PropTest: Automatic Property Testing for Improved Visual Programming

This is the code for the paper PropTest: Automatic Property Testing for Improved Visual Programming

Paper (EMNLP 2024 Findings) | Project Page

Environmnet

Clone recursively:

git clone --recurse-submodules https://github.com/uvavision/PropTest.git

After cloning:

cd PropTest
export PATH=/usr/local/cuda/bin:$PATH
bash setup.sh  # This may take a while. Make sure the vipergpt environment is active
cd GLIP
python setup.py clean --all build develop --user
cd ..
echo YOUR_OPENAI_API_KEY_HERE > api.key

This code was built on top of ViperGPT. We follow the same installation steps as ViperGPT. For detailed installation, please refer to the ViperGPT repository.

You need to download two pretrained models and store it in ./pretrained_models. You can use download_models.sh to download the models.

Running the Code

The code can be run using the following command:

CONFIG_NAMES=your_config_name python main_batch.py

CONFIG_NAMES is an environment variable that specifies the configuration files to use.

Citation

Please cite our paper if you find our method or code useful:

@article{koo2024proptest,
      title={PropTest: Automatic Property Testing for Improved Visual Programming}, 
      author={Jaywon Koo and Ziyan Yang and Paola Cascante-Bonilla and Baishakhi Ray and Vicente Ordonez},
      journal={arXiv preprint arXiv:2403.16921},
      year={2024}
}