nvidia-segformer-b0-finetuned-ade-512-512-finetuned-ISIC17

This model is a fine-tuned version of nvidia/segformer-b0-finetuned-ade-512-512 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1948
• Mean Iou: 0.8064
• Mean Accuracy: 0.8726
• Overall Accuracy: 0.9381
• Per Category Iou: [0.6841604127643356, 0.9285439643646547]
• Per Category Accuracy: [0.7721651141608432, 0.9729809595315688]

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Per Category Iou	Per Category Accuracy
0.481	0.16	10	0.4235	0.6191	0.6970	0.8761	[0.3719409076673884, 0.8662862424406493]	[0.42270204900152314, 0.9713331864930521]
0.4147	0.32	20	0.3894	0.7067	0.8502	0.8853	[0.5464942438498753, 0.8668431573745645]	[0.7965579529885418, 0.9038859083170013]
0.356	0.48	30	0.3148	0.7467	0.8513	0.9107	[0.5963581593534901, 0.897077797385972]	[0.7603709174964982, 0.9422313184595918]
0.3039	0.63	40	0.3024	0.7620	0.8671	0.9162	[0.6211722830632663, 0.9028139512386881]	[0.7918407335685692, 0.9422883932404167]
0.2545	0.79	50	0.2849	0.7766	0.8898	0.9201	[0.6468577863419183, 0.9063792530493855]	[0.8432862096150755, 0.9362151542385662]
0.2635	0.95	60	0.2504	0.7828	0.8644	0.9279	[0.6487213857926865, 0.9168129696986418]	[0.7671470887645524, 0.9616549114054705]
0.2175	1.11	70	0.2497	0.7849	0.8682	0.9283	[0.6526705030304356, 0.9171225024239068]	[0.7762677096648272, 0.9602225755678137]
0.2025	1.27	80	0.2400	0.7840	0.8632	0.9288	[0.6501844204669202, 0.9178944798865282]	[0.7627291445016801, 0.9636411137781736]
0.2035	1.43	90	0.2288	0.7931	0.8749	0.9313	[0.6657367286733036, 0.9203778068784213]	[0.7885027822639286, 0.9612655167036179]
0.2488	1.59	100	0.2110	0.7978	0.8719	0.9341	[0.6717638717220313, 0.923859975121704]	[0.7766611302038285, 0.9672003292652145]
0.1954	1.75	110	0.2067	0.7962	0.8597	0.9354	[0.666599427783381, 0.9258672754383861]	[0.7436428904928473, 0.9757231213956472]
0.1806	1.9	120	0.2047	0.7926	0.8525	0.9349	[0.6596059897565958, 0.925563006736469]	[0.726197674685608, 0.9787940661520825]
0.161	2.06	130	0.2047	0.7903	0.8505	0.9342	[0.6558737849234609, 0.9247714617107691]	[0.7223974159771602, 0.9786951901233297]
0.1736	2.22	140	0.2023	0.7948	0.8588	0.9349	[0.6643652721485811, 0.9252950591002775]	[0.742124317828686, 0.9754152391272543]
0.1947	2.38	150	0.2077	0.7985	0.8656	0.9355	[0.6712414223331253, 0.9257326708494226]	[0.7585178608332249, 0.9726888331181641]
0.1464	2.54	160	0.1960	0.8030	0.8680	0.9373	[0.678274892507806, 0.9276935390097538]	[0.7620104248788739, 0.9740685958478499]
0.1644	2.7	170	0.1964	0.8064	0.8751	0.9377	[0.6847175060674714, 0.9279857318627613]	[0.7791196258677832, 0.9710404169835255]
0.1803	2.86	180	0.1948	0.8064	0.8726	0.9381	[0.6841604127643356, 0.9285439643646547]	[0.7721651141608432, 0.9729809595315688]

Framework versions

• Transformers 4.24.0
• Pytorch 1.12.0+cu116
• Datasets 2.7.0
• Tokenizers 0.12.1