Modelpublic

NaSugu/Pathe-asr-RbData-fcb

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face

0likes4downloads

Model Card

Pathe-asr-RbData-fcb

This model is a fine-tuned version of openai/whisper-small on the audiofolder dataset. It achieves the following results on the evaluation set:

—Loss: 0.0005
—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: 2
—evalbatchsize: 2
—seed: 42
—gradientaccumulationsteps: 4
—totaltrainbatch_size: 8
—optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: constant
—lrschedulerwarmup_steps: 1000
—num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.4889	1.0	3	4.5425	100.0
2.3946	2.0	6	3.4017	100.0
0.9603	3.0	9	2.1974	100.0
0.2824	4.0	12	1.3106	100.0
0.0716	5.0	15	0.2906	100.0
0.0059	6.0	18	0.0091	0.0
0.0008	7.0	21	0.0015	0.0
0.0003	8.0	24	0.0009	0.0
0.0001	9.0	27	0.0016	0.0
0.0001	10.0	30	0.0030	0.0
0.0	11.0	33	0.0037	0.0
0.0	12.0	36	0.0042	0.0
0.0	13.0	39	0.0042	0.0
0.0	14.0	42	0.0039	0.0
0.0	15.0	45	0.0034	0.0
0.0	16.0	48	0.0029	0.0
0.0	17.0	51	0.0025	0.0
0.0	18.0	54	0.0022	0.0
0.0	19.0	57	0.0020	0.0
0.0	20.0	60	0.0016	0.0
0.0	21.0	63	0.0016	0.0
0.0	22.0	66	0.0014	0.0
0.0	23.0	69	0.0011	0.0
0.0	24.0	72	0.0012	0.0
0.0	25.0	75	0.0011	0.0
0.0	26.0	78	0.0010	0.0
0.0	27.0	81	0.0009	0.0
0.0	28.0	84	0.0009	0.0
0.0	29.0	87	0.0008	0.0
0.0	30.0	90	0.0008	0.0
0.0	31.0	93	0.0008	0.0
0.0	32.0	96	0.0008	0.0
0.0	33.0	99	0.0008	0.0
0.0	34.0	102	0.0007	0.0
0.0	35.0	105	0.0007	0.0
0.0	36.0	108	0.0007	0.0
0.0	37.0	111	0.0007	0.0
0.0	38.0	114	0.0006	0.0
0.0	39.0	117	0.0006	0.0
0.0	40.0	120	0.0007	0.0
0.0	41.0	123	0.0006	0.0
0.0	42.0	126	0.0006	0.0
0.0	43.0	129	0.0005	0.0
0.0	44.0	132	0.0006	0.0
0.0	45.0	135	0.0005	0.0
0.0	46.0	138	0.0005	0.0
0.0	47.0	141	0.0005	0.0
0.0	48.0	144	0.0005	0.0
0.0	49.0	147	0.0005	0.0
0.0	50.0	150	0.0005	0.0

Framework versions

—Transformers 4.49.0
—Pytorch 2.5.1+cu124
—Datasets 3.3.2
—Tokenizers 0.21.0