large language model

We collected programming problems and their solutions from previous studies. After applying some pre-processing steps, we queried advanced LLMs, such as GPT4, with the collected problems to produce machine-generated codes, while the original solutions were labeled as human-written codes. Finally, the entire collected dataset was divided into training, validation, and test sets, ensuring that there is no overlap among these sets, meaning no solutions in two different sets that solve the same programming problem.