Neural Template Repair

Download source code and datasets from Google One. 

Before you start, please install the necessary dependencies which will help to run NTR successfully in the following.

Note that since CodeLlama requires the latest version of Transformers, follow the commands below to install it:

1-Transformers

pip install git+https://github.com/huggingface/transformers.git@main

2-peft

conda install -c conda-forge peft

3-bitsandbytes

conda install -c conda-forge bitsandbytes 

4-accelerate

conda install -c conda-forge accelerate

5-sentencepiece

conda install -c conda-forge sentencepiece

Before conducting experiments, install the necessary test benchmarks (Defects4J, HumanEval-Java, Vul4J, and CBRepair) and initialize them.

1-Defects4J

Please visit the GitHub link for more installation details.

2-HumanEval-Java

Please visit the GitHub link for more installation details.

3-Vul4J

Please visit the GitHub link for more installation details.

4-CBRepair

Please visit the GitHub link for more installation details.

Quickly reproduce NTR and fine-tuned LLMs by running the following script files.

To promote open science, we have released StarCoder-15B and CodeLlama-70B, two LLMs that have been fine-tuned with the NTR framework and basic NMT fine-tuning strategy. We encourage future researchers to leverage these foundational models to develop more powerful APR tools. 

Table 1. Repair templates used for the Transfer/Recoder dataset

Note that we use the SC4FT tool to extract repair templates from the Recoder dataset. Similarly, the templates we extracted are consistent with previous works (e.g., TBar, TRANSFER, TENURE), which contained 15 general categories, and specific examples of repair templates can be found in the descriptions provided by TENURE.

Table 2. Repair templates used for the VulGen dataset

Note that we manually checked and summarized 14 general templates from the VulGen dataset.