Reproduce “Decode-Time Grammars: Constrained LLM Generation over a Refinement Order of Grammar Fragments” 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
Shuoming Zhang, Ruiyuan Xu, Haofeng Li, Qiuchu Yu, Yangyu Zhang, Chunwei Xia, Xiaobing Feng, Chenxi Wang and 2 more
The authors introduce an environment-indexed decoding policy that restricts generated code references to names available in the current runtime environment. It selects and specializes grammar fragments for each region, updating available names after declarations. On 60 Spider samples, its SQL mask achieved 100% execution match with no ghost columns; this does not guarantee full query correctness.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.