Modelpublic

distill-lab/train_dit02_initials_3cat_cls_3.5k_naflex

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face

0likes5downloads

Model Card

traindit02initials3catcls3.5knaflex

This model is a fine-tuned version of google/siglip2-so400m-patch16-naflex on an unknown dataset. It achieves the following results on the evaluation set:

—Loss: 2.2541
—Accuracy: 0.6474
—Precision: 0.6451
—Recall: 0.6474
—F1: 0.6312
—Roc Auc: 0.7542

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: 5e-06
—trainbatchsize: 64
—evalbatchsize: 64
—seed: 1337
—optimizer: Use adamwtorch with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: cosine
—num_epochs: 20.0
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Roc Auc
0.1606	2.0833	100	0.2993	0.6530	0.6537	0.6530	0.6395	0.7772
0.0511	4.1667	200	0.6150	0.6101	0.6770	0.6101	0.6038	0.7704
0.0104	6.25	300	0.7966	0.6549	0.6360	0.6549	0.6394	0.7630
0.0114	8.3333	400	0.9993	0.6604	0.6599	0.6604	0.6489	0.7559
0.0016	10.4167	500	1.2927	0.6530	0.6350	0.6530	0.6267	0.7524
0.0002	12.5	600	1.4876	0.6287	0.6400	0.6287	0.6162	0.7600
0.0	14.5833	700	2.0675	0.6455	0.6336	0.6455	0.6271	0.7547
0.0	16.6667	800	2.1974	0.6493	0.6390	0.6493	0.6305	0.7562
0.0	18.75	900	2.2512	0.6493	0.6473	0.6493	0.6331	0.7541

Framework versions

—Transformers 4.56.2
—Pytorch 2.7.1+cu126
—Datasets 4.0.0
—Tokenizers 0.22.1