Code Similarity Visualization with GraphCodeBERT

This interactive application visualizes token-level embeddings generated by GraphCodeBERT for classical sorting algorithms. It supports pairwise comparison of algorithms based on their representation in the model’s embedding space, using PCA for dimensionality reduction.

✒️ Reference

Martinez-Gil, J. (2025).
Augmenting the Interpretability of GraphCodeBERT for Code Similarity Tasks.
International Journal of Software Engineering and Knowledge Engineering, 35(05), 657–678.

🚀 Features

Selection of two classical sorting algorithms.
Automatic tokenization and embedding via GraphCodeBERT.
PCA-based projection into 2D space for visualization.
Clean, static matplotlib plots showing token overlap and divergence.

🧠 Technical Overview

Model: microsoft/graphcodebert-base
Tokenizer: RobertaTokenizer
Embeddings: Last hidden layer of GraphCodeBERT
Reduction Technique: Principal Component Analysis (PCA)
Interface: Gradio
Languages: Python 3.10+

🔬 Research Context

This tool supports research on code similarity, clone detection, and representation learning for source code. It offers insight into how GraphCodeBERT encodes common algorithmic patterns, providing a visual companion to embedding-based analysis.

🛠 Dependencies

All required libraries are listed in requirements.txt:


transformers
torch
scikit-learn
numpy
matplotlib
gradio
Pillow

🖥️ Intended Use

Academic teaching and demonstration of code embeddings
Qualitative evaluation of pretrained models for source code
Supplementary visualization for software engineering publications

📬 Contact

Jorge Martinez-Gil
Senior Research Scientist in Computer Science