undefined | Better HN

0 pointsvlovich1233y ago0 comments

Can someone explain to me Microsoft’s decision here to use GPL code in the training set? It would seem like sticking to non-attribution / non-viral licenses would have kept them in the clear. Was that an insufficient size data set?

0 comments

az2263y ago

It only trains on the GPL code, it doesn't reproduce entire code files verbatim. So it's fair use.

vlovich123OP3y ago

Except when it does

j / k navigate · click thread line to collapse

0 pointsvlovich1233y ago0 comments

Can someone explain to me Microsoft’s decision here to use GPL code in the training set? It would seem like sticking to non-attribution / non-viral licenses would have kept them in the clear. Was that an insufficient size data set?

0 comments

az2263y ago

It only trains on the GPL code, it doesn't reproduce entire code files verbatim. So it's fair use.

vlovich123OP3y ago

Except when it does

j / k navigate · click thread line to collapse