I use GPT-3.5-turbo's API to generate a Chinese-Python educational dataset as Magicoder did with the lack of coding data in Chinese. The license also follows Magicoder-OSS-Instruct's mit.
One should notice:
the quality is not promised and has not been tested.
this dataset is mainly in Python since the seeds are Python only, but checking the results shows there are also other languages like PHP, Java, Ruby, etc. Substrings like "```{lang}" may be useful for filtering or categorizing.
