AnalysisAI Models
2 days ago
The sample efficiency black hole
Dwarkesh Patel argues that progress on training sample efficiency has stagnated over the last few years despite scaling. The post questions whether current approaches are sufficient for achieving general intelligence.
2 days ago