List your company
ResearchFormerly WizWand

Image Classification on Patching Task Suite Average

84.4Average AccuracyMultiple models
53.261.369.477.5Aug 10, 2022
Updated 8mo ago

Evaluation results

MethodLinks
Multiple modelsBackbone=ViT-L/142022.0884.4
Joint patchingBackbone=ViT-L/142022.0883.8
Seq. patchingBackbone=ViT-L/14, Seed=22022.0880.9
Seq. patchingBackbone=ViT-L/14, Seed=02022.0880.1
Seq. patchingBackbone=ViT-L/14, Seed=12022.0880.1
Multiple modelsBackbone=ViT-B/162022.0880
Joint patchingBackbone=ViT-B/162022.0878.2
Parallel patchingBackbone=ViT-L/142022.0877.2
Multiple modelsBackbone=ViT-B/322022.0876.1
Joint patchingBackbone=ViT-B/322022.0874.1
Seq. patchingBackbone=ViT-B/16, Seed=22022.0871.8
Seq. patchingBackbone=ViT-B/16, Seed=12022.0871.1
Seq. patchingBackbone=ViT-B/16, Seed=02022.0870.7
Parallel patchingBackbone=ViT-B/162022.0868.5
UnpatchedBackbone=ViT-L/142022.0868
Seq. patchingBackbone=ViT-B/32, Seed=22022.0866.4
Seq. patchingBackbone=ViT-B/32, Seed=12022.0865.6
Seq. patchingBackbone=ViT-B/32, Seed=02022.0865.4
Parallel patchingBackbone=ViT-B/322022.0862.5
UnpatchedBackbone=ViT-B/162022.0860.4
UnpatchedBackbone=ViT-B/322022.0854.4