bagasshw/asr_sunda_result_full

Whisper Tiny Sunda

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

• Loss: 0.4974
• Wer: 0.5419

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: 64
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 128
• 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: 30
• training_steps: 150
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.8761	0.0219	30	1.2365	0.7810
0.9096	0.0438	60	0.7216	0.5673
0.6491	0.0657	90	0.5795	0.5316
0.5444	0.0876	120	0.5178	0.5609
0.4887	0.1095	150	0.4975	0.5418

Framework versions

• Transformers 4.50.0.dev0
• Pytorch 2.6.0+cu126
• Datasets 3.3.2
• Tokenizers 0.21.0