arXiv Machine Learning
New Model, Old Habits: Behavior Transfer under Model Swapping in Multi-Session LLM Agents
Read the original on arXiv Machine Learning →The Flow has not summarised this story yet — read it at arXiv Machine Learning.
Related stories
arXiv AI
2d ago
SUMONE: A Simple and Efficient Summarizer for Plug-and-Play Agent Memory
SUMONE is a lightweight summarizer designed to compress long interaction histories of language‑model agents. It is trained via masked reconstruction, where a frozen reconstructor predicts randomly masked parts of the original trajectory; reconstruction quality plus a length penalty yields a task‑agnostic reward for optimizing the summarizer with GRPO. At inference, only the summarizer remains, enabling plug‑and‑play context compression that reduces inference cost by roughly 78 % on DeepSearchQA and 75 % on SWE‑bench without affecting the underlying agent.
arXiv Machine Learning
Jul 24
LLMs Get Lost in Evolving User Intent
arXiv:2607. 20734v1 Announce Type: new Abstract: As LLMs become more capable, they are increasingly deployed as collaborative agents, taking on user-delegated tasks through iterative interaction.
arXiv AI
Aug 6
FinPerMA: A Theory-Informed, Event-Grounded Personalized-Memory Benchmark for LLM Agents
arXiv:2608. 04095v1 Announce Type: new Abstract: Large language model (LLM) agents are increasingly used as personalized assistants in high-stakes domains such as financial advising, yet it remains unclear whether they can maintain and update an individualized user model over long horizons.
arXiv Computation and Language
Oct 9
DPPM: Dual-Path Parametric Memory for Personalized Language Models
The paper introduces Dual-Path Parametric Memory (DPPM), a method for long‑term personalization of language models that leverages interaction history without requiring it in the inference context. DPPM uses two complementary paths: an Evidence path that pools historical representations to retain earlier evidence, and a Delta path that sequentially updates an associative state to capture changes. By fusing these outputs into history‑conditioned LoRA adapters, DPPM achieves superior performance on PersonaMem‑v2 (54.22%) and PrefEval (86.79%) compared to existing baselines.
arXiv AI
Sep 10
EdgeMem: LLM-Free Agent Memory Construction and Retrieval via Evidence-Preserving Multi-Anchor Hypergraph
EdgeMem is a new agent-memory method that preserves original interaction turns and organizes them using complementary content, temporal, and episodic cues via a multi‑anchor hypergraph. It performs lightweight local processing, returning source evidence directly and reserving LLM use only for final answer generation. Experiments on LoCoMo and LongMemEval‑S demonstrate strong retrieval and memory‑grounded question answering, with EdgeMem achieving the highest strict‑judge score among seven systems on LoCoMo while requiring no generative‑LLM calls for construction and retrieval.
arXiv AI
Jun 12
ReSum: Synergizing LLM Reasoning and Summarization with Reinforcement Learning
arXiv:2606. 13316v1 Announce Type: new Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) is a central technique for improving long-horizon reasoning in Large Language Models (LLMs).