---
title: "Axis-LoRA: Probing and Adapting Embedding Models for Personal-Memory Retrieval"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2611.07820v1"
permalink: "https://feed.astralor.com/items/itm_691f78eb768e45175386c0c2"
published: "2026-11-05T10:59:23.000Z"
score: 57
channel: "tech"
tags: ["LLM","RAG","Axis-LoRA","Probing","Adapting Embedding Models","Personal-Memory Retrieval","LLMs","Current"]
---

# Axis-LoRA: Probing and Adapting Embedding Models for Personal-Memory Retrieval

## At a glance

As LLMs are increasingly asked questions about a user's personal experiences, retrieval must distinguish memories by time, place, person, and event. Current personal-memory pipelines rely on frozen dense embedding model…

## Key points

- As LLMs are increasingly asked questions about a user's personal experiences, retrieval must distinguish memories by time, place, person, and event.
- Current personal-memory pipelines rely on frozen dense embedding models to embed memories for retrieval, but we find that these models do not capture the four axes well.
- Swapping a memory's date or place barely shifts its embedding, and a trained projection on top of the frozen embedding does not recover the missing signal.

## Full text

As LLMs are increasingly asked questions about a user's personal experiences, retrieval must distinguish memories by time, place, person, and event. Current personal-memory pipelines rely on frozen dense embedding models to embed memories for retrieval, but we find that these models do not capture the four axes well. Swapping a memory's date or place barely shifts its embedding, and a trained projection on top of the frozen embedding does not recover the missing signal. The problem lives inside the model, where each axis is encoded by a distinct attention-head pattern. Guided by these findings, we propose Axis-LoRA, a per-axis adapter that places four LoRA branches inside the embedding model and combines them via a per-query gate. Across three standard embedding models evaluated on LongMemEval, ATM-Bench, and LoCoMo, Axis-LoRA improves 4-benchmark mean R@10 by $+1.28$ to $+1.56$~pp while training only a small fraction of the base parameters on a small axis-counterfactual training set. Code is available at https://github.com/dohoonkim-ai/axis-lora.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_691f78eb768e45175386c0c2