Datasets:
Mining The Stack vs Synthetically Generating Dataset
#4
by
TylerHilbert
- opened
This is a great dataset!
I was wondering why you decided to mine The Stack instead of synthetically generating the code? Any insight / reflection on the dataset creation would be greatly appreciated!
I'm looking to do something similar, just with OpenCL instead of Triton. I was wondering if you have any guidance to offer?
Hi! It was for legal reasons (everything on the stack is permissibly licensed).
Generally if you’re trying to create a dataset that is free for folks to use without legal trouble I’d use something like the stack or something similar. If it’s for personal things, feel free to use whatever fits your needs.