The Inference Budget: How Spending Compute After the Prompt Changes Everything
中文摘要
AI 扩展定律正从参数规模转向推理时的计算预算，这彻底改变了模型性能优化的逻辑。
English Summary
AI scaling laws are shifting from parameters and tokens to inference-time compute, fundamentally changing how models are optimized.
原文节选
For a decade, scaling laws were measured in parameters and tokens. Continue reading on Medium »