The dataset includes: 60 small Python programs generated by OpenAI GPT-3.5 (10 functionally correct program variations for each of the selected 6 publicly available introductory programming assignments) 399 program comprehension questions that target the generated programs 798 answers to those questions by OpenAI GPT-3.5 and GPT-4 Researcher annotation on the correctness of each answer and developed codes for the error types In addition, scripts for processing and analyzing the data are included.

2024