whisper-small-khmer

This model is a fine-tuned version of openai/whisper-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.2805
• Wer: 83.8261

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: 16
• eval_batch_size: 8
• 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: linear
• lr_scheduler_warmup_steps: 200
• training_steps: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4839	4.5455	200	1.3586	84.7826
0.0291	9.0909	400	1.7237	82.2609
0.0116	13.6364	600	1.8644	80.2609
0.0044	18.1818	800	2.0192	81.9130
0.0025	22.7273	1000	2.0997	80.6957
0.0007	27.2727	1200	2.2802	82.0870
0.0002	31.8182	1400	2.2096	79.8261
0.0001	36.3636	1600	2.2580	79.6522
0.0001	40.9091	1800	2.2754	83.7391
0.0001	45.4545	2000	2.2805	83.8261

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2