RSS

Raffaello Fornasiere

Karma: 49

Ac­ti­va­tion Or­a­cles sig­nifi­cantly un­der­perform with­out a safe base model

28 Aug 2026 1:39 UTC
20 points
0 comments9 min readLW link

The Model Or­ganism Lot­tery: Model Or­ganism In­ter­pretabil­ity Strongly Depends on Train­ing Methodology

23 Jul 2026 22:37 UTC
45 points
0 comments6 min readLW link
(arxiv.org)