---
title: "CAMR: Compatibility-Aware Memory Replay for Continual Learning in Large Language Models"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2610.36734v1"
permalink: "https://feed.astralor.com/items/itm_56be047b3f8a64a1cb2280cb"
published: "2026-10-27T06:52:54.000Z"
score: 57
channel: "finance"
tags: ["LLM","多模态","CAMR","Compatibility-Aware Memory Replay","Continual Learning","Large Language Models","Continual","Fine-tuning"]
---

# CAMR: Compatibility-Aware Memory Replay for Continual Learning in Large Language Models

## At a glance

Continual learning requires large language models to acquire new knowledge and capabilities while retaining what they have learned. Fine-tuning and replay support acquisition and retention, but deciding whether incoming…

## Key points

- Continual learning requires large language models to acquire new knowledge and capabilities while retaining what they have learned.
- Fine-tuning and replay support acquisition and retention, but deciding whether incoming examples need supervision and which historical examples to revisit remains a challenge.
- We propose Compatibility-Aware Memory Replay (CAMR), an adaptive framework that jointly controls current-example supervision and historical replay.

## Full text

Continual learning requires large language models to acquire new knowledge and capabilities while retaining what they have learned. Fine-tuning and replay support acquisition and retention, but deciding whether incoming examples need supervision and which historical examples to revisit remains a challenge. We propose Compatibility-Aware Memory Replay (CAMR), an adaptive framework that jointly controls current-example supervision and historical replay. CAMR combines compatibility checks with a learned memory gate to guide an online policy over three actions: learning incoming examples with general replay, jointly reinforcing compatible related examples, and skipping direct supervision of verified mastered duplicates. The policy learns from acquisition, retention, and training-cost feedback under a bounded per-arrival replay budget. We evaluate CAMR with Qwen3-8B-Base on continual factual updating and cross-task and cross-domain learning. On WikiBigEdit, CAMR consistently leads evaluated methods in Original Match from 3K to 100K facts, reaching 93.66% after just 3K facts and surpassing Full-history Replay (LoRA) after 100K facts. It also achieves the highest final scores on all four selected TRACE tasks and maintains or improves earlier-task performance relative to post-acquisition levels in a sequence spanning financial question answering, mathematical reasoning, code generation, and summarization.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_56be047b3f8a64a1cb2280cb