dynann/whisper-small-khm-asr-cultural-v2

dynann/whisper-small-kh-final

This model is a fine-tuned version of openai/whisper-small on the KHM ASR Cultural dataset. It achieves the following results on the evaluation set:

• Loss: 0.3025
• Wer: 102.0349

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• 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: 50
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.3845	7.6923	500	0.4710	103.3430
0.2341	15.3846	1000	0.3025	102.0349

Framework versions

• Transformers 5.7.0
• Pytorch 2.11.0+cu128
• Datasets 4.8.5
• Tokenizers 0.22.2