shreyasdesaisuperU/whisper-medium-fine-tuned-names-3000

Whisper Medium Fine Tuned 3000 Names SSD superU

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: 0.1246
• Wer: 26.1364

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: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4987	0.2392	100	0.4556	51.7045
0.4092	0.4785	200	0.3531	48.5795
0.3696	0.7177	300	0.3179	48.0114
0.2694	0.9569	400	0.2911	38.6364
0.2162	1.1962	500	0.2809	36.6477
0.2378	1.4354	600	0.2682	34.9432
0.2057	1.6746	700	0.1950	28.9773
0.1681	1.9139	800	0.2118	36.3636
0.1217	2.1531	900	0.1847	26.9886
0.1235	2.3923	1000	0.1722	25.5682
0.1203	2.6316	1100	0.1655	26.7045
0.1182	2.8708	1200	0.1704	28.9773
0.062	3.1100	1300	0.1566	26.9886
0.0835	3.3493	1400	0.1455	23.8636
0.0738	3.5885	1500	0.1387	24.1477
0.0849	3.8278	1600	0.1354	25.0
0.0419	4.0670	1700	0.1298	24.4318
0.0512	4.3062	1800	0.1302	26.4205
0.0524	4.5455	1900	0.1251	26.4205
0.0411	4.7847	2000	0.1246	26.1364

Framework versions

• Transformers 4.46.2
• Pytorch 2.2.2+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3