Reproduce “Recurrent Relational Networks” 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
Rasmus Berg Palm, Ulrich Paquet, Ole Winther
The recurrent relational network (RRN) models relational reasoning as repeated learned message-passing updates among interacting objects. For Sudoku, cells exchange messages through row, column, and box connections, updating hidden states over multiple steps. On the authors’ generated 17-given test puzzles, it solves 96.6% after 64 steps, versus 0% for a matched one-step baseline; it was trained for 32 steps.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.