ResearchBenchmarksClassification on ImageNet Blended-TextClassification on ImageNet Blended-TextFollow73.7Clean AccuracyCLIP64.8667.15569.4571.745Dec 1, 2024Evaluation resultsMethodMethodLinksClean AccuracyASRCLIPBackbone=ViT-L/14, Res...Backbone=ViT-L/14, Resolution=336x3362024.1273.799.3CleanCLIPBackbone=ViT-L/14, Res...Backbone=ViT-L/14, Resolution=336x3362024.1268.983.4PARBackbone=ViT-L/14, Res...Backbone=ViT-L/14, Resolution=336x3362024.1265.217