segformer-b0-finetuned-segments-stamp-verification

This model is a fine-tuned version of nvidia/mit-b0 on the AliShah07/stamp-verification dataset. It achieves the following results on the evaluation set:

• Loss: 0.0535
• Mean Iou: 0.1317
• Mean Accuracy: 0.2635
• Overall Accuracy: 0.2635
• Accuracy Unlabeled: nan
• Accuracy Stamp: 0.2635
• Iou Unlabeled: 0.0
• Iou Stamp: 0.2635

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: 6e-05
• train_batch_size: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Unlabeled	Accuracy Stamp	Iou Unlabeled	Iou Stamp
0.6502	0.8333	20	0.6958	0.4685	0.9370	0.9370	nan	0.9370	0.0	0.9370
0.4529	1.6667	40	0.5458	0.0754	0.1508	0.1508	nan	0.1508	0.0	0.1508
0.3716	2.5	60	0.3818	0.0021	0.0041	0.0041	nan	0.0041	0.0	0.0041
0.3238	3.3333	80	0.2932	0.0126	0.0252	0.0252	nan	0.0252	0.0	0.0252
0.2167	4.1667	100	0.2326	0.0008	0.0015	0.0015	nan	0.0015	0.0	0.0015
0.1948	5.0	120	0.2029	0.0033	0.0065	0.0065	nan	0.0065	0.0	0.0065
0.1643	5.8333	140	0.1609	0.0	0.0	0.0	nan	0.0	0.0	0.0
0.1642	6.6667	160	0.1428	0.0	0.0	0.0	nan	0.0	0.0	0.0
0.1326	7.5	180	0.1222	0.0001	0.0002	0.0002	nan	0.0002	0.0	0.0002
0.1012	8.3333	200	0.0981	0.0	0.0	0.0	nan	0.0	0.0	0.0
0.0981	9.1667	220	0.0972	0.0058	0.0117	0.0117	nan	0.0117	0.0	0.0117
0.0838	10.0	240	0.0781	0.0015	0.0031	0.0031	nan	0.0031	0.0	0.0031
0.0771	10.8333	260	0.0708	0.0060	0.0120	0.0120	nan	0.0120	0.0	0.0120
0.0743	11.6667	280	0.0696	0.0298	0.0596	0.0596	nan	0.0596	0.0	0.0596
0.0655	12.5	300	0.0630	0.0398	0.0795	0.0795	nan	0.0795	0.0	0.0795
0.0673	13.3333	320	0.0613	0.0856	0.1712	0.1712	nan	0.1712	0.0	0.1712
0.0573	14.1667	340	0.0538	0.0725	0.1450	0.1450	nan	0.1450	0.0	0.1450
0.0623	15.0	360	0.0543	0.1008	0.2016	0.2016	nan	0.2016	0.0	0.2016
0.0557	15.8333	380	0.0559	0.1474	0.2947	0.2947	nan	0.2947	0.0	0.2947
0.0594	16.6667	400	0.0492	0.1019	0.2039	0.2039	nan	0.2039	0.0	0.2039
0.056	17.5	420	0.0479	0.1235	0.2470	0.2470	nan	0.2470	0.0	0.2470
0.0499	18.3333	440	0.0481	0.1124	0.2248	0.2248	nan	0.2248	0.0	0.2248
0.0516	19.1667	460	0.0477	0.1465	0.2930	0.2930	nan	0.2930	0.0	0.2930
0.0517	20.0	480	0.0535	0.1317	0.2635	0.2635	nan	0.2635	0.0	0.2635

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1