ReContext: Recursive Evidence Replay as LLM Harness for Long-Context Reasoning

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Yanjun Zhao, Ruizhong Qiu, Tianxin Wei, Yuanchen Bei, Zhining Liu, Lingjie Chen, Ismini Lourentzou, Hanghang Tong, Jingrui He
• Venue: arXiv: cs.AI
• Published: 2026-07-02
• Collected: 2026-07-03
• URL: http://arxiv.org/abs/2607.02509v1
• PDF: https://arxiv.org/pdf/2607.02509v1
• DOI: UNKNOWN

Plain-Language Brief

Researchers propose a new method called Recursive Evidence Replay (RECONTEXT) to improve the ability of large language models (LLMs) to understand and reason over long contexts. This method doesn't require training and helps the model utilize relevant evidence from the input more effectively by creating a query-conditioned evidence pool that is replayed before final generation.

What the Document Says

• The text discusses a new method called Recursive Evidence Replay (RECONTEXT) for improving long-context reasoning in large language models (LLMs).
• This method doesn't require training and works during the inference phase.
• RECONTEXT organizes evidence into a query-conditioned pool, which is then replayed before final generation while preserving the full original context.
• The recursive selection process separates evidence organization from answer generation without using external memory or pruning the context.
• A theoretical analysis based on associative memory is provided to characterize the method.
• Experiments were conducted on eight long-context datasets with a context length of 128K, showing that RECONTEXT improves evidence utilization across several LLMs and achieves the best average rank on all tested backbones.

Key Concepts Mentioned

• Recursive Evidence Replay (RECONTEXT)
• Large Language Models (LLMs)
• Long-context reasoning
• Query-conditioned evidence pool
• Theoretical analysis based on associative memory

Claims Made by the Document

• RECONTEXT improves long-context reasoning in LLMs.
• This improvement is achieved without training, external memory, or context pruning.

Evidence Used

• Improvements are shown through experiments on eight long-context datasets with a context length of 128K.
• RECONTEXT achieves the best average rank on all three backbones tested.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Understanding and reasoning over long contexts has become a key requirement for deploying large language models (LLMs) in realistic applications. Although recent LLMs support increasingly long context windows, they often fail to use relevant evidence that is already present in the input, revealing a gap between context access and effective context utilization. In this work, we propose Recursive Evidence Replay as LLM Harness for Long-Context Reasoning (RECONTEXT), a training-free inference method for improving long-context reasoning. RECONTEXT uses model-internal relevance signals to construct a query-conditioned evidence pool and replays it before final generation while preserving the full original context. This recursive selection process separates evidence organization from answer generation without training, external memory, or context pruning. We also provide a theoretical analysis based on associative memory, which characterizes the context as a memory store, the question as a retrieval cue, attention as cue-trace association, and replay as trace reactivation. Experiments on eight long-context datasets with 128K context length show that RECONTEXT consistently improves evidence utilization across Qwen3-4B, Qwen3-8B, and Llama3-8B, achieving the best average rank on all three backbones. Code is available at https://github.com/Yanjun-Zhao/ReContext.