# Cooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM Conversations

Source: [Cooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM Conversations](<https://www.sota2.com/research/paper/69df1b757d0172140ec7b7fb>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-27T05:00:13.467Z

Authors: Ziyang Liu

Published: 2026-04-14T07:06:35.000Z

## Abstract

When LLM conversations grow beyond the context window, old content must be evicted -- but how does the model recover it when needed? We propose cooperative paging: evicted segments are replaced with minimal keyword bookmarks (\[pN:keywords\], ~8-24 tokens each), and the model is given a recall() tool to retrieve full content on demand. On the LoCoMo benchmark (10 real multi-session conversations, 300+ turns), cooperative paging achieves the highest answer quality among six methods -- outperforming truncation, BM25, word-overlap retrieval, a search-tool baseline, and full context -- on four models (GPT-4o-mini, DeepSeek-v3.2, Claude Haiku, GLM-5), confirmed by four independent LLM judges ($p=0.017$, paired bootstrap). We then study the paging design space with a 5x4 ablation over boundary strategies and eviction policies (3,176 synthetic probes, 1,600 LoCoMo probes). Key findings: (1) coarse fixed-size pages (fixed\_20) reach 96.7% while content-aware topic\_shift collapses to 56.7%; (2) eviction policy choice is data-dependent (FIFO best on synthetic, LFU on LoCoMo); (3) two bookmark generation strategies improve over the heuristic baseline (+4.4 and +8.7 E2E points); (4) the remaining bottleneck is bookmark discrimination -- the model triggers recall() 96% of the time but selects the correct page only 57% when bookmarks are insufficiently distinctive. Keyword specificity alone accounts for a 25 percentage point accuracy difference.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Multi-session Conversation Question Answering on LoCoMo 98 QA probes](<https://www.sota2.com/research/sota/multi-session-conversation-question-answering-on-locomo-98-qa-probes.md>): Multi-session Conversation Question Answering; LoCoMo 98 QA probes; Score (1-5): 2.18.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.12376>)

[PDF](<https://arxiv.org/pdf/2604.12376.pdf>)