volkan-aslan/whisper-tiny-tr

Whisper Tiny TR V2 - Fast - Volkan ASLAN

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

• Loss: 0.4938
• Wer: 39.8229

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.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
0.4471	0.6894	1000	0.5801	46.5537
0.2842	1.3785	2000	0.5294	42.3316
0.1894	2.0676	3000	0.5052	41.5703
0.1838	2.7570	4000	0.4918	40.5484
0.1404	3.4461	5000	0.4966	40.4803
0.115	4.1351	6000	0.4938	39.8229

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1