nurzhanit/whisper-omg-2.5.1

Whisper Small Hi - Sanchit Gandhi

This model is a fine-tuned version of nurzhanit/whisper-enhanced-ml on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0006
• Wer: 0.0162

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 25
• training_steps: 500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1023	0.9804	50	0.0441	8.3765
0.0402	1.9608	100	0.0186	3.5969
0.0209	2.9412	150	0.0087	2.8354
0.0064	3.9216	200	0.0046	0.6481
0.0044	4.9020	250	0.0020	0.2106
0.0018	5.8824	300	0.0013	0.1620
0.0009	6.8627	350	0.0009	0.1134
0.001	7.8431	400	0.0007	0.0648
0.0007	8.8235	450	0.0006	0.0162
0.0007	9.8039	500	0.0006	0.0162

Framework versions

• Transformers 4.40.0
• Pytorch 2.5.0+cu124
• Datasets 3.0.2
• Tokenizers 0.19.1