Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

EXG: Self-Evolving Agents with Experience Graphs

About

Large language model (LLM)-based agents have demonstrated strong capabilities in complex reasoning and problem solving through multi-step interactions, yet most deployed agents remain behaviorally static, with knowledge acquired during execution rarely translating into systematic improvement over time. In response, a growing line of work on self-evolving agents explores how agents can improve through experience during deployment, but most existing approaches either rely on ad hoc reflection limited to single-task correction or adopt unstructured memory that accumulates fragmented experience with delayed usability. To address this limitation, we introduce EXG, an experience graph framework for self-evolving agents that explicitly organizes accumulated successes and failures into a structured, relational representation. EXG is the first experience graph designed for self-evolving agents, supporting both online, real-time graph growth during execution for immediate cross-task experience reuse, and offline reuse of a consolidated experience graph as an external memory module. This design also enables EXG to serve as a plug-and-play component for existing self-evolving agents, organizing prior experience into a unified experience graph and improving both solution quality and resource efficiency as deployment progresses. Extensive experiments across code generation and reasoning benchmarks show that EXG attains more favorable performance-efficiency trade-offs than reflection- and memory-based baselines in both online and offline evaluations. Our results suggest that structuring experience as a graph provides a principled foundation for scalable and transferable self-evolving agent behavior.

Yuxin Jin, Siyuan Zhang, Hanchen Wang, Lu Qin, Ying Zhang, Wenjie Zhang• 2026

Related benchmarks

TaskDatasetResultRank
Code GenerationEvalPlus
Pass@181.7
118
Code GenerationHumanEval
pass@186
21
Multi-hop Question AnsweringMuSiQue
Pass@169.4
21
Multi-hop Question AnsweringHotpotQA
pass@1 Success Rate65.9
21
Code GenerationHumanEval Online Collection
pass@182.4
2
Multi-hop Question AnsweringHotpotQA (Online Collection Split)
Pass@161.4
2
Multi-hop Question AnsweringHotpotQA Offline (test)
Pass@161
2
Code GenerationHumanEval Offline (test)
pass@187.9
2
Showing 8 of 8 rows

Other info

Follow for update