Reproduce “pplx-embed-v2-late Technical Report” 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
Markus Krimmel, Louis Milliken, Ekaterina Glazkova, Maximilian Schall, Antoine Chaffin, Sedigheh Eslami, Jonas Golde, Denis Bykov and 3 more
The authors train two late-interaction retrieval models to share a token-level embedding space, allowing a smaller model to encode live queries against an index built by a larger model. On six text domains, the hybrid averages 79.6% nDCG@10, versus 78.0% with the smaller model on both sides; this adds index-time work, whose total cost is unquantified.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.