AI Paper Atlas

Research map / Benchmarks / MultiNLI

MultiNLI leaderboard

20 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on MultiNLI. Best worst-group accuracy: 78.95 (ISR-Mean GroupDRO, ICML 2022).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

MultiNLI · worst-group accuracy

20 papers, each with its best reported number for its own method; higher is better.

#worst-group accuracy ↑MethodPaperTopic
178.95ISR-Mean GroupDRO
BERT
Provable Domain Generalization via Invariant-Feature Subspace Recovery
ICML 2022 · Table 1
Domain adaptation and generalization
278.9PG-DRO
BERT-base-uncased
Distributionally Robust Optimization with Probabilistic Group
AAAI 2023 · Table 3
Recognition and long-tail
378.6RCSV
BERT
Breaking Correlation Shift via Conditional Invariant Regularizer
ICLR 2023 · Table 2
OOD and anomaly detection
478.5GSR
BERT
Group-robust Sample Reweighting for Subpopulation Shifts via Influence Functions
ICLR 2025 · Performance comparison table
Recognition and long-tail
578gDRO-nWhat You See is What You Get: Principled Deep Learning via Distributional Generalization
NeurIPS 2022 · Table 1
Implicit bias and generalization
677.6difFOCI+ERM
ERM
A Differentiable Rank-Based Objective for Better Feature Learning
ICLR 2025 · Table: Worst group accuracy ac
Prediction and decision losses
776.8SCER
BERT
Spurious Correlation-Aware Embedding Regularization for Worst-Group Robustness
ICLR 2026 · Table 5
Recognition and long-tail
876.75SGD batch size 8
BERT
ON THE ROLE OF IMPLICIT REGULARIZATION OF STOCHASTIC GRADIENT DESCENT IN GROUP ROBUSTNESS
ICLR 2026 · Table 2
Implicit bias and generalization
976.6SSA
BERT
Spread Spurious Attribute: Improving Worst-group Accuracy with Spurious Attribute Estimation
ICLR 2022 · Table 2
Recognition and long-tail
1076.33LA/TTLSABeyond Invariance: Test-Time Label-Shift Adaptation for Addressing "Spurious" Correlations
NeurIPS 2023 · Table 2
Test-time adaptation
10 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.