And in this example below, you can see another problem. Does Laura Bates know about this? Privacy issues!! #gdpr https://twitter.com/GaryMarcus/status/1188150000364032000 …pic.twitter.com/JGphiCHqzg
may actually be a fictional person; i don't think there are game developers at Twitter? but GPT-2 as implemented here completely indiffferent to differences between fact and fiction and probably does (in an extended way) do something like cut and paste from its (public) databases
-
-
Yes, there's no reason to think that GPT-2 doesn't encode literal copies of portions of text that are either identifying or copywritten, right?.
-
There was a discussion here [1]. It does memorize some stuff which is unavoidable/needed but copyright issues loom because of the transformer's insatiable appetite for data. [1] https://old.reddit.com/r/MachineLearning/comments/dfky70/discussion_exfiltrating_copyright_notices_news/ …
- 1 more reply
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.