Reproduce “Towards Reliable, Scalable, and Value-Aligned Generative Recommendation for Large-Scale Advertising” 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
Yatong Xiao, Yibo Sheng, Lixiang Wang, Mingjie Sun, Yunhao Sha, Pengfei Zhang, Shengyu Wang, Peng Wang and 2 more
GR4AD-X improves advertising retrieval by giving ads multiple semantic-ID paths, expanding model capacity through complementary mechanisms, and using commercial-value estimates to train prefixes, guide beam selection, and rerank candidates. With UniVGD, full-level HR@10 rises from 0.021 to 0.298. Sequential online tests on Kuaishou report 3.07% higher revenue than deployed GR4AD; this measures the combined system.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.