Scaling LLM Multi-turn RL with End-to-end Summarization-based Context Management

• URL: http://arxiv.org/abs/2510.06727v1
• Date: Wed, 08 Oct 2025 07:29:22 GMT
• Title: Scaling LLM Multi-turn RL with End-to-end Summarization-based Context Management
• Authors: Miao Lu, Weiwei Sun, Weihua Du, Zhan Ling, Xuesong Yao, Kang Liu, Jiecao Chen,
• Abstract summary: We introduce summarization-based context management to training.<n>We instantiate this framework with underlineSUmmarization augmented underlinePolicy underlineOptimization (textttSUPO)<n>Our results establish summarization-based context management as a principled and scalable approach for training RL agents beyond a fixed context length limit.
• Score: 19.980762483472354
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: We study reinforcement learning (RL) fine-tuning of large language model (LLM) agents for long-horizon multi-turn tool use, where context length quickly becomes a fundamental bottleneck. Existing RL pipelines can suffer from degraded instruction following, excessive rollout costs, and most importantly, strict context limits. To address these challenges, we introduce summarization-based context management to training. In specific, it periodically compresses the tool using history by LLM-generated summaries that retain task-relevant information to keep a compact context while enabling the agent to scale beyond the fixed context window. Building on this formulation, we derive a policy gradient representation that seamlessly enables standard LLM RL infrastructures to optimize both tool-use behaviors as well as summarization strategies in an end-to-end fashion. We instantiate this framework with \underline{SU}mmarization augmented \underline{P}olicy \underline{O}ptimization (\texttt{SUPO}), an LLM RL algorithm that enables long-horizon training beyond a fixed context limit. Experiments on interactive function calling and searching tasks demonstrate that \texttt{SUPO} significantly improves the success rate while maintaining the same or even lower working context length compared to baselines. We also demonstrate that for complex searching tasks, \texttt{SUPO} can further improve the evaluation performance when scaling test-time maximum round of summarization beyond that of training time. Our results establish summarization-based context management as a principled and scalable approach for training RL agents beyond a fixed context length limit.

Related papers

• Memory-Based Advantage Shaping for LLM-Guided Reinforcement Learning [18.215893951726166]
  In environments with sparse or delayed rewards, reinforcement learning incurs high sample complexity.<n>This limitation has motivated the use of large language models (LLMs) for subgoal discovery and trajectory guidance.<n>We address these challenges by constructing a memory graph that encodes subgoals and trajectories from both LLM guidance and the agent's own successful rollouts.
  arXiv  Detail & Related papers  (2026-02-20T01:44:35Z)
• Zero-Shot Instruction Following in RL via Structured LTL Representations [54.08661695738909]
  Linear temporal logic (LTL) is a compelling framework for specifying complex, structured tasks for reinforcement learning (RL) agents.<n>Recent work has shown that interpreting instructions as finite automata, which can be seen as high-level programs monitoring task progress, enables learning a single generalist policy capable of executing arbitrary instructions at test time.<n>We propose a novel approach to learning a multi-task policy for following arbitrary instructions that addresses this shortcoming.
  arXiv  Detail & Related papers  (2025-12-02T10:44:51Z)
• Scaling Long-Horizon LLM Agent via Context-Folding [46.685552398338295]
  We introduce Context-Folding, a framework that empowers agents to actively manage their working context.<n>An agent can procedurally branch into a sub-trajectory to handle a subtask and then fold it upon completion, collapsing the intermediate steps while retaining a concise summary of the outcome.
  arXiv  Detail & Related papers  (2025-10-13T22:00:58Z)
• ContextNav: Towards Agentic Multimodal In-Context Learning [85.05420047017513]
  ContextNav is an agentic framework that integrates the scalability of automated retrieval with the quality and adaptiveness of human-like curation.<n>It builds a resource-aware multimodal embedding pipeline, maintains a retrievable vector database, and applies agentic retrieval and structural alignment to construct noise-resilient contexts.<n> Experimental results demonstrate that ContextNav achieves state-of-the-art performance across various datasets.
  arXiv  Detail & Related papers  (2025-10-06T07:49:52Z)
• Agentic Reinforced Policy Optimization [66.96989268893932]
  Large-scale reinforcement learning with verifiable rewards (RLVR) has demonstrated its effectiveness in harnessing the potential of large language models (LLMs) for single-turn reasoning tasks.<n>Current RL algorithms inadequately balance the models' intrinsic long-horizon reasoning capabilities and their proficiency in multi-turn tool interactions.<n>We propose Agentic Reinforced Policy Optimization (ARPO), a novel agentic RL algorithm tailored for training multi-turn LLM-based agents.
  arXiv  Detail & Related papers  (2025-07-26T07:53:11Z)
• Scalable In-Context Q-Learning [68.9917436397079]
  We propose textbfScalable textbfIn-textbfContext textbfQ-textbfLearning (textbfSICQL) to steer in-context reinforcement learning.<n>textbfSICQL harnesses dynamic programming and world modeling to steer ICRL toward efficient reward and task generalization.
  arXiv  Detail & Related papers  (2025-06-02T04:21:56Z)
• Can Long-Context Language Models Subsume Retrieval, RAG, SQL, and More? [54.667202878390526]
  Long-context language models (LCLMs) have the potential to revolutionize our approach to tasks traditionally reliant on external tools like retrieval systems or databases. We introduce LOFT, a benchmark of real-world tasks requiring context up to millions of tokens designed to evaluate LCLMs' performance on in-context retrieval and reasoning. Our findings reveal LCLMs' surprising ability to rival state-of-the-art retrieval and RAG systems, despite never having been explicitly trained for these tasks.
  arXiv  Detail & Related papers  (2024-06-19T00:28:58Z)
• Long Context Alignment with Short Instructions and Synthesized Positions [56.1267385315404]
  This paper introduces Step-Skipping Alignment (SkipAlign) It is a new technique designed to enhance the long-context capabilities of Large Language Models (LLMs) With a careful selection of the base model and alignment datasets, SkipAlign with only 6B parameters achieves it's best performance and comparable with strong baselines like GPT-3.5-Turbo-16K on LongBench.
  arXiv  Detail & Related papers  (2024-05-07T01:56:22Z)
• How Can LLM Guide RL? A Value-Based Approach [68.55316627400683]
  Reinforcement learning (RL) has become the de facto standard practice for sequential decision-making problems by improving future acting policies with feedback. Recent developments in large language models (LLMs) have showcased impressive capabilities in language understanding and generation, yet they fall short in exploration and self-improvement capabilities. We develop an algorithm named LINVIT that incorporates LLM guidance as a regularization factor in value-based RL, leading to significant reductions in the amount of data needed for learning.
  arXiv  Detail & Related papers  (2024-02-25T20:07:13Z)
• AMAGO: Scalable In-Context Reinforcement Learning for Adaptive Agents [36.71024242963793]
  We introduce AMAGO, an in-context Reinforcement Learning agent that uses sequence models to tackle the challenges of generalization, long-term memory, and meta-learning. Our agent is scalable and applicable to a wide range of problems, and we demonstrate its strong performance empirically in meta-RL and long-term memory domains.
  arXiv  Detail & Related papers  (2023-10-15T22:20:39Z)
• On Context Distribution Shift in Task Representation Learning for Offline Meta RL [7.8317653074640186]
  We focus on context-based OMRL, specifically on the challenge of learning task representation for OMRL. To overcome this problem, we present a hard-sampling-based strategy to train a robust task context encoder.
  arXiv  Detail & Related papers  (2023-04-01T16:21:55Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.