2026-09-18 03:01 UTC
holy shit lmao:
«OpenAI's VP of Research explained: "We train our networks to memorize the training data — that's their objective."»
:neofox_facepalm:
Replies (2)
-
@SnoopJ@hachyderm.io 2026-09-18 03:13
【by 2021, OpenAI considered the prevention of memorization important "for fair use [compliance] and minimizing copyright violations in model output.】
-
@cthos@mastodon.cthos.dev 2026-09-18 03:02
@SnoopJ@hachyderm.io Hang on a second! I was told this was "overfitting" and "would be corrected" (Sarcasm, of course, but good gods)