Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Chen, Hongzheng, Wang, Yingheng, Cai, Yaohui, Hu, Hins, Li, Jiajie, Huang, Shirley, Deng, Chenhui, Liang, Rongjian, Kong, Shufeng, Ren, Haoxing, Samaranayake, Samitha, Gomes, Carla P., Zhang, Zhiru
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2506.07972
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912855266164736
author Chen, Hongzheng
Wang, Yingheng
Cai, Yaohui
Hu, Hins
Li, Jiajie
Huang, Shirley
Deng, Chenhui
Liang, Rongjian
Kong, Shufeng
Ren, Haoxing
Samaranayake, Samitha
Gomes, Carla P.
Zhang, Zhiru
author_facet Chen, Hongzheng
Wang, Yingheng
Cai, Yaohui
Hu, Hins
Li, Jiajie
Huang, Shirley
Deng, Chenhui
Liang, Rongjian
Kong, Shufeng
Ren, Haoxing
Samaranayake, Samitha
Gomes, Carla P.
Zhang, Zhiru
contents While Large Language Models (LLMs) have demonstrated significant advancements in reasoning and agent-based problem-solving, current evaluation methodologies fail to adequately assess their capabilities: existing benchmarks either rely on closed-ended questions prone to saturation and memorization, or subjective comparisons that lack consistency and rigor. In this work, we introduce HeuriGym, an agentic framework designed for evaluating heuristic algorithms generated by LLMs for combinatorial optimization problems, characterized by clearly defined objectives and expansive solution spaces. HeuriGym empowers LLMs to propose heuristics, receive evaluative feedback via code execution, and iteratively refine their solutions. We evaluate nine state-of-the-art models on nine problems across domains such as computer systems, logistics, and biology, exposing persistent limitations in tool use, planning, and adaptive reasoning. To quantify performance, we propose the Quality-Yield Index (QYI), a metric that captures both solution pass rate and quality. Even top models like GPT-o4-mini-high and Gemini-2.5-Pro attain QYI scores of only 0.6, well below the expert baseline of 1. Our open-source benchmark aims to guide the development of LLMs toward more effective and realistic problem-solving in scientific and engineering domains.
format Preprint
id arxiv_https___arxiv_org_abs_2506_07972
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle HeuriGym: An Agentic Benchmark for LLM-Crafted Heuristics in Combinatorial Optimization
Chen, Hongzheng
Wang, Yingheng
Cai, Yaohui
Hu, Hins
Li, Jiajie
Huang, Shirley
Deng, Chenhui
Liang, Rongjian
Kong, Shufeng
Ren, Haoxing
Samaranayake, Samitha
Gomes, Carla P.
Zhang, Zhiru
Machine Learning
Artificial Intelligence
Computation and Language
While Large Language Models (LLMs) have demonstrated significant advancements in reasoning and agent-based problem-solving, current evaluation methodologies fail to adequately assess their capabilities: existing benchmarks either rely on closed-ended questions prone to saturation and memorization, or subjective comparisons that lack consistency and rigor. In this work, we introduce HeuriGym, an agentic framework designed for evaluating heuristic algorithms generated by LLMs for combinatorial optimization problems, characterized by clearly defined objectives and expansive solution spaces. HeuriGym empowers LLMs to propose heuristics, receive evaluative feedback via code execution, and iteratively refine their solutions. We evaluate nine state-of-the-art models on nine problems across domains such as computer systems, logistics, and biology, exposing persistent limitations in tool use, planning, and adaptive reasoning. To quantify performance, we propose the Quality-Yield Index (QYI), a metric that captures both solution pass rate and quality. Even top models like GPT-o4-mini-high and Gemini-2.5-Pro attain QYI scores of only 0.6, well below the expert baseline of 1. Our open-source benchmark aims to guide the development of LLMs toward more effective and realistic problem-solving in scientific and engineering domains.
title HeuriGym: An Agentic Benchmark for LLM-Crafted Heuristics in Combinatorial Optimization
topic Machine Learning
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2506.07972