Modelpublic

PostsDesert/segformer-b5-finetuned-segments-instryde-foot-test

sourceHugging Faceotherupdated 3y agoView on Hugging Face

0likes6downloads

Model Card

segformer-b5-finetuned-segments-instryde-foot-test

This model is a fine-tuned version of nvidia/mit-b5 on the inStryde/inStrydeSegmentationFoot dataset. It achieves the following results on the evaluation set:

—Loss: 0.0149
—Mean Iou: 0.4800
—Mean Accuracy: 0.9599
—Overall Accuracy: 0.9599
—Per Category Iou: [0.0, 0.9599216842864238]
—Per Category Accuracy: [nan, 0.9599216842864238]

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
—trainbatchsize: 8
—evalbatchsize: 8
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
0.1024	0.27	20	0.2085	0.4534	0.9067	0.9067	[0.0, 0.9067344993758137]	[nan, 0.9067344993758137]
0.0431	0.53	40	0.0487	0.4604	0.9207	0.9207	[0.0, 0.9207331455341442]	[nan, 0.9207331455341442]
0.0354	0.8	60	0.0319	0.4577	0.9155	0.9155	[0.0, 0.9154662028576415]	[nan, 0.9154662028576415]
0.0389	1.07	80	0.0276	0.4629	0.9257	0.9257	[0.0, 0.9257162800419576]	[nan, 0.9257162800419576]
0.0208	1.33	100	0.0244	0.4702	0.9404	0.9404	[0.0, 0.9403945317069335]	[nan, 0.9403945317069335]
0.0241	1.6	120	0.0212	0.4703	0.9406	0.9406	[0.0, 0.9406131407017349]	[nan, 0.9406131407017349]
0.0167	1.87	140	0.0208	0.4761	0.9521	0.9521	[0.0, 0.9521215619420916]	[nan, 0.9521215619420916]
0.0156	2.13	160	0.0205	0.4612	0.9224	0.9224	[0.0, 0.9224359945462809]	[nan, 0.9224359945462809]
0.0156	2.4	180	0.0208	0.4734	0.9468	0.9468	[0.0, 0.9467575875538612]	[nan, 0.9467575875538612]
0.0167	2.67	200	0.0182	0.4833	0.9667	0.9667	[0.0, 0.9666659635383208]	[nan, 0.9666659635383208]
0.0145	2.93	220	0.0243	0.4351	0.8702	0.8702	[0.0, 0.8702122233110058]	[nan, 0.8702122233110058]
0.0114	3.2	240	0.0176	0.4686	0.9373	0.9373	[0.0, 0.93726765603217]	[nan, 0.93726765603217]
0.0155	3.47	260	0.0161	0.4770	0.9541	0.9541	[0.0, 0.9540767701096305]	[nan, 0.9540767701096305]
0.0158	3.73	280	0.0169	0.4684	0.9368	0.9368	[0.0, 0.9368239181251786]	[nan, 0.9368239181251786]
0.0114	4.0	300	0.0162	0.4777	0.9554	0.9554	[0.0, 0.9554348305492647]	[nan, 0.9554348305492647]
0.0112	4.27	320	0.0159	0.4839	0.9678	0.9678	[0.0, 0.9677532556440432]	[nan, 0.9677532556440432]
0.0131	4.53	340	0.0154	0.4811	0.9622	0.9622	[0.0, 0.9622032718479555]	[nan, 0.9622032718479555]
0.0101	4.8	360	0.0156	0.4683	0.9367	0.9367	[0.0, 0.9366846987126999]	[nan, 0.9366846987126999]
0.0102	5.07	380	0.0152	0.4758	0.9517	0.9517	[0.0, 0.9516509773164403]	[nan, 0.9516509773164403]
0.0101	5.33	400	0.0169	0.4884	0.9768	0.9768	[0.0, 0.9768393358121804]	[nan, 0.9768393358121804]
0.0082	5.6	420	0.0150	0.4761	0.9522	0.9522	[0.0, 0.9522462074215836]	[nan, 0.9522462074215836]
0.01	5.87	440	0.0152	0.4788	0.9576	0.9576	[0.0, 0.9575745140264517]	[nan, 0.9575745140264517]
0.0098	6.13	460	0.0148	0.4783	0.9565	0.9565	[0.0, 0.9565489693736469]	[nan, 0.9565489693736469]
0.0088	6.4	480	0.0153	0.4795	0.9591	0.9591	[0.0, 0.959051850601846]	[nan, 0.959051850601846]
0.0091	6.67	500	0.0152	0.4828	0.9656	0.9656	[0.0, 0.965590177169167]	[nan, 0.965590177169167]
0.0102	6.93	520	0.0149	0.4800	0.9599	0.9599	[0.0, 0.9599216842864238]	[nan, 0.9599216842864238]

Framework versions

—Transformers 4.37.2
—Pytorch 2.0.1
—Datasets 2.16.1
—Tokenizers 0.15.1