Modelpublic

Hanhpt23/whisper-small-chinesemed-free_E0-8D3-11

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face

0likes5downloads

Model Card

openai/whisper-small

This model is a fine-tuned version of openai/whisper-small on the pphuc25/ChiMed dataset. It achieves the following results on the evaluation set:

—Loss: 0.9620
—Wer: 81.1395
—Cer: 23.7745

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: 0.0001
—trainbatchsize: 8
—evalbatchsize: 8
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 100
—num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.5356	1.0	161	0.7160	303.1434	129.8351
0.2895	2.0	322	0.7218	328.0943	132.3084
0.1755	3.0	483	0.8070	279.5678	115.5526
0.0867	4.0	644	0.8045	264.8330	112.0544
0.0545	5.0	805	0.8648	115.7171	36.7201
0.0331	6.0	966	0.9064	125.5403	47.9947
0.0266	7.0	1127	0.8990	119.8428	42.2237
0.0131	8.0	1288	0.8769	123.3792	46.3012
0.0103	9.0	1449	0.9286	83.1041	26.1809
0.0053	10.0	1610	0.9243	82.9077	24.2647
0.0102	11.0	1771	0.9249	84.0864	24.3761
0.0031	12.0	1932	0.9252	80.3536	23.4848
0.003	13.0	2093	0.9434	119.0570	45.2986
0.0011	14.0	2254	0.9481	156.5815	65.7754
0.0017	15.0	2415	0.9521	79.7642	23.9305
0.0004	16.0	2576	0.9535	80.5501	23.8191
0.0005	17.0	2737	0.9579	81.1395	23.9082
0.0004	18.0	2898	0.9603	81.1395	23.8636
0.0004	19.0	3059	0.9616	81.1395	23.7968
0.0004	20.0	3220	0.9620	81.1395	23.7745

Framework versions

—Transformers 4.41.1
—Pytorch 2.3.0
—Datasets 2.19.1
—Tokenizers 0.19.1