Reproduce “Towards Foundation Decision Models: Drex” 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
Nischay Dhankhar, Alisher Amantay, Zhanibek Datbayev, Darkhan Akhmetov, Yash Shah, Nurmukhamed Ubaidullayev, Raman Saparkhan, Dos Baha and 1 more
Drex returns probabilities over user-supplied options for bounded software decisions, avoiding generated prose that must be parsed. Its pointer models cache shared state, while Max re-encodes each question with full bidirectional context. On the public Decision Index 0.4.1, Max scores 67.52 versus Jev 1.14.0 at 60.70; this weighted benchmark comparison does not isolate causes or establish universal decision accuracy.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.