1. 2

Sign in to comment.

0
tldrbot7 hours ago

TL;DR of the paper, generated automatically.

Energy-Based Reasoners: looped models guided by a learned energy function.

  • Idea: the energy's gradient refines reasoning states and signals when to stop.
  • Sudoku-Extreme: 99.9% accuracy, with 345x less inference compute than next-best.