EvoArena: Tracking Memory Evolution for Robust LLM Agents in Dynamic Environments
Summary
EvoArena introduces a benchmark for evaluating LLM agents in dynamic environments with progressive updates across terminal, software, and social domains, while EvoMem proposes a patch-based memory paradigm that records structured evolution; experiments show current agents achieve only 39.6% accuracy on EvoArena, and EvoMem yields average gains of 1.5% on the benchmark and improvements on GAIA and LoCoMo.
View Cached Full Text
Cached at: 06/12/26, 06:51 AM
Paper page - EvoArena: Tracking Memory Evolution for Robust LLM Agents in Dynamic Environments
Source: https://huggingface.co/papers/2606.13681 Published on Jun 11
#2 Paper of the day Authors:
,
,
,
,
,
,
,
,
,
,
,
,
Abstract
EvoArena benchmark and EvoMem memory paradigm address the challenge of dynamic environments in LLM agents by modeling progressive updates and structured memory evolution, showing improved performance on evolving tasks.
Large language model (LLM) agents have achieved strong performance on a wide range of benchmarks, yet most evaluations assume static environments. In contrast, real-world deployment is inherently dynamic, requiring agents to continually align their knowledge, skills, and behavior with changing environments and updated task conditions. To address this gap, we introduceEvoArena, a benchmark suite that modelsenvironment changesas sequences ofprogressive updatesacross terminal, software, andsocial domains. We further proposeEvoMem, a patch-based memory paradigm that recordsmemory evolutionas structured update histories, enabling agents to reason about environmental evolution through changes in their memory. Experiments show that current agents struggle onEvoArena, achieving an average accuracy of 39.6% across evolving terminal, software, and social-preference domains.EvoMemconsistently improves performance, yielding an average gain of 1.5% onEvoArenaand also improving standard benchmarks such as GAIA and LoCoMo by 6.1% and 4.8%. Beyond individual tasks,EvoMemfurther improveschain-level accuracyby 3.7% onEvoArena, where success requires completing a consecutive sequence of related evolutionary subtasks.Mechanistic analysisshows thatEvoMemimproves evidence capture in the memory, indicating better preservation of complete evolving environment states. Our results highlight the importance of modeling evolution in both evaluation and memory for reliable agent deployment.
View arXiv pageView PDFAdd to collection
Get this paper in your agent:
hf papers read 2606\.13681
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2606.13681 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2606.13681 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2606.13681 in a Space README.md to link it from this page.
Collections including this paper1
Similar Articles
MemEvoBench: Benchmarking Memory MisEvolution in LLM Agents
MemEvoBench introduces the first benchmark for evaluating memory safety in LLM agents, measuring behavioral degradation from adversarial memory injection, noisy outputs, and biased feedback across QA and workflow tasks. The work reveals that memory evolution significantly contributes to safety failures and that static defenses are insufficient.
EvolveMem:Self-Evolving Memory Architecture via AutoResearch for LLM Agents
EvolveMem introduces a self-evolving memory architecture for LLM agents that optimizes retrieval configurations through LLM-powered diagnosis and iterative research cycles, achieving significant performance improvements on benchmarks like LoCoMo and MemBench.
MemGym: a Long-Horizon Memory Environment for LLM Agents
MemGym is a benchmark for evaluating memory formation in LLM agents over long-horizon tasks, unifying existing agent gyms and synthetic pipelines with memory-isolated scores. It spans tool-use dialogue, multi-turn search, coding, and computer use, and includes a lightweight reward model (MemRM) for efficient evaluation.
@hyunji_amy_lee: LLM agents & memory systems operate in continuously updated environments (Git repos, evolving docs). They must process …
MINTEval is a new benchmark for evaluating LLM agents and memory systems in continuously updated environments with frequent context changes. It shows that current systems perform poorly, with an average accuracy of 27.9% across representative systems.
Scaling Self-Evolving Agents via Parametric Memory
Researchers from Alibaba/Qwen and Peking University introduce TMEM, a self-evolving parametric memory framework that uses online LoRA weight updates to let LLM agents genuinely learn from experience within a single episode, rather than relying solely on prompt-space memory. TMEM outperforms summary-based and retrieval-based baselines across multiple benchmarks including LoCoMo, LongMemEval-S, and CL-Bench.