CheeseES/whisper_model

Whisper_FT_V1

This model is a fine-tuned version of openai/whisper-small on the LLM Fine Tuning Dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.0892
• Wer: 51.6190

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-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 33
• 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: 300
• training_steps: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5149	0.8547	100	0.2080	80.4524
0.2	1.7094	200	0.1883	82.9286
0.1807	2.5641	300	0.1701	84.7143
0.1561	3.4188	400	0.1553	82.6667
0.1363	4.2735	500	0.1458	75.3571
0.1152	5.1282	600	0.1367	71.3095
0.0994	5.9829	700	0.1284	68.6190
0.0865	6.8376	800	0.1214	64.5238
0.073	7.6923	900	0.1136	69.5714
0.0656	8.5470	1000	0.1091	66.6905
0.0598	9.4017	1100	0.1049	69.8810
0.0512	10.2564	1200	0.1025	65.0
0.0481	11.1111	1300	0.0977	64.8571
0.0429	11.9658	1400	0.0955	59.5238
0.0385	12.8205	1500	0.0930	61.3810
0.0338	13.6752	1600	0.0916	65.3810
0.0334	14.5299	1700	0.0905	63.0952
0.0298	15.3846	1800	0.0892	51.6190

Framework versions

• PEFT 0.15.2
• Transformers 4.51.3
• Pytorch 2.8.0.dev20250319+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1