Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Natural Language Inference research benchmarks – Page 15 · SOTA2 Research

Natural Language Inference

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

182.36Training Time (hours)

3

Feb 26, 2026

63.41Training Time (hours)

3

Feb 26, 2026

-0.03Delta Accuracy

3

Mar 18, 2026

SNLI synthetic contrast (val)

Naive Finetuning

90.4Accuracy

3

Feb 26, 2026

SNLI original (val)

Baseline (Pre-trained)

89.4Accuracy

3

Feb 26, 2026

84.3NLI Accuracy

3

Feb 26, 2026

0.867QNLI

3

Feb 26, 2026

38.1Macro-F1

3

Feb 26, 2026

20 languages All transfers

38.1Accuracy

3

Feb 26, 2026

20 languages Different script

37.8Accuracy

3

Feb 26, 2026

Multi-Genre Natural Language Inference (MultiNLI) targeted BERT-base (test)

0.006AUC-DEL+

3

Feb 26, 2026

SNLI contrastive (test)

76Accuracy

3

Feb 26, 2026

CommitmentBank (CB)

46.43Accuracy

2

May 12, 2026

85.46MRL Score

2

Apr 28, 2026

81.06Accuracy

2

Apr 27, 2026

3E[D]

2

Apr 20, 2026

MI-based distillation

12.27ECE

2

Feb 26, 2026

ANLI (R1+R2+R3 Average)

fs-X-ICL (ChatGPT)

59.6ANLI Avg Accuracy

2

Feb 26, 2026

Contract NLI SCROLLS (val)

77.7Exact Match

2

Feb 26, 2026

AdversarialNLI r3

67.83Accuracy

2

Feb 26, 2026

2.89Relative CPU Speed

2

Feb 26, 2026

Composite Distance

3.5Performance Loss

1

Mar 24, 2026

XNLI 2.0 78 (test)

—

—Accuracy

0

Feb 26, 2026

XNLI Māori 2.0

—

—Accuracy

0

Feb 26, 2026

Page 15 of 15

Popular datasets

MNLI ANLI SNLI XNLI MultiNLI

Follow for updates

Join SOTA2 Contact