Papers
arxiv:2604.11610

Self-Evolving LLM Memory Extraction Across Heterogeneous Tasks

Published on Apr 13
· Submitted by
Yuqing Yang
on Apr 23
Authors:
,
,
,
,

Abstract

LLM-based assistants require heterogeneous memory extraction capabilities, which are evaluated through the BEHEMOTH benchmark, with CluE offering improved performance through cluster-based prompt optimization.

AI-generated summary

As LLM-based assistants become persistent and personalized, they must extract and retain useful information from past conversations as memory. However, the types of information worth remembering vary considerably across tasks. We formalize the heterogeneous memory extraction task and introduce BEHEMOTH, a benchmark that repurposes 18 existing datasets spanning personalization, problem-solving, and agentic tasks, using a downstream utility-driven metric for systematic evaluation. Our empirical analysis confirms that no single static extraction prompt dominates across all task categories, and that existing self-evolving prompt optimization frameworks, originally designed for homogeneous distributions, degrade when training tasks are heterogeneous. To address this, we propose CluE, a cluster-based self-evolving strategy that groups training examples into clusters by extraction scenarios, analyzes each cluster independently, and synthesizes cross-cluster insights to update the extraction prompt. Experiments on BEHEMOTH show that CluE generalizes effectively across heterogeneous tasks (+9.04\% relative gain), consistently outperforming prior self-evolving frameworks.

Community

Paper author Paper submitter

We propose BEHEMOTH, an 18-dataset benchmark for heterogeneous memory extraction, and CluE, a cluster-based self-evolving method that consistently outperforms compared baselines.

Sign up or log in to comment

Get this paper in your agent:

hf papers read 2604.11610
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2604.11610 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2604.11610 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2604.11610 in a Space README.md to link it from this page.

Collections including this paper 1