Add high quality coding data

[rombodawg]

@jondurbin I know I previously asked to add a dataset, but as you stated the set was very large and comprised of many smaller sets. This dataset I am suggesting is extremely high quality and was used to train the very good OpenCodeInterpreter coding models.

• https://huggingface.co/datasets/Replete-AI/OpenCodeInterpreterData

If you dont want to use my version feel free to use the originals bellow:

• https://huggingface.co/datasets/m-a-p/Code-Feedback

and

• https://huggingface.co/datasets/m-a-p/CodeFeedback-Filtered-Instruction

[rombodawg]

I also made code_bagel

https://huggingface.co/datasets/Replete-AI/code_bagel
800b tokens of coding data. deduped and uncensored