COSMIR: Chain Orchestrated Structured Memory for Iterative Reasoning over Long Context

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gupta, Naman, Gowaikar, Shreeyash, Iyer, Arun, Shiragur, Kirankumar, Bairi, Ramakrishna B, Maurya, Rishikesh, Maiti, Ritabrata, Damle, Sankarshan, Gupta, Shachee Mishra
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918154810163200
author Gupta, Naman
Gowaikar, Shreeyash
Iyer, Arun
Shiragur, Kirankumar
Bairi, Ramakrishna B
Maurya, Rishikesh
Maiti, Ritabrata
Damle, Sankarshan
Gupta, Shachee Mishra
author_facet Gupta, Naman
Gowaikar, Shreeyash
Iyer, Arun
Shiragur, Kirankumar
Bairi, Ramakrishna B
Maurya, Rishikesh
Maiti, Ritabrata
Damle, Sankarshan
Gupta, Shachee Mishra
contents Reasoning over very long inputs remains difficult for large language models (LLMs). Common workarounds either shrink the input via retrieval (risking missed evidence), enlarge the context window (straining selectivity), or stage multiple agents to read in pieces. In staged pipelines (e.g., Chain of Agents, CoA), free-form summaries passed between agents can discard crucial details and amplify early mistakes. We introduce COSMIR (Chain Orchestrated Structured Memory for Iterative Reasoning), a chain-style framework that replaces ad hoc messages with a structured memory. A Planner agent first turns a user query into concrete, checkable sub-questions. worker agents process chunks via a fixed micro-cycle: Extract, Infer, Refine, writing all updates to the shared memory. A Manager agent then Synthesizes the final answer directly from the memory. This preserves step-wise read-then-reason benefits while changing both the communication medium (structured memory) and the worker procedure (fixed micro-cycle), yielding higher faithfulness, better long-range aggregation, and auditability. On long-context QA from the HELMET suite, COSMIR reduces propagation-stage information loss and improves accuracy over a CoA baseline.
format Preprint
id arxiv_https___arxiv_org_abs_2510_04568
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle COSMIR: Chain Orchestrated Structured Memory for Iterative Reasoning over Long Context
Gupta, Naman
Gowaikar, Shreeyash
Iyer, Arun
Shiragur, Kirankumar
Bairi, Ramakrishna B
Maurya, Rishikesh
Maiti, Ritabrata
Damle, Sankarshan
Gupta, Shachee Mishra
Artificial Intelligence
Machine Learning
Reasoning over very long inputs remains difficult for large language models (LLMs). Common workarounds either shrink the input via retrieval (risking missed evidence), enlarge the context window (straining selectivity), or stage multiple agents to read in pieces. In staged pipelines (e.g., Chain of Agents, CoA), free-form summaries passed between agents can discard crucial details and amplify early mistakes. We introduce COSMIR (Chain Orchestrated Structured Memory for Iterative Reasoning), a chain-style framework that replaces ad hoc messages with a structured memory. A Planner agent first turns a user query into concrete, checkable sub-questions. worker agents process chunks via a fixed micro-cycle: Extract, Infer, Refine, writing all updates to the shared memory. A Manager agent then Synthesizes the final answer directly from the memory. This preserves step-wise read-then-reason benefits while changing both the communication medium (structured memory) and the worker procedure (fixed micro-cycle), yielding higher faithfulness, better long-range aggregation, and auditability. On long-context QA from the HELMET suite, COSMIR reduces propagation-stage information loss and improves accuracy over a CoA baseline.
title COSMIR: Chain Orchestrated Structured Memory for Iterative Reasoning over Long Context
topic Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2510.04568