shripadbhat/whisper-small-pa-IN

Whisper Small Panjabi

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: 0.6084
• Wer: 36.1004

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: 64
• 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
0.349	5.86	100	0.4664	49.1929
0.0175	11.74	200	0.4633	39.1494
0.0052	17.63	300	0.5317	37.7146
0.0014	23.51	400	0.5521	36.4079
0.0009	29.4	500	0.5731	35.4599
0.0002	35.29	600	0.5806	35.6649
0.0001	41.17	700	0.5933	35.7161
0.0001	47.06	800	0.6016	35.9211
0.0001	52.91	900	0.6067	36.0492
0.0001	58.8	1000	0.6084	36.1004

Framework versions

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