DYCP: Dynamic Context Pruning for Long-Form Dialogue with LLMs

TL;DR

DyCP is a lightweight, outside-the-model method that dynamically manages dialogue context by selectively retrieving relevant segments, improving efficiency and maintaining dialogue flow in long-form conversations.

Contribution

It introduces DyCP, a novel method for dynamic, outside-the-LLM context management that adapts to dialogue shifts without offline memory construction.

Findings

DyCP achieves competitive answer quality across benchmarks.

It improves inference efficiency by more selective context usage.

DyCP maintains dialogue sequentiality without predefined topic boundaries.

Abstract

Large Language Models (LLMs) increasingly operate over long-form dialogues with frequent topic shifts. While recent LLMs support extended context windows, efficient management of dialogue history in practice is needed due to inference cost and latency constraints. We present DyCP, a lightweight context management method implemented outside the LLM that dynamically identifies and retrieves relevant dialogue segments conditioned on the current turn, without offline memory construction. DyCP manages dialogue context while preserving the sequential nature of dialogue without predefined topic boundaries, enabling adaptive and efficient context selection. Across three long-form dialogue benchmarks-LoCoMo, MT-Bench+, and SCM4LLMs-and multiple LLM backends, DyCP achieves competitive answer quality in downstream generation, with more selective context usage and improved inference efficiency.