factual
Original scaling laws paper was slightly wrong
The original OpenAI scaling laws paper was slightly wrong due to issues with learning rate schedules, and Chinchilla later showed a more correct compute-optimal version, after which practitioners deviated again to optimize for inference budget rather than pure compute-optimality.
factualpending
Speaker
Aman SangerEvidence Quote
“The original scaling laws paper by open AI was slightly wrong... And then Chinchilla showed a more correct version. And then from then people have again deviated from doing the compute optimal thing.”
Created: 6/13/2026, 3:31:02 AM
Your notes
Loading notes...