NeFut Logo NeFut
Admin Login

[CS.AI] Revolutionary ARC Framework: Enhancing Long-Term Memory Management for AI Agents

Published at: 2026-07-29 22:00 Last updated: 2026-07-30 03:24
#AI #Machine Learning #Open Source

Abstract

Long-horizon LLM agents accumulate reasoning traces, actions, and tool observations that can eventually exceed a model's fixed context window. Existing compaction methods address this limitation by discarding, summarizing, or retrieving earlier information, but they may remove task-critical details or fail to recover them reliably.

We propose ARC (Addressable Recall Compaction), a context-management framework that separates archival storage from active-context presentation. ARC stores tool observations in an append-only, ID-addressable log and replaces older observations with compact citations when compaction is required. The agent can subsequently use these identifiers to request stored content without re-executing the corresponding tools or depending solely on similarity-based retrieval.

We evaluate ARC using Qwen3-8B with a 16k context window and Qwen3-32B with a 32k context window. On the Needle-in-a-Haystack evaluation, ARC achieves an average exact-answer accuracy of 99.40%, compared with 88.12% for the best-performing baseline in our evaluation. ARC also reduces estimated serving time and HBM traffic under our hardware-cost model. On the LongBench-v2 Hard subset, ARC obtains an average accuracy of 29.97%, compared with 28.25% for the best-performing baseline. These results indicate that explicit, address-based recall can improve information retention and serving efficiency relative to the evaluated context-management baselines under the tested settings.

Blogger's Review: The introduction of the ARC framework presents an innovative approach to addressing the information management challenges faced by LLM agents in long-horizon tasks. Its effective compression and storage methods for context could significantly enhance the performance and responsiveness of agents in real-world applications. Future research may further explore optimizing ARC's performance in more complex tasks.

Original Source: https://arxiv.org/abs/2607.25066

[h] Back to Home