List your company
ResearchFormerly WizWand

Accuracy Prediction on NAS-Bench-101 (depth=2~7, all samples)

0.891Kendall's TauNAR-Former
0.647640.710820.7740.83718Nov 15, 2022
Updated 8mo ago

Evaluation results

MethodLinks
NAR-FormerBackbone=Transformer, Training Samples=5% (21180)2022.110.891
NAR-FormerBackbone=Transformer, Training Samples=1% (4236)2022.110.871
TNASPBackbone=Transformer, Training Samples=1% (4236)2022.110.82
TNASP + SEBackbone=Transformer, Training Samples=1% (4236)2022.110.82
ReNASBackbone=CNN, Training Samples=1% (4236)2022.110.816
NAO+SEBackbone=LSTM, Training Samples=1% (4236)2022.110.787
NAOBackbone=LSTM, Training Samples=1% (4236)2022.110.775
NP + SEBackbone=GCN, Training Samples=1% (4236)2022.110.773
NPBackbone=GCN, Training Samples=1% (4236)2022.110.769
NAR-FormerBackbone=Transformer, Training Samples=0.1% (424)2022.110.765
TNASP + SEBackbone=Transformer, Training Samples=0.1% (424)2022.110.722
TNASPBackbone=Transformer, Training Samples=0.1% (424)2022.110.705
NP + SEBackbone=GCN, Training Samples=0.1% (424)2022.110.684
NAO+SEBackbone=LSTM, Training Samples=0.1% (424)2022.110.68
NPBackbone=GCN, Training Samples=0.1% (424)2022.110.679
NAOBackbone=LSTM, Training Samples=0.1% (424)2022.110.666
ReNASBackbone=CNN, Training Samples=0.1% (424)2022.110.657