Tag
The paper introduces Coffee, a plug-and-play framework that guides discrete diffusion models at inference time using compiled finite-state objectives, avoiding exponential enumeration of token completions while supporting hard constraints and learned soft objectives across symbolic, language, and biological benchmarks.
The paper introduces a minimal-core-guided repair method for neuro-symbolic constraint solving, where language models use proofs from unsatisfiable cores to correct translation errors, reducing fabrication in solutions.
A blog post explaining how building Brio train track layouts with a toddler inspired the author to explore constraint solving and backtracking search algorithms.