Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Peisong, Liu, Bowen, Li, Zehua, Wang, Yuyao, Ma, Zhiwei, Li, Yuhan, Li, Jia
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.24693
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910252288442368
author Wang, Peisong
Liu, Bowen
Li, Zehua
Wang, Yuyao
Ma, Zhiwei
Li, Yuhan
Li, Jia
author_facet Wang, Peisong
Liu, Bowen
Li, Zehua
Wang, Yuyao
Ma, Zhiwei
Li, Yuhan
Li, Jia
contents Large language models still struggle with contest-level programming, while many agentic remedies rely on massive inference-time sampling or expensive multi-stage post-training. We study when execution feedback reliably helps an LLM CP solver and which mechanisms govern the gains. We model feedback-driven solving as a calibrated stopped process and identify three quantities: false-admission risk, program-level evidence against bad programs, and the active-state success hazard. Under held-out trace calibration and selection from a pre-declared finite controller manifest, the resulting structural certificate lower-bounds the clean success probability before false admission. We instantiate mechanisms targeting these quantities as Dual-Granularity Verification, Test Augmentation, and Experience-Driven Self-Evolving, yielding CP-Agent. Without updating any parameters, CP-Agent raises Pass@1 from 25.8\% to 48.5\% on LiveCodeBench Pro and improves Refine@5 by 11.0\% on ICPC-Eval. Across three LLM backbones, CP-Agent lies on the cost--accuracy efficiency frontier, and ablations show that each component primarily affects its corresponding certificate quantity.
format Preprint
id arxiv_https___arxiv_org_abs_2605_24693
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle CP-Agent: A Calibrated Risk-Controlled Agent for Feedback-Driven Competitive Programming
Wang, Peisong
Liu, Bowen
Li, Zehua
Wang, Yuyao
Ma, Zhiwei
Li, Yuhan
Li, Jia
Computation and Language
Large language models still struggle with contest-level programming, while many agentic remedies rely on massive inference-time sampling or expensive multi-stage post-training. We study when execution feedback reliably helps an LLM CP solver and which mechanisms govern the gains. We model feedback-driven solving as a calibrated stopped process and identify three quantities: false-admission risk, program-level evidence against bad programs, and the active-state success hazard. Under held-out trace calibration and selection from a pre-declared finite controller manifest, the resulting structural certificate lower-bounds the clean success probability before false admission. We instantiate mechanisms targeting these quantities as Dual-Granularity Verification, Test Augmentation, and Experience-Driven Self-Evolving, yielding CP-Agent. Without updating any parameters, CP-Agent raises Pass@1 from 25.8\% to 48.5\% on LiveCodeBench Pro and improves Refine@5 by 11.0\% on ICPC-Eval. Across three LLM backbones, CP-Agent lies on the cost--accuracy efficiency frontier, and ablations show that each component primarily affects its corresponding certificate quantity.
title CP-Agent: A Calibrated Risk-Controlled Agent for Feedback-Driven Competitive Programming
topic Computation and Language
url https://arxiv.org/abs/2605.24693