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

Self-correction research benchmarks · SOTA2 Research

Self-correction

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

Llama-3.3-70B math n=30 (failure pool)

L_MEMORY (Source-Conditioned Role Relabeling)

86.7Correction Rate

5

Jun 5, 2026

Qwen-72B math failure pool n=30

L_MEMORY (Source-Conditioned Role Relabeling)

80Correction Rate

5

Jun 5, 2026

WikiText-2 and OpenWebText

82.2NLI Score

3

May 15, 2026

Popular datasets

Llama WikiText-2 and OpenWebText Qwen-72B math

Follow for updates

Join SOTA2 Contact