1/31/2026
•
EN
Quoting Andrej Karpathy
Andrej Karpathy notes a 600x cost reduction in training a GPT-2 level model over 7 years, highlighting rapid efficiency gains in AI.