whisper-small-burmese-v2

This model is a fine-tuned version of myatsu/whisper-small-burmese on the Myanmar Speech Dataset (OpenSLR-80) dataset. It achieves the following results on the evaluation set:

• Loss: 0.1623
• Wer: 67.3642
• Cer: 24.9554

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: 5e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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: 100
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.2344	1.4	200	0.2328	81.5227	33.3180
0.1545	2.8	400	0.1866	73.7756	28.1360
0.0954	4.1965	600	0.1688	70.5699	26.8591
0.0747	5.5965	800	0.1629	68.8780	25.2430
0.0654	6.9965	1000	0.1623	67.3642	24.9554

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.4.2
• Tokenizers 0.22.1