@HuaxiuYaoML
๐ง Can agent memory scale without losing reasoning? ๐ฅ Weโre excited to share our latest work, SimpleMem, a principled memory framework for LLM agents built around semantic lossless compression. ๐ 30ร fewer inference tokens ๐ +26.4% avg F1 (vs Mem0) โก 50.2% faster retrieval (vs Mem0) Instead of storing raw interaction history ๐๏ธ or relying on costly iterative reasoning loops ๐, SimpleMem treats memory as a structured, evolving representation whose primary objective is ๐ฏ maximizing information density per token. ๐ Paper: https://t.co/4ZmiXMgzLA ๐ Code: https://t.co/hhdgvViDwh ๐ฆ Website๏ผhttps://t.co/B3AhdoTBtS Nice work @JiaqiLiu835914, Yaofeng Su, @richardxp888, @lillianwei423, and great collab. w/ @cihangxie, Zeyu Zheng, @dingmyu