Reproduce “OBLIQ-IR: Training a Dense Retriever for Oblique Queries” 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
Mahmoud Abdalla, Abdelrahman Abdallah, Shaimaa Sedek, Adam Jatowt
OBLIQ-IR trains a shared retriever to search for latent document attributes, using synthetic attribute-focused queries and, for writing style, nearest-neighbor pairs from a specialist authorship encoder. On Writing, its Gold NDCG@10 is 0.211, versus 0.096 without those pairs. Gains are significant on Writing and Congress but marginal on Math and Twitter; transfer to unseen tasks remains untested.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.