sridhar1ga/whisper-small-en

Whisper Small En - Sridhar Vanga

This model is a fine-tuned version of openai/whisper-small on the sridhar1ga/UCLASS dataset. It achieves the following results on the evaluation set:

• Loss: 2.0974
• Wer: 41.4951

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: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0513	18.5185	500	1.5771	61.3218
0.0056	37.0370	1000	1.8013	42.1452
0.0089	55.5556	1500	1.8905	65.0054
0.0054	74.0741	2000	1.7860	44.9621
0.0016	92.5926	2500	1.9571	41.9285
0.0001	111.1111	3000	2.0281	41.0618
0.0001	129.6296	3500	2.0595	41.9285
0.0001	148.1481	4000	2.0805	41.3868
0.0001	166.6667	4500	2.0927	41.4951
0.0001	185.1852	5000	2.0974	41.4951

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1