BriefGPT.xyz
Jul, 2017
生成神经语法分析的有效推论
Effective Inference for Generative Neural Parsing
HTML
PDF
Mitchell Stern, Daniel Fried, Dan Klein
TL;DR
通过改进候选选择和使用粗略的修剪功能，我们描述了一种可行的搜索过程，及其用于Choe和Charniak（2016）模型的推理过程在Penn Treebank的23节上能够获得92.56 F1，超过以前的最先进结果。
Abstract
generative neural models
have recently achieved state-of-the-art results for
constituency parsing
. However, without a feasible
search procedure
→