Reproduce “DRanker: A Transformer-Based Multi-Task Ranking Model for the Disney+ Homepage” 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
Priya Nirmal Singh Khokher, Daniel Nemirovsky, Benjamin P. Whitesell, Ethan Sukrae Lee, Eric Wulf, Tom Chee-Duffy, Nicholas Flores
DRanker unifies Disney+’s homepage ranking with a model that uses recent viewing history and session context to order retrieved titles. A transformer represents viewing sequences while an auxiliary tower supplies learned embeddings. A synthetic test shows recommendations changing with session context. Production tests report +1.4% engagement and +0.5% daily active users; offline, the tower lowers AP@10 but lowers Top-50 Overlap.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.