Could you give more info about the dataset?

[baleksey]

First of all - thank you for the great dataset. I haven't tested it yet but your results looks great.
Can you answer several questions about this dataset?

1. How did you make it? ChatGpt model? And how you generated your instructions and prompts for it so I can do the same?
2. Have you checked it for wrong answers or impossible instructions (like url links in inputs or "describe the image" etc)?
3. Can you describe roughly what percent different kind of instructions present? (math:x%, qa:y% code:z% etc..)

Thank you!