segformer-b0-finetuned-100by100PNG-50epochs-attempt2-removeNAN

This model is a fine-tuned version of nvidia/mit-b0 on the JCAI2000/100By100BranchPNG dataset. It achieves the following results on the evaluation set:

• Loss: 0.1268
• Mean Iou: 0.8754
• Mean Accuracy: 1.0
• Overall Accuracy: 1.0
• Accuracy Branch: 1.0
• Iou Branch: 0.8754

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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Branch	Iou Branch
0.4487	1.05	20	0.6365	0.8754	1.0	1.0	1.0	0.8754
0.466	2.11	40	0.5024	0.8754	1.0	1.0	1.0	0.8754
0.4431	3.16	60	0.4013	0.8754	1.0	1.0	1.0	0.8754
0.3967	4.21	80	0.3739	0.8754	1.0	1.0	1.0	0.8754
0.2476	5.26	100	0.3191	0.8754	1.0	1.0	1.0	0.8754
0.3577	6.32	120	0.3235	0.8754	1.0	1.0	1.0	0.8754
0.2501	7.37	140	0.2839	0.8754	1.0	1.0	1.0	0.8754
0.3382	8.42	160	0.2674	0.8754	1.0	1.0	1.0	0.8754
0.3191	9.47	180	0.2512	0.8754	1.0	1.0	1.0	0.8754
0.1632	10.53	200	0.2197	0.8754	1.0	1.0	1.0	0.8754
0.1888	11.58	220	0.2095	0.8754	1.0	1.0	1.0	0.8754
0.1443	12.63	240	0.1975	0.8754	1.0	1.0	1.0	0.8754
0.1348	13.68	260	0.1836	0.8754	1.0	1.0	1.0	0.8754
0.1772	14.74	280	0.1742	0.8754	1.0	1.0	1.0	0.8754
0.1524	15.79	300	0.1893	0.8754	1.0	1.0	1.0	0.8754
0.1135	16.84	320	0.1710	0.8754	1.0	1.0	1.0	0.8754
0.1676	17.89	340	0.1789	0.8754	1.0	1.0	1.0	0.8754
0.131	18.95	360	0.1604	0.8754	1.0	1.0	1.0	0.8754
0.1693	20.0	380	0.1531	0.8754	1.0	1.0	1.0	0.8754
0.1031	21.05	400	0.1572	0.8754	1.0	1.0	1.0	0.8754
0.1432	22.11	420	0.1571	0.8754	1.0	1.0	1.0	0.8754
0.1711	23.16	440	0.1542	0.8754	1.0	1.0	1.0	0.8754
0.1287	24.21	460	0.1469	0.8754	1.0	1.0	1.0	0.8754
0.1228	25.26	480	0.1493	0.8754	1.0	1.0	1.0	0.8754
0.1316	26.32	500	0.1568	0.8754	1.0	1.0	1.0	0.8754
0.0737	27.37	520	0.1455	0.8754	1.0	1.0	1.0	0.8754
0.0914	28.42	540	0.1454	0.8754	1.0	1.0	1.0	0.8754
0.1122	29.47	560	0.1467	0.8754	1.0	1.0	1.0	0.8754
0.1482	30.53	580	0.1500	0.8754	1.0	1.0	1.0	0.8754
0.1006	31.58	600	0.1351	0.8754	1.0	1.0	1.0	0.8754
0.1069	32.63	620	0.1513	0.8754	1.0	1.0	1.0	0.8754
0.0985	33.68	640	0.1417	0.8754	1.0	1.0	1.0	0.8754
0.0794	34.74	660	0.1364	0.8754	1.0	1.0	1.0	0.8754
0.1065	35.79	680	0.1343	0.8754	1.0	1.0	1.0	0.8754
0.0993	36.84	700	0.1346	0.8754	1.0	1.0	1.0	0.8754
0.0904	37.89	720	0.1430	0.8754	1.0	1.0	1.0	0.8754
0.1159	38.95	740	0.1342	0.8754	1.0	1.0	1.0	0.8754
0.1787	40.0	760	0.1343	0.8754	1.0	1.0	1.0	0.8754
0.0621	41.05	780	0.1363	0.8754	1.0	1.0	1.0	0.8754
0.0844	42.11	800	0.1301	0.8754	1.0	1.0	1.0	0.8754
0.0919	43.16	820	0.1318	0.8754	1.0	1.0	1.0	0.8754
0.0728	44.21	840	0.1348	0.8754	1.0	1.0	1.0	0.8754
0.1073	45.26	860	0.1391	0.8754	1.0	1.0	1.0	0.8754
0.0563	46.32	880	0.1310	0.8754	1.0	1.0	1.0	0.8754
0.0827	47.37	900	0.1303	0.8754	1.0	1.0	1.0	0.8754
0.0633	48.42	920	0.1304	0.8754	1.0	1.0	1.0	0.8754
0.1452	49.47	940	0.1268	0.8754	1.0	1.0	1.0	0.8754

Framework versions

• Transformers 4.33.0
• Pytorch 2.0.1+cu117
• Datasets 2.14.4
• Tokenizers 0.13.3