Bounded Personas Match Retrieval on Classification but Not Regression for a Frozen Agent
Source
JaeHa Yoon, Minjun Park, Seoyeon Kim, Jiwoo Lee, Hyunwoo Choi, Dohyun Kang
Author
JaeHa Yoon, Minjun Park, Seoyeon Kim, Jiwoo Lee, Hyunwoo Choi, Dohyun Kang
Date
Terms in this piece · Glossary
distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.
Why it matters
Retrieval-augmented personalization gets expensive as user history grows. This shows a bounded, self-refining persona can match retrieval's accuracy on classification-style personalization tasks, but not yet on numeric regression tasks, a useful boundary for choosing between the two.
articleParametric Multimodal User Memory: Storing What Captions Cannot CarryBojie Li, Noah Shi
articleThe Memory Trust Gap: Capability-Dependent Failures in Persistent-Memory AgentsJundong Hu, Shekar Ramachandran
articleFinPerMA: A Theory-Informed, Event-Grounded Personalized-Memory Benchmark for LLM AgentsBen Wang, Kang Zhou, Lifan Guo, Feng Chen, Chi Zhang