Dataset for the ASE 2025 Challenge on Code Completion Context Collection Optimization

This dataset was used in the ASE 2025 Challenge on Code Completion Context Collection Optimization organized by JetBrains and Mistral AI. It includes the ground truth data and the associated repositories for all three phases: practice, public, and private. It also covers both tracks, Python and Kotlin. In addition, the dataset contains submitted solutions from the private phase as well as baseline solutions (BM25 and recent files) for further study. Overall: 102 repositories, 1,746 revisions, and 1,764 completion points Python: 47 points for Practice, 247 for Public, 394 for Private (total: 688 points from 52 repos) Kotlin: 30 points for Practice, 400 for Public, 646 for Private (total: 1,076 points from 50 repos)