List your company

Toxicity detection on CivilComments-WILDS (test)

92.7Average AccuracyERM
87.588.8590.291.55Apr 20, 2021Aug 24, 2021Dec 29, 2021May 5, 2022Sep 8, 2022Jan 13, 2023May 20, 2023

Evaluation results

MethodLinks
ERMGroup annotated in train?=false, Group annotated in val?=false2023.0592.7—51.6
ERMGroup annotated in train?=false, Group annotated in val?=true2023.0592.6—57.4
ERM (ours)Backbone=BERT2021.0492.558.1—
CVaR DROGroup annotated in train?=false, Group annotated in val?=true2023.0592.5—60.5
LfFGroup annotated in train?=false, Group annotated in val?=true2023.0592.5—58.8
ERMBackbone=BERT2021.0492.258—
JTTGroup annotated in train?=false, Group annotated in val?=false2023.0592.1—61.5
LfFGroup annotated in train?=false, Group annotated in val?=false2023.0592—55.9
CVaR DROGroup annotated in train?=false, Group annotated in val?=false2023.0591.9—56.5
FishBackbone=BERT2021.0491.474.2—
EIILGroup annotated in train?=false, Group annotated in val?=false2023.0591.2—63.8
JTTGroup annotated in train?=false, Group annotated in val?=true2023.0591.1—69.3
EIILGroup annotated in train?=false, Group annotated in val?=true2023.0590.5—67
Group DROBackbone=BERT2021.0489.470.4—
SSAGroup annotated in train?=false, Group annotated in val?=false2023.0589.1—69.5
ReweightedBackbone=BERT2021.0488.967.3—
Q-DiversityGroup annotated in train?=false, Group annotated in val?=false2023.0588.7—73.5
SSAGroup annotated in train?=false, Group annotated in val?=true2023.0588.2—69.9
Oracle DROGroup annotated in train?=true, Group annotated in val?=true2023.0587.7—69.1