Reproduce “DPE-Rec: Diversity-Preserving and Efficient Diffusion Language Model Recommendation for Bundle Construction” 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
Pingchuan Ma, Bohan Jiang, Chengshuai Zhao, Dawei Li, Zhen Tan, Huan Liu
DPE-Rec improves bundle completion by training diffusion language models to choose an item’s coarse semantic branch before its finer identifier tokens, then distilling denoising trajectories so a student can predict multiple tokens in parallel. Across five settings, DPE-Rec-Fast reports 25.6× average inference speedup over three generative baselines. It also reports 63% average relative quality improvement across nine baselines, five settings, and three metrics, not a single benchmark score.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.