whisper-small-npsc

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: 0.2028
• Wer: 12.9254

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 6000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3922	0.18	500	0.3975	24.2055
0.2893	0.36	1000	0.3139	20.1507
0.2471	0.54	1500	0.2733	17.4449
0.2159	0.72	2000	0.2488	16.2681
0.2195	0.89	2500	0.2304	15.0577
0.1178	1.07	3000	0.2245	14.5968
0.1099	1.25	3500	0.2183	14.1118
0.1059	1.43	4000	0.2136	13.7914
0.1156	1.61	4500	0.2072	13.7491
0.1025	1.79	5000	0.2034	13.1515
0.1123	1.97	5500	0.2006	13.0284
0.0734	2.15	6000	0.2028	12.9254

Framework versions

• Transformers 4.25.0.dev0
• Pytorch 1.12.1+cu113
• Datasets 2.6.1
• Tokenizers 0.13.1