Reproduce “TRANX: A Transition-based Neural Abstract Syntax Parser for Semantic Parsing and Code Generation” 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
Pengcheng Yin, Graham Neubig
TRANX provides a reusable transition-based parser for translating natural-language requests into Python, SQL, and structured outputs. It uses a supplied ASDL grammar to construct an abstract syntax tree, then converts that tree into the target representation. On DJANGO, TRANX with parent feeding reports 73.7 ACC., versus 71.6 for YN17 and 72.7 without parent feeding. On three of four datasets, it outperforms existing neural approaches; WikiSQL results depend on table access.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.