	56.7216	48.7121
0.7320	5.6285	3000	0.5359	58.5016	50.9688
0.6755	6.5666	3500	0.5275	56.1637	49.0084
0.5877	7.5047	4000	0.5242	53.9320	46.2731

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 5.0.0
• Tokenizers 0.22.2