Reproduce “Rethinking User Encoders: Routing Diffusion States for Scenario-Adaptive User Representation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Junru Zhang, Jinyong Wen, Bin Dou, Wuliang Huang, Baokun Wang, Yu Cheng, Weiqiang Wang, Cheng Peng

FINDERS addresses user modeling for decisions that depend on different evidence by preserving multiple behavioral views and routing them according to the incoming query before compression. The method uses diffusion-model reconstruction states and learned slots to represent complementary evidence, then forms a query-conditioned representation in one clean forward pass. On the offline Promo task, AUC is 0.7938 versus 0.7868 for Q-Anchor + Prompt Tuning; this supports the design in the evaluated setting, not a universal advantage.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.