Adjusted RACE Framework

This framework is based on: https://github.com/DeepSoftwareAnalytics/RACE

Our research paper can be found here.

Author:

• Peiyi Jiang ([email protected])
• Haoyi Zhang
• Jiahao Chen ([email protected])

Directions

Our experiments are on separate branches. Please checkout to the correcponding branch and follow the instructions there.

Model(C+G): model-direction

AST: ast-direction

Dataset: dataset-direction

We also create corresponding demo/deployment branch for each direction.

Demo-Model(C+G): demo-model

Demo-AST: demo-ast

Demo-Dataset: demo-go

Changes

1. Add conditions to determine if computer support cuda in the ECMG model. This will enable the model to be tested on a machine that doesn't have NVIDIA GPU
2. Decrease the batch number to solve the out of memory error on the Machines with inadequate GPU performance

Instructions

Downloading Dataset

wget https://zenodo.org/record/7196966/files/dataset.tar.gz
tar zxvf dataset.tar.gz

dataset folder must be at the same level as the run script

1. Running on ARM64 MAC machine

# Create virtual env and activate
conda create --name py38 python=3.8
conda activate py38

# Install Rust compiler if you don't have it
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

# Install packages
pip install -r requirements.txt

# Training
language=java
bash run.sh $language

Run small result:

2. Running on colab free tire

# Open GPU in the Runtime

# Upload dataset to your google drivee

# Refer the steps in run_colab.ipynb and examples in 
examples/run_colab.ipynb