rishabhjain16/whisper_medium_en_to_pf10h

openai/whisper-medium.en

This model is a fine-tuned version of openai/whisper-medium.en on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1748
• Wer: 2.7097

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: 32
• eval_batch_size: 32
• 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0329	5.0	500	0.1343	4.0125
0.0013	10.01	1000	0.1531	2.8810
0.0002	15.01	1500	0.1609	2.7321
0.0002	20.01	2000	0.1608	2.7544
0.0001	25.01	2500	0.1688	2.7321
0.0002	30.02	3000	0.1722	2.7172
0.0001	35.02	3500	0.1742	2.7172
0.0001	40.02	4000	0.1748	2.7097

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu117
• Datasets 2.9.1.dev0
• Tokenizers 0.13.2