AI Paper Atlas

Research map / Large language models / Agents, code and math

Benchmarking at the Edge of Comprehension

Samuele Marro, Jialin Yu, Emanuele La Malfa, Oishi Deb, Jiawei Li, Yibo Yang, Ebey Abraham, Sunando Sengupta, Eric Sommerlade, Michael J. Wooldridge, Philip Torr

ICML 2026 · Large language models › Agents, code and math · approach: Mathematical and structural reasoning benchmarks · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use critique-resilient answerer Elo as a model scoreCritique-Resilient Benchmarking · AIME 2025, BRUMO 2025, HMMT Feb 2025Spearman rho: 0.851, 0.830, 0.819 correlation with external benchmarksImproved
fit Bradley-Terry scores versus a base-rate predictor on held-out questionsitemized bipartite Bradley-Terry model · 5-fold validationaccuracy: 0.646 -> 0.922Improved
replace human adjudicators with weak models (GPT-3.5, GPT-4o) (ablation)adjudication stage · agreement with human verdictsSpearman rho of rankings: rho in [0.98, 1], tau in [0.93, 1]No clear effect
vary prompt design (three-way prompt ablation) (ablation)answerer rankings · prompt sensitivityKendall W: Kendall W = 0.952No clear effect
stronger models as pair members (frontier vs GPT-3.5 pairs) on mutual incompatibilitynon-proven-wrong answers · answer-pair compatibilityincompatible pairs (%): 14-18% (GPT-3.5 pairs) -> 0-0.4% (frontier pairs)Improved
compare critique agreement to answer correctness across modelscritics · GPT-5.4-high as ground truth proxyagreement rate: critique agreement 0.88-0.96 vs answer correctness 0.16-0.99No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy