parambharat/whisper-small-ml

Whisper Small ML - Bharat Ramanathan

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

• Loss: 0.2308
• Wer: 36.7397

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: 64
• 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: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1275	4.03	500	0.1630	35.4015
0.09	9.02	1000	0.1821	40.0243
0.062	14.01	1500	0.2004	37.7129
0.0441	19.0	2000	0.2105	36.2530
0.0335	23.03	2500	0.2250	37.7129
0.0276	28.02	3000	0.2308	36.7397

Framework versions

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