README.md

metadata

license: apache-2.0
datasets:
  - AISE-TUDelft/Capybara
tags:
  - code

BinT5

• Repository: https://github.com/AISE-TUDelft/Capybara-BinT5
• Paper: https://huggingface.co/papers/2301.01701
• Point of Contact: https://huggingface.co/aalkaswan
• Raw Data: https://zenodo.org/records/7229913

BinT5 is a Binary Code Summarization model, the base models are CodeT5 and fine-tuned with Capybara.

We offer 5 variations of the model:

Name	Training Data
BinT5-C	C Source
BinT5-Decom	Decompiled C Binaries
BinT5-Stripped	Stripped Decompiled C Binaries
BinT5-Demi	Demi-stripped Decompiled C Binaries
BinT5-NoFunName	Decompiled C Binaries with the Function Name removed

Citation Information

@inproceedings{alkaswan2023extending,
  title={Extending Source Code Pre-Trained Language Models to Summarise Decompiled Binaries},
  author={Al-Kaswan, Ali and Ahmed, Toufique and Izadi, Maliheh and Sawant, Anand Ashok and Devanbu, Premkumar and van Deursen, Arie},
  booktitle={2023 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER)},
  pages={260--271},
  year={2023},
  organization={IEEE}
}