lohitava/whisper-tiny-bn

Whisper Tiny Bn - Lohitava Ghosh

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

• Loss: 0.2394
• Wer: 63.4318

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: Use OptimizerNames.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: 500
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3354	0.6365	1000	0.3555	77.6737
0.2373	1.2731	2000	0.2772	69.6010
0.2246	1.9096	3000	0.2479	65.2452
0.2028	2.5461	4000	0.2394	63.4318

Framework versions

• Transformers 4.51.3
• Pytorch 2.5.1+cu124
• Datasets 3.5.1
• Tokenizers 0.21.0