Skip to yearly menu bar Skip to main content


Poster
in
Workshop: Multi-Agent Systems in the Era of Foundation Models: Opportunities, Challenges and Futures

Long Context Scaling: Divide and Conquer via Multi-Agent Question-driven Collaboration

Sibo Xiao · Zixin Lin · Wenyang Gao · Hui Chen · Yue Zhang


Abstract:

Processing long contexts has become a critical capability for modern large language models (LLMs). Existing works leverage agent-based divide-and-conquer methods for processing long contexts. But these methods face crucial limitations, including prohibitive accumulated latency and amplified information loss from excessive agent invocations, and the disruption of inherent textual dependencies by immoderate partitioning. XpandA tackles long-text challenges using: 1) dynamic partitioning for flexible context handling; 2) question-driven updates for consistent knowledge across partitions; and 3) selective replay of partitions to resolve complex information order (e.g., flashbacks). We perform a comprehensive evaluation of XpandA on multiple long-context benchmarks with length varying from 1k to 1M, demonstrating XpandA's feasibility for processing ultra-long sequences and its significant effectiveness in enhancing the long-context capabilities of various LLMs by achieving 20\% improvements and 1.5x inference speedup over baselines of full-context, RAG and previous agent-based methods.

Chat is not available.