Whisper-Small-Inbrowser-Proctor

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

• Loss: 0.3093
• Wer: 16.9481

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-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use 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: 20
• training_steps: 250
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2855	0.4545	25	0.4320	24.4186
0.1728	0.9091	50	0.3271	17.4896
0.0925	1.3636	75	0.3101	14.5428
0.1021	1.8182	100	0.3059	16.8366
0.054	2.2727	125	0.3039	15.1641
0.083	2.7273	150	0.3050	14.6703
0.0355	3.1818	175	0.3055	14.7818
0.0502	3.6364	200	0.3074	15.6897
0.0287	4.0909	225	0.3089	17.0596
0.0347	4.5455	250	0.3093	16.9481

Framework versions

• Transformers 4.48.1
• Pytorch 2.2.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0