datdo2717/whisper-small-en-23-11

Whisper Small Ori vi

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: 0.4093
• Wer: 15.2829

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 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: 200
• training_steps: 1300
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4964	0.2222	100	0.4513	18.4622
0.4607	0.4444	200	0.4251	18.1479
0.4223	0.6667	300	0.4054	17.3732
0.4186	0.8889	400	0.3998	17.2562
0.2411	1.1111	500	0.3978	16.8616
0.2425	1.3333	600	0.3946	16.8396
0.2194	1.5556	700	0.3926	14.8882
0.238	1.7778	800	0.3905	16.5034
0.2323	2.0	900	0.3904	15.2755
0.1294	2.2222	1000	0.4076	15.0709
0.1139	2.4444	1100	0.4080	15.3706
0.1108	2.6667	1200	0.4102	15.1732
0.1197	2.8889	1300	0.4093	15.2829

Framework versions

• Transformers 4.46.3
• Pytorch 2.4.0
• Datasets 3.1.0
• Tokenizers 0.20.0