Dataset and Model Artifacts for "From Execution to Embedding: Enriching Code Representations with Data Difference Signals for Comment Generation"

This Zenodo record contains the dataset, trained model checkpoints, and the data-collection pipeline associated with the paper: Giacomo Fantino, Antonio Vetrò, Marco Torchiano, and Federica Cappelluti. 2026. From Execution to Embedding: Enriching Code Representations with Data Difference Signals for Comment Generation. In 2026 IEEE/ACM 48th International Conference on Software Engineering (ICSE-NIER ’26), April 12–18, 2026, Rio de Janeiro, Brazil. https://doi.org/10.1145/3786582.3786826 Contents of this repository:- Dataset of executed Python notebook cells with data-difference annotations- Pretrained and fine-tuned model checkpoints- Pipeline to reproduce the data collection process Source code, training scripts, and full usage instructions are available at:https://github.com/giacomofantino/execution-to-embedding Artifact package:The replicability artifact (scripts and instructions to reproduce the paper’s results) is available on Zenodo at: https://zenodo.org/records/18233481