Reproduce “RecoReward: Recommender-Guided Multimodal Description Generation for 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
Guohong Mu, Yueyang Liu, Jiangxia Cao, Changxin Lao, Zijie Zhuang, Yuhui Zhang, Jiaqi Feng, Ruochen Yang and 3 more
RecoReward trains a live-stream description model with recommender feedback from target and non-target users, while keeping user data out of generation at serving. It produces one shared description from stream frames and speech. Offline evaluation shows RecoReward-9B raises NDCG@128 from 0.013166 to 0.018199 versus Qwen3.5-9B, with gains across seven recall metrics; a one-week online test also reports gains on one platform.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.