Maple is an open-source full-stack code dataset developed and released by Tudor Iustin.
It is designed to support code generation, web development, supervised fine-tuning, instruction tuning, post-training, dataset research, and evaluation workflows for code-capable AI systems.
Maple contains 16,000 full-stack code samples totaling approximately 102 million tokens. It focuses on realistic software-building tasks, including web applications, product interfaces… See the full description on the dataset page:
https://huggingface.co/datasets/tudor-iustin22/maple.
