How to Train Your Agent to Read and Write

This repository contains the code for the AAAI paper: "How to Train Your Agent to Read and Write".

This repository contains experimental software and is published for the sole purpose of giving additional background details on the respective publication.

This project is implemented using the framework OpenNMT-py and the library PyTorch Geometric. Please, refer to their websites for further details on the installation and dependencies.

Environments and Dependencies

python 3.6
PyTorch 1.1.0
PyTorch Geometric 1.3.1
subword-nmt 0.3.6

Datasets

In our experiments, we use the following datasets: AGENDA.

Preprocess

First, convert the dataset into the format required for the model.

For the AGENDA dataset, run:

./preprocess_AGENDA.sh <dataset_folder>

For the WebNLG dataset, run:

./preprocess_WEBNLG.sh <dataset_folder>

Training

For traning the model using the AGENDA dataset, execute:

./train_AGENDA.sh <graph_encoder> <gpu_id>

Options for <graph_encoder> is cge-lw.

Examples:

./train_AGENDA.sh 0 cge-lw

Decoding

For decoding, run:

./decode_AGENDA.sh <gpu_id> <model> <nodes_file> <graph_file> <output>

Example:

./decode_AGENDA.sh 0 model_agenda_cge_lw.pt test-nodes.txt test-graph.txt output-agenda-testset.txt

For more details regading hyperparameters, please refer to OpenNMT-py and PyTorch Geometric.