Chollet Describes Two Axes of Test-Time Scaling
Researcher notes breadth as key alongside depth for hard search problems.
François Chollet posted that test-time scaling has two axes. Depth means running agents over longer timeframes. Breadth means running a larger number of agents. He stated that most people know about the first axis, but the second is equally important for solving hard problems that require broad search. Chollet created the Keras deep learning library and the ARC-AGI benchmark. He co-founded ARC Prize and Ndea.
Combined views
13.3K
1 post, first seen 3h ago
Chollet Describes Two Axes of Test-Time Scaling
Researcher notes breadth as key alongside depth for hard search problems.
François Chollet posted that test-time scaling has two axes. Depth means running agents over longer timeframes. Breadth means running a larger number of agents. He stated that most people know about the first axis, but the second is equally important for solving hard problems that require broad search. Chollet created the Keras deep learning library and the ARC-AGI benchmark. He co-founded ARC Prize and Ndea.
Reactions
1 postFC 
François Chollet@fcholletTest-time scaling has two axes: running agents over longer timeframes (depth), and running a larger number of agents (breadth). Everybody knows about the first axis, but the second one is just as important when solving hard problems that require broad search.
- likes: 136
- replies: 36
- bookmarks: 43
- reposts: 12