tracked author

Aritra Mazumder

University of Utah PhD student in the UtahDB Lab, advised by Anna Fariha, and LRE coauthor. Explicit paper links are required because the same name appears in unrelated research areas.

1 archived notes X: not-found Homepage GitHub

Representative Papers

来自作者已核验个人主页的重点论文；本站单篇归档见下方 Related Notes。

01 AgentCollabBench: Diagnosing When Good Agents Make Bad Collaborators
02 Learning What Not to Forget: Long-Horizon Agent Memory from a Few Kilobytes of Learning

Related Notes

按论文归档时间排序，展示该作者在本站已经出现的材料。

归档 Jul 27, 2026

Learning What Not to Forget: Long Horizon Agent Memory from a Few Kilobytes of Learning

LRE 用日志中的未来复用信号训练轻量逻辑回归，对历史单元做查询无关打分并按预算逐字保留；单随机种子 AppWorld 中以零压缩器调用得到 41.1% 任务目标完成率、接近全历史的 44.0%，LoCoMo 上也是最佳受预算策略，但 Hard 与 LongMemEval 结果显示它尚未跨域占优，2048 token 只约束较旧历史。

待审阅 2606.20954-lre-learned-relevance-eviction Agent Memory Long Context Agent Workflow