JointCore：按交互感知为 LLM Agent 选择充分记忆子集
Beyond Marginal Influence: Interaction-Aware Memory Selection for LLM Agents
一篇讲 Agent 记忆筛选的论文，JointCore 这个基线能把要评估的记忆子集次数砍掉六成多，省 87.8% 的活跃记忆，做 Agent 的可以看看思路。
arXiv 论文提出按决策的记忆充分性问题：从 Agent 记忆中选出小规模子集并保持全记忆行为。作者构建 JointCoreBench，含 9 个交互家族、6 个工作流域和精确充分集 oracle。配套的 JointCore 基线先测试记忆块再测试单条记录，在 6,912 个条件下 100.0% 找到最小成本充分集，子集评估次数从 30.00 降至 11.13（62.9%），主动记忆节省 87.8%。在四个 12B-70B 模型家族、每个模型 432 个任务上，相对成本感知单条删除方法，其模型查询均值从 16.00 降到 10.28，最小成本恢复率从 43.1% 升至 50.6%。
Beyond Marginal Influence: Interaction-Aware Memory Selection for LLM Agents
LLM-agent memory selection often uses retrieval rankings based on record relevance or individual influence. Such rankings cannot identify sufficient subsets under redundant or jointly necessary evidence. We study per-decision memory sufficiency: selecting a small subset that preserves full-memory behavior. To separate search error from model error, we introduce JointCoreBench, a benchmark with nine interaction families, six workflow domains, natural-language decisions, and exact sufficient-set oracles, and JointCore, a relevance cost-ordered group-deletion baseline that tests blocks before individual records. On 6,912 exact conditions, cost-ordered JointCore recovers a minimum-cost sufficient set in 100.0% of cases, reduces subset evaluations from 30.00 to 11.13 (62.9%), and achieves 87.8% active-memory savings. Across four 12B-70B model families and 432 tasks per model, its unweighted mean is 10.28 versus 16.00 model queries and 50.6% versus 43.1% minimum-cost recovery relative to cost aware singleton deletion. Both methods preserve full-memory behavior under the deterministic primary interface. Tests with physical deletion, UNKNOWN masks, PENDING sentinels, counterfactual values, and alternative labels identify interface dependent behavior. JointCore is an efficient benchmark baseline; the study does not establish deployment-ready memory compression.
