Reproduce “From Interests to Semantic IDs: Retrieval-Grounded Credit Assignment for Generative Recommendation” 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
Mengdan Zhu, Yufan Zhao, Yao Zhao, Sophie Di, Tao Di, Yulan Yan, Sridhar Iyer, Liang Zhao
The paper introduces query-attributed retrieval reward for reasoning-enhanced generative recommendation. A model states future interests, a frozen catalog retriever tests each interest against the held-out item, and routes credit to lines that retrieve it. On Video Games, Recall@10 reaches 0.1195 versus 0.0856 for a matched no-retrieval control, with gains across three Amazon Reviews categories; credit is conditional on retriever hits.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.