cstghitpku/egn3ldg

EGN3LDG

A lightweight neural network library based on dynamic graph for natural language processing

• The library supports only cpu, where tensors are implemented based on eigen.
• The library is fast if only cpu is available. (eigen can be speeded up by MKL)
• To use this library, just include the directory in your code and call it by "#include N3LDG.h"

Installation:

Download and include the directory

Prerequisitions:

EIGEN

You can get EIGEN from http://eigen.tuxfamily.org/index.php?title=Main_Page

If you have any problem, please send an email to [email protected]

Examples:

Some examples are realeased at:

• https://github.com/zhangmeishan/N3LDGClassifier
• https://github.com/zhangmeishan/NNTranJSTagger

Histories:

• LibN3L:A Lightweight Package for Neural NLP. (Zhang et al., LREC 2016); initial
• LibN3L-2.0, auto-differentiation
• EGN3LDG, dynamic batching, much faster (if limited gpu resource like me, a good choice)
• N3LDG (enable gpu), in progress, just for interest since dynet exists.

Authors:

Zhang Meishan, Yu Nan