Hacker News Reader

>what the training dataset is

All non-private repositories on GitHub.

Is it though? I'd assume that too, but we don't really know, do we? I mean to ask, what have Microsoft stated that leads you to believe this? (Maybe there's a press release I missed?)

https://twitter.com/NoraDotCodes/status/1412741339771461635

(this doesn't say they didn't use other sources too, though)

"source code from publicly available sources, including code in public repositories on GitHub." (from the copilot home page)

Pretty much says they did also use other sources.

Reply on news.ycombinator.com