Julius AI's Rahul Sonwalkar questions if legacy models wrote better, but researcher Aidan McLaughlin argues they just plagiarized
Early base models often directly reproduced their training data.
--0--
Sentiment
Sentiment unavailable for this story.
Cluster Engagement
-
Views
-
Comments
-
Reposts
-
Bookmarks
Expand data
Posts from X
Most Activity
Most ActivityTimeline
VIEWS757LIKES22RETWEETS1REPLIES3
Aidan McLaughlin@aidan_mclau
@0interestrates i think we forget that older and closer-to-base-model writing is often just plagiarism
rahul@0interestrates
do you think the older base models (davinci and older) were actually better at creative writing or our slopradars weren’t as tuned
1hViews 757Likes 22Bookmarks 1
BOOKMARKS2
🎭@deepfates
@aidan_mclau @0interestrates I was thinking the other day how the models have definitely read all of the books I'm asking them about but they are not allowed to reproduce any text from it so they just basically don't mention it at all. This seems suboptimal. to some degree the persona must be incoherent
Aidan McLaughlin@aidan_mclau
@0interestrates i think we forget that older and closer-to-base-model writing is often just plagiarism
33mViews 133Likes 5Bookmarks 2