Oxford's Toby Ord argues Anthropic's uniform scaling claim fails for non-verifiable tasks, which face a lower performance plateau · Digg
/Tech
4h ago
Oxford's Toby Ord argues Anthropic's uniform scaling claim fails for non-verifiable tasks, which face a lower performance plateau
-
RLVR improvements do not fully transfer to non-verifiable tasks.
0
19
1
0
1.1K
Original post unavailable.
Sentiment
Sentiment building, check back later.
Cluster Engagement
1.1K
Views
0
Comments
1
Reposts
0
Bookmarks
Expand data
Posts from X
Most Activity
Most Activity
Timeline
No ranked X posts are available for this story yet.
/Tech
4h ago
Oxford's Toby Ord argues Anthropic's uniform scaling claim fails for non-verifiable tasks, which face a lower performance plateau
-
RLVR improvements do not fully transfer to non-verifiable tasks.
0
19
1
0
1.1K
Original post unavailable.