List your company
ResearchFormerly WizWand

Image Classification on ImageNet-Rendition (IN-R) (val)

92.5Top-1 AccEVA CLIP-g
23.038441.071759.10577.1383Oct 27, 2021Dec 29, 2021Mar 3, 2022May 6, 2022Jul 9, 2022Sep 11, 2022Nov 14, 2022
Updated 8mo ago

Evaluation results

MethodLinks
EVA CLIP-gzero-shot evaluation protocol=True2022.1192.5
ALIGNzero-shot evaluation protocol=True2022.1192.2
Open CLIP-Hzero-shot evaluation protocol=True2022.1189.3
Open CLIP-gzero-shot evaluation protocol=True2022.1188.7
OpenAI CLIP-Lzero-shot evaluation protocol=True2022.1187.8
MAEModel=ViT-H, Resolution=448, Training Objective=MAE pre-training2021.1166.5
MAEModel=ViT-H, Resolution=224, Training Objective=MAE pre-training2021.1164.4
MAEModel=ViT-L, Resolution=224, Training Objective=MAE pre-training2021.1159.9
Supervised BaselineModel=ViT-L, Resolution=224, Training Objective=Supervised2021.1150.9
Supervised BaselineModel=ViT-H, Resolution=224, Training Objective=Supervised2021.1150.3
Supervised BaselineModel=ViT-B, Resolution=224, Training Objective=Supervised2021.1149.4
Noisy StudentNotes=Previous Best2021.1148.7
MAEModel=ViT-B, Resolution=224, Training Objective=MAE pre-training2021.1148.3
Patch-base NSalpha=2, base_method=MoCo-v22021.1028.95
MoCo-v22021.1027.81
MoCHibase_method=MoCo-v22021.1025.71