Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Amory: Building Coherent Narrative-Driven Agent Memory through Agentic Reasoning

Yue Zhou, Xiaobo Guo, Belhassen Bayar, Srinivasan H. Sengamedu

TL;DR

Amory is introduced, a working memory framework that actively constructs structured memory representations through enhancing agentic reasoning during offline time and achieves considerable improvements over previous state-of-the-art approaches.

Abstract

Long-term conversational agents face a fundamental scalability challenge as interactions extend over time: repeatedly processing entire conversation histories becomes computationally prohibitive. Current approaches attempt to solve this through memory frameworks that predominantly fragment conversations into isolated embeddings or graph representations and retrieve relevant ones in a RAG style. While computationally efficient, these methods often treat memory formation minimally and fail to capture the subtlety and coherence of human memory. We introduce Amory, a working memory framework that actively constructs structured memory representations through enhancing agentic reasoning during offline time. Amory organizes conversational fragments into episodic narratives, consolidates memories with momentum, and semanticizes peripheral facts into semantic memory. At retrieval time, the system employs coherence-driven reasoning over narrative structures. Evaluated on the LOCOMO benchmark for long-term reasoning, Amory achieves considerable improvements over previous state-of-the-art, with performance comparable to full context reasoning while reducing response time by 50%. Analysis shows that momentum-aware consolidation significantly enhances response quality, while coherence-driven retrieval provides superior memory coverage compared to embedding-based approaches.

Amory: Building Coherent Narrative-Driven Agent Memory through Agentic Reasoning

TL;DR

Amory is introduced, a working memory framework that actively constructs structured memory representations through enhancing agentic reasoning during offline time and achieves considerable improvements over previous state-of-the-art approaches.

Abstract

Long-term conversational agents face a fundamental scalability challenge as interactions extend over time: repeatedly processing entire conversation histories becomes computationally prohibitive. Current approaches attempt to solve this through memory frameworks that predominantly fragment conversations into isolated embeddings or graph representations and retrieve relevant ones in a RAG style. While computationally efficient, these methods often treat memory formation minimally and fail to capture the subtlety and coherence of human memory. We introduce Amory, a working memory framework that actively constructs structured memory representations through enhancing agentic reasoning during offline time. Amory organizes conversational fragments into episodic narratives, consolidates memories with momentum, and semanticizes peripheral facts into semantic memory. At retrieval time, the system employs coherence-driven reasoning over narrative structures. Evaluated on the LOCOMO benchmark for long-term reasoning, Amory achieves considerable improvements over previous state-of-the-art, with performance comparable to full context reasoning while reducing response time by 50%. Analysis shows that momentum-aware consolidation significantly enhances response quality, while coherence-driven retrieval provides superior memory coverage compared to embedding-based approaches.
Paper Structure (32 sections, 5 figures, 4 tables, 2 algorithms)

This paper contains 32 sections, 5 figures, 4 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Existing memory framework for long-term interactions.
  4. A cognitive perspective on memory.
  5. Amory: Agentic Memory Framework
  6. Preliminary
  7. Enhancing agentic reasoning for memory formation.
  8. Framework Overview and Components
  9. Memory Initialization.
  10. Memory Binding.
  11. Memory Consolidation with Momentum.
  12. Semanticization.
  13. Memory Retrieval.
  14. Experiments
  15. Settings
  16. ...and 17 more sections

Figures (5)

  • Figure 1: Overview of Amory framework.
  • Figure 2: Offline memory processing complexity.
  • Figure 3: Episodic memory evolvement.
  • Figure 4: Memory coverage rate using coherence-driven and embedding-based retriever (Left and Middle) and context compression rate at different top-k for memory retrieving (Right).
  • Figure 5: Tree structure of plot and subplot of some episodic memories.