Reproduce “HiGenTrans: Hierarchical LLM Knowledge Transfer 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

Zhuohang Jiang, Yuxin Chen, Haohao Qu, ZeFei luo, Yunlong Wang, Xu Yuan, Haojie Ding, Boheng Zhang and 8 more

HiGenTrans improves next-item recommendation by transferring frozen language-model semantics into item-code generation and reranking generated candidates. It adds a history summary at the generator’s start, injects intermediate-layer features, and calibrates beam-search candidates using semantic, generation, and context signals. On Amazon Toys and Games, it reaches Recall@10 0.1047 and NDCG@10 0.0674; reranking cannot recover items missing from the candidate set.

Summary by alphaXiv.

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