Chained Recursive Language Models for Multi-Iteration Reasoning
2026-08-05 • Computation and Language
Computation and LanguageArtificial IntelligenceInformation TheoryMachine Learning
AI summaryⓘ
The authors address a problem where large language models (LLMs) struggle to handle long and complex tasks all at once because mistakes early on can affect the final answer. They propose a method called Chained Recursive Language Models (Chained RLM), which breaks the task into smaller pieces, with the model working on each piece separately but sharing summaries and notes from previous steps. This approach lets the model revisit and improve on intermediate results before producing the final answer. The authors explain how their system works and test when this method improves accuracy compared to normal LLM use.
large language modelslong context reasoninginference trajectorymulti-hop reasoningrecursive modelstask decompositionblackboard architectureintermediate artifactscontext management
Authors
Purbesh Mitra, Sennur Ulukus
Abstract
Long context reasoning in large language models (LLMs) is usually constrained by the fact that a single inference trajectory has to simultaneously explore the context, store intermediate state, verify evidence, and produce the final answer. This becomes particularly difficult in tasks that require extraction, counting, ordering, or multi-hop reasoning, where an early mistake can propagate until the final response. In this work, we propose Chained Recursive Language Models (Chained RLM), an inference-time architecture, in which the same underlying model is called repeatedly as a sequence of fresh reasoning roots. Each root receives the original problem and context, but does not inherit the full conversational history. Instead, it receives a compact plain-text summary, a plain-text blackboard, and some durable task-specific artifacts written by predecessor roots. The motivation is to manage the context by chopping into partial tasks rather than one large inference response; in each staged computation, intermediate artifacts can be inspected, corrected, and extended by a later fresh inference by the same model. We describe the system model, handoff mechanism, artifact workspace, and evaluation protocol for this system. We study when fresh-context artifact continuation gives a measurable gain in accuracy over direct LLM answering even with recursive tool-calling.