Reproduce “Try Before You Evolve: Ranking-Scored Trials for Recursive Self-Improvement of Recommender Systems” 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
Xu Huang, Jiawei Feng, Jiancan Wu
RecRSI uses short training trials to rank recommender recipe edits before committing to full retraining. It trains each valid edit from a shared initialization and scores full-catalog NDCG@10 on probe users. Across two backbones and three datasets, it kept 23 of 30 retrained edits, versus 37 of 90 for three existing agents; these results do not establish performance beyond the studied setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.