@hengyuf
PhD student of Computer Sciences at UC Berkeley
AI
The author says mechanistic interpretability shows why looping helps implicit reasoning but still struggles to generalize.
AI
X
