List your company

Natural Language Inference on MultiNLI (test) (Matched & Mismatched)

88.05Average Worst-Group AccuracyNaive + P
63.547669.908876.2782.6312Oct 4, 2024

Evaluation results

MethodLinks
Naive + PContext size=642024.1088.05——————0.38
Naive + PContext size=2562024.1087.99——————0.2
Proposed + G + PContext size=642024.1087.97——————0.3
Proposed + PContext size=642024.1087.86——————0.62
Proposed + G + PContext size=162024.1087.81——————0.56
Proposed + G + PContext size=2562024.1087.78——————0.31
Proposed + PContext size=162024.1087.75——————0.24
Proposed + G + PContext size=322024.1087.74——————0.5
Proposed + PContext size=5122024.1087.73——————0.3
Proposed + PContext size=322024.1087.7——————0.43
Proposed + GContext size=2562024.1087.69——————0.51
Proposed + G + PContext size=5122024.1087.68——————0.12
Proposed + G + PContext size=82024.1087.57——————0.34
Naive + PContext size=162024.1087.56——————0.77
Proposed + PContext size=2562024.1087.56——————0.46
Proposed + G + PContext size=1282024.1087.56——————0.27
ERMContext size=5122024.1087.55——————0.41
Naive + PContext size=5122024.1087.54——————0.34
ProposedContext size=162024.1087.54——————0.42
Naive + PContext size=322024.1087.5——————0.57
Proposed + GContext size=5122024.1087.5——————0.55
ProposedContext size=82024.1087.45——————0.58
Naive + PContext size=82024.1087.43——————1.05
Proposed + PContext size=82024.1087.43——————0.48
Proposed + GContext size=162024.1087.37——————0.35
Naive + PContext size=1282024.1087.35——————0.27
ProposedContext size=2562024.1087.35——————0.19
Proposed + G + PContext size=42024.1087.32——————0.76
Proposed + PContext size=1282024.1087.31——————0.53
ERMContext size=642024.1087.3——————0.78
Naive + PContext size=42024.1087.29——————0.95
Proposed + GContext size=82024.1087.23——————0.42
GroupDROContext size=5122024.1087.23——————0.49
Proposed + PContext size=42024.1087.22——————0.55
GroupDROContext size=642024.1087.16——————0.55
Proposed + GContext size=22024.1087.15——————0.82
GroupDROContext size=1282024.1087.14——————0.37
ERMContext size=2562024.1087.1——————0.48
GroupDROContext size=2562024.1087.1——————0.14
ERMContext size=1282024.1087.04——————0.55
Proposed + GContext size=322024.1087.02——————0.27
ProposedContext size=322024.1086.99——————0.26
ProposedContext size=5122024.1086.89——————0.3
ERMContext size=162024.1086.82——————0.54
ERMContext size=322024.1086.76——————0.54
GroupDROContext size=162024.1086.75——————0.47
ProposedContext size=22024.1086.62——————0.65
GroupDROContext size=322024.1086.55——————0.3
1-NNContext size=5122024.1086.53——————0.29
ProposedContext size=642024.1086.51——————0.27
Proposed + GContext size=642024.1086.51——————0.58
Proposed + GContext size=1282024.1086.47——————0.49
1-NNContext size=1282024.1086.44——————0.81
1-NNContext size=162024.1086.39——————0.76
Proposed + GContext size=42024.1086.23——————0.65
ProposedContext size=1282024.1086.11——————0.29
1-NNContext size=642024.1086.04——————0.93
ProposedContext size=42024.1086.02——————0.33
1-NNContext size=322024.1086.01——————0.4
1-NNContext size=2562024.1085.9——————0.43
Proposed + PContext size=22024.1085.84——————1.12
1-NNContext size=82024.1085.84——————0.84
GroupDROContext size=82024.1085.45——————1.1
Proposed + G + PContext size=22024.1085.41——————1.38
Naive + PContext size=22024.1085.3——————1.58
NaiveContext size=642024.1085.2——————0.24
ERMContext size=82024.1085.1——————1.13
NaiveContext size=82024.1084.74——————0.37
NaiveContext size=42024.1084.58——————0.37
NaiveContext size=1282024.1084.38——————0.32
NaiveContext size=2562024.1084.3——————0.33
NaiveContext size=22024.1084.15——————0.3
NaiveContext size=162024.1084.13——————0.45
NaiveContext size=5122024.1084.05——————0.39
NaiveContext size=322024.1083.92——————0.25
1-NNContext size=42024.1082.41——————0.81
ERMContext size=42024.1081.71——————0.58
GroupDROContext size=42024.1081.56——————0.75
GroupDROContext size=22024.1076.54——————0.42
ERMContext size=22024.1076.4——————0.65
1-NNContext size=22024.1064.49——————0.9
AugmentationMitigation Approach=Augmentation, Base Model=ELECTRA-small2026.04———81.78————
AugmentationMitigation Approach=Augmentation, Base Model=ELECTRA-small2026.04———81————
BERT2018.04—86.785.9—————
CAFEEmbedding Dimension=300D, Model Architecture Class=Integrated neural network2019.07—78.777.9—————
CAFEEmbedding Dimension=300D, Model Architecture Class=Ensemble2019.07—80.279—————
CB last-layer retrainingAnnotations Group=false, Annotations Class=true2023.09—————64.7——
Class-balanced ERMAnnotations Group=false, Annotations Class=false2023.09—————69.2——
CRTGroup Info (Train/Val)=✗/✗2026.02—————65.480.2—
CVaR DROGroup annotated in train?=false, Group annotated in val?=true2023.05———8268———
CVaR DROGroup annotated in train?=false, Group annotated in val?=false2023.05———81.861.8———
CVaRDROGroup Info (Train/Val)=✗/✗2026.02—————48.275.4—
DFRAnnotations Group=true, Annotations Class=true2023.09—————72.6——
DFRGroup Info (Train/Val)=✗/✗2026.02—————63.880.2—
DFR2026.04———82.4—57.1——
DFR (our impl.)Annotations Group=true, Annotations Class=true2023.09—————70.8——
DIIN2018.04—78.877.8—————
DIINModel Architecture Class=Integrated neural network2019.07—78.877.8—————
DIINModel Architecture Class=Ensemble2019.07—8078.7—————
DMANModel Architecture Class=Integrated neural network2019.07—78.978.2—————