shripadbhat/whisper-small-hi

Whisper Small Hindi - Shripad Bhat

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

• Loss: 0.3909
• Wer: 21.4519

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
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4337	0.73	100	0.4874	47.5868
0.1894	1.47	200	0.3264	23.9482
0.1007	2.21	300	0.3101	22.5267
0.0984	2.94	400	0.3064	21.5723
0.0555	3.67	500	0.3325	22.0251
0.029	4.41	600	0.3439	21.4863
0.0163	5.15	700	0.3668	21.6468
0.0153	5.88	800	0.3756	21.4662
0.0081	6.62	900	0.3888	21.5035
0.0059	7.35	1000	0.3909	21.4519

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2