SPIGOT

C++ implementation for Backpropagating through Structured Argmax using a SPIGOT. More detailed documentation coming soon.

Required software

The following software needs to be installed:

• A C++ compiler supporting the [C++11 language standard] (https://en.wikipedia.org/wiki/C%2B%2B11)
• CMake (tested with version 3.6.2)
• Boost libraries (tested with version 1.61.0)
• Git LFS (optional)

Other dependencies including Eigen, DyNet, gflags, glog, and AD3, and included in this repo to keep it self-contained.

Checking out the project for the first time

git clone https://github.com/Noahs-ARK/SPIGOT
cd SPIGOT
./install_deps.sh

To compile the parser

mkdir -p SyntacticSemanticParsing/build
cd SyntacticSemanticParsing/build
cmake ..; make -j4
cd ../..

To fetch the GloVe embeddings and pretrained pruners:

git lfs fetch
git lfs checkout

Or you can simply download the files and put them to the corresponding places.

Data

The data format follows that by NeurboParser. Several training samples are included under 'data/'.

Training/running

You can use the scripts in './SyntacticSemanticParsing' to train/evaluate the parser.