AI Paper Atlas

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

Beyond Problem Solving: UOJ-Bench for Evaluating Code Generation, Hacking, and Repair in Competitive Programming

Tingqiang Xu, Hangrui Zhou, Tianle Cai, Alex Gu, Kaifeng Lyu

ICML 2026 · Large language models › Agents, code and math · approach: Code generation 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
sample many candidates (Pass@k) instead of oneGemini-3-pro-preview · Hacking (Hard) 100-sample subsetsuccess rate: 93 out of 100 hacks foundImproved
sample many candidates (Pass@k) instead of oneGemini-3-pro-preview · Repair (Hard) 50-sample subsetsuccess rate: 41 out of 50 repairedImproved
use agentic ReAct feedback for hacking instead of parallel samplingGPT-OSS-120B · Hacking (Hard)success rate over 10 turnsNo clear effect
use agentic iterative debugging for repair instead of direct samplingGPT-OSS-120B · Repair (Hard)success rate over 10 turnsImproved
hard (covert error) instead of easy (overt error) hackingGPT-5.2-high · UOJ-Bench hackingPass@1: 71.61 -> 58.60Hurt
hard instead of easy repairGPT-OSS-120B · UOJ-Bench repairPass@1: 20.20 -> 9.72Hurt
problem difficulty from Easy to UltrahardGPT-5.2-high · UOJ-Bench generationPass@1: 88.89 -> 27.33Hurt
evaluate on more recent problemsLLMs · Hard problems generationPass@1Hurt
evaluate on more recent problemsLLMs · Hard problems hacking and repairPass@1No clear effect
run GPT-OSS-120B hacking with test-time scaling on zero-day submissionsGPT-OSS-120B · Zero-Day-Hacking-5Kfraction of submissions with new errors: over 10%Improved

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