This repository contains code for GAN training on the gSCAN dataset, and is based off of the code from the original gSCAN directory.

Make a virtualenvironment that uses Python 3.7 or higher:

virtualenv --python=/usr/bin/python3.7 <path/to/virtualenv>

Activate the environment and install the requirements with a package manager:

{ source <path/to/virtualenv>/bin/activate; python3.7 -m pip install -r requirements; }

Note that this repository depends on the grounded SCAN implementation to load the dataset from a dataset.txt with the function GroundedScan.load_dataset_from_file(). Before actually training models, unzip the data you want to use from this repo and put it in a folder data.

To train a model on a grounded SCAN dataset with a simple situation representation, run:

python3.7 -m seq2seq --mode=train --data_directory=<path/to/folder/with/dataset.txt/> --output_directory=<path/where/models/will/be/saved> --attention_type=bahdanau --max_training_iterations=200000

This will train a model and save the results in output_directory.

To test with a model and obtain predictions, run the following command.

python3.7 -m seq2seq --mode=test --data_directory=<path/to/folder/with/dataset.txt/> --attention_type=bahdanau --no_auxiliary_task --conditional_attention --output_directory=<path/containing/trained/models> --resume_from_file=adverb_k_1_run_3/model_best.pth.tar --splits=test,dev,visual,visual_easier,situational_1,situational_2,contextual,adverb_1,adverb_2 --output_file_name=predict.json --max_decoding_steps=120

NB: the output .json file generated by this can be passed to the error analysis or execute commands mode in the dataset generation repo (found here). The repository at that link also contains a file example_prediction.json with 1 data example prediction as generated with the test mode of this repository.