Replication package for the paper Improving Deep Learning Library Testing with Machine Learning
Install Python 3.10
Download and uncompress the file specs4freedll.tar.gz. The file contains two folders:
specs4freedll: implementation of our training data generation strategies and the process for training the ML models.
ACETest-ml-predictor: extension of ACETest that uses the trained ML models to predict input validity.
Once uncompressed, complete the setup and installation following these instructions:
cd specs4freedll
python -m venv venv
python -m pip install -r requirements.txt
To generate the training data and train the ML models for a given function run the following command:
python specs4free.py addr Pytorch
More details can be found in the README.md file.Â
To replicate the experiments to obtain the results of our paper follow the instructions in the file experiments.md within our artifact.