CREX is a tool to match semantically similar functions based on transfer learning.
We recommend conda to setup the environment and install the required packages.
First, create the conda environment,
conda create -n crex python=3.8 numpy scipy scikit-learn requests
and activate the conda environment:
conda activate crex
Then, install the latest PyTorch (assume you have GPU):
conda install pytorch torchvision torchaudio cudatoolkit=11.1 -c pytorch -c nvidia
Enter the crex root directory: e.g., path/to/crex, and install crex:
pip install --editable .
For large datasets install PyArrow:
pip install pyarrow
Create the checkpoints and checkpoints/similarity subdirectory in path/to/crex
mkdir checkpoints, mkdir checkpoints/similarity
Download pretrained weight parameters and put in checkpoints/pretrain
For full Micro-traces:
./tools/preprocess.sh
For function Micro-traces:
./tools/extract02.sh
For embeddings:
./command/inference/get_embedding.sh
We put our dataset here
dataset-funtion.zip
. ├── bugcode : buggy file ├── Coconut : APR patch file ├── GT : Ground-truth(correct) file
Files in dataset-funtion/xxx(Bug ID)/xxx(File type)/modify are the functional Micro-traces