segformer-b0-scene-parse-150-lr-4-e-15

This model is a fine-tuned version of DiTo97/binarization-segformer-b3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1773
• Mean Iou: 0.5116
• Mean Accuracy: 0.5539
• Overall Accuracy: 0.9486
• Per Category Iou: [0.07467818861526594, 0.9484318643687625]
• Per Category Accuracy: [0.13278359055139496, 0.9749314802690082]

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
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
No log	1.0	112	0.3321	0.4844	0.5000	0.9686	[5.913750483660308e-05, 0.968644931717587]	[5.9410243004868244e-05, 0.9998514102409571]
No log	2.0	224	0.1448	0.4844	0.5	0.9688	[0.0, 0.9687870873345269]	[0.0, 1.0]
No log	3.0	336	0.1467	0.4855	0.5011	0.9687	[0.0024028604839131528, 0.9686745745655791]	[0.002417148172540925, 0.9998084247604243]
No log	4.0	448	0.1597	0.4974	0.5136	0.9673	[0.02761431295696444, 0.9672534071470754]	[0.029766229180953417, 0.9974892869900998]
0.4196	5.0	560	0.1483	0.4945	0.5101	0.9683	[0.02072799899238894, 0.9682597471616551]	[0.021509902838791155, 0.9987846484301768]
0.4196	6.0	672	0.1300	0.4973	0.5131	0.9682	[0.026546808517533143, 0.9681413453315052]	[0.02781078346833604, 0.9984659761718246]
0.4196	7.0	784	0.1407	0.5063	0.5244	0.9659	[0.04665771796171021, 0.9658509666995633]	[0.05345563922026602, 0.995305832396877]
0.4196	8.0	896	0.1377	0.5014	0.5186	0.9662	[0.036728661127978124, 0.9661516368135028]	[0.041295211194926705, 0.995994201663374]
0.174	9.0	1008	0.1632	0.5096	0.5382	0.9570	[0.06234910880338227, 0.9568704542992275]	[0.09161908189107895, 0.984874907876537]
0.174	10.0	1120	0.1424	0.5102	0.5323	0.9627	[0.05773026579725805, 0.9625824124413115]	[0.07327829115771892, 0.9913228393342741]
0.174	11.0	1232	0.1553	0.5035	0.5223	0.9644	[0.04268206669259935, 0.9643468862627879]	[0.05084668083459509, 0.9938369430563793]
0.174	12.0	1344	0.1607	0.5086	0.5330	0.9600	[0.057171934641356385, 0.95994904570909]	[0.07762033120361757, 0.9884765551939039]
0.174	13.0	1456	0.1619	0.5095	0.5358	0.9589	[0.060308850859297915, 0.958769171435925]	[0.08455435528004292, 0.9870474246884537]
0.1457	14.0	1568	0.1625	0.5123	0.5476	0.9534	[0.07133326653200926, 0.9531840662639103]	[0.11479756384054969, 0.9803688154229716]
0.1457	15.0	1680	0.1773	0.5116	0.5539	0.9486	[0.07467818861526594, 0.9484318643687625]	[0.13278359055139496, 0.9749314802690082]

Framework versions

• Transformers 4.37.0
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0