Dataset Open Access
Le, Kim Tuyen; Rashidi, Gabriel; Andrzejak, Artur
Code Token Type Taxonomy (CT3) is a methodology for refined evaluation of ML-based code completion approaches.
We published the CT3-enhanced dataset with pre-computed token types for each token in the Python150k dataset.
The dataset was obtained from an empirical study of the below paper:
Kim Tuyen Le, Gabriel Rashidi, and Artur Andrzejak. A Methodology for Refined Evaluation of ML-based Code Completion Approaches. In KDD Workshop on Programming Language Processing (PLP), August 14-18, 2021 (Virtual).
Please read the README.txt file for detailed information of structuring the enhanced dataset.