Modelpublic

1Developer/cali-whisper-tiny.en-drop-004-production

sourceHugging Faceapache-2.0updated 2mo agoView on Hugging Face

0likes20downloads

Model Card

whisper-tiny-calista-basics-v10

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

—Loss: 0.0001
—Wer Ortho: 0.0
—Wer: 0.0

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
—trainbatchsize: 16
—evalbatchsize: 16
—seed: 42
—optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
—lrschedulertype: constantwithwarmup
—lrschedulerwarmup_steps: 50
—training_steps: 500
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
No log	0.3509	20	4.3009	87.1212	38.6364
5.0821	0.7018	40	0.2389	10.6061	9.8485
0.9579	1.0526	60	0.0018	0.0	0.0
0.0159	1.4035	80	0.0007	0.0	0.0
0.0006	1.7544	100	0.0005	0.0	0.0
0.0006	2.1053	120	0.0003	0.0	0.0
0.0001	2.4561	140	0.0002	0.0	0.0
0.0001	2.8070	160	0.0002	0.0	0.0
0.0081	3.1579	180	0.0004	0.0	0.0
0.0002	3.5088	200	0.0002	0.0	0.0
0.0002	3.8596	220	0.0003	0.0	0.0
0.0039	4.2105	240	0.0002	0.0	0.0
0.0008	4.5614	260	0.0001	0.0	0.0
0.0000	4.9123	280	0.0001	0.0	0.0
0.0000	5.2632	300	0.0001	0.0	0.0
0.0000	5.6140	320	0.0001	0.0	0.0
0.0000	5.9649	340	0.0030	0.0	0.0
0.0021	6.3158	360	0.0001	0.0	0.0
0.0000	6.6667	380	0.0001	0.0	0.0
0.0049	7.0175	400	0.0001	0.0	0.0
0.0049	7.3684	420	0.0002	0.0	0.0
0.0005	7.7193	440	0.0002	0.0	0.0
0.0000	8.0702	460	0.0001	0.0	0.0
0.0000	8.4211	480	0.0001	0.0	0.0
0.0000	8.7719	500	0.0001	0.0	0.0

Framework versions

—Transformers 5.12.1
—Pytorch 2.11.0+cu128
—Datasets 5.0.0
—Tokenizers 0.22.2