SaliMory: Orchestrating Cognitive Memory for Conversational Agents
2026-06-02 • Computation and Language
Computation and LanguageArtificial Intelligence
AI summaryⓘ
The authors developed SALIMORY, a new method to help chatbots remember and use important information about users over time. Instead of just adding more text to remember, their approach organizes memories into stages like filtering, saving, and recalling facts, making the chatbot smarter. They trained one model to manage this process using special rewards for each memory step. Their method reduced memory mistakes by a third and made the chatbot better at personalizing responses by more than twice compared to older models.
conversational agentspersistent memoryreinforcement learningcredit assignmenthierarchical rewardsmemory consolidationcontrastive refinementpersonalizationlanguage models
Authors
Kai Zhang, Xinyuan Zhang, Hongda Jiang, Shiun-Zu Kuo, Hyokun Yun, Ejaz Ahmed, Shereen Oraby, Ziyun Li, Sanat Sharma, Ann Lee, Ahmed A Aly, Anuj Kumar, Raffay Hamid, Xin Luna Dong
Abstract
Conversational agents that serve as lifelong companions must maintain persistent memory across all interactions. However, simply expanding context windows with raw retrieval degrades reasoning quality, while training memory agents via standard reinforcement learning creates a severe credit assignment bottleneck in a multi-stage pipeline. To solve this, we introduce SALIMORY, a framework that trains a single language model to manage a cognitively-structured memory-spanning user facts, preferences, and working memory. By introducing a hierarchical stage-wise process reward and reward-decomposed contrastive refinement, SALIMORY provides isolated supervision for distinct memory operations (selective filtering, consolidation, and cue-driven recall) end-to-end. SALIMORY cuts memory-attributed failures by one-third, outperforms the state-of-the-art by over 10% in end-to-end accuracy, and more than doubles the Good Personalization rate.