SmartWhatt/rangsitra-whisper-sm-selected-random-0.5

Rangsitra Whisper

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

• Cer: 7.0065
• Wer: 23.7324
• Loss: 0.2051

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: 16
• seed: 42
• 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: warmup_stable_decay
• lr_scheduler_warmup_steps: 1013
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Cer	Wer	Validation Loss
0.1644	1.0	2028	10.0526	33.6618	0.2230
0.0987	2.0	4056	8.9654	28.8740	0.1969
0.0627	3.0	6084	8.2169	27.0582	0.2021
0.0182	4.0	8112	7.2727	24.5310	0.1994
0.0029	5.0	10140	7.0065	23.7324	0.2051

Framework versions

• Transformers 4.52.4
• Pytorch 2.9.1+cu128
• Datasets 3.6.0
• Tokenizers 0.21.4