Saved in:
Bibliographic Details
Main Authors: Gupta, Naman, Singh, Vaibhav, Iyer, Arun, Shiragur, Kirankumar, Grover, Pratham, Bairi, Ramakrishna B., Maiti, Ritabrata, Damle, Sankarshan, Gupta, Shachee Mishra, Maurya, Rishikesh, C, Vageesh D.
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.09835
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • Sequential multi-agent reasoning frameworks such as Chain-of-Agents (CoA) handle long-context queries by decomposing inputs into chunks and processing them sequentially using LLM-based worker agents that read from and update a bounded shared memory. From a probabilistic perspective, CoA aims to approximate the conditional distribution corresponding to a model capable of jointly reasoning over the entire long context. CoA achieves this through a latent-state factorization in which only bounded summaries of previously processed evidence are passed between agents. The resulting bounded-memory approximation introduces a lossy information bottleneck, making the final evidence state inherently dependent on the order in which chunks are processed. In this work, we study the problem of chunk ordering for long-context reasoning. We use the well-known Chow-Liu trees to learn a dependency structure that prioritizes strongly related chunks. Empirically, we show that a breadth-first traversal of the resulting tree yields chunk orderings that reduce information loss across agents and consistently outperform both default document-chunk ordering and semantic score-based ordering in answer relevance and exact-match accuracy across three long-context benchmarks.