Scaling in Inference and RL Training Outperforms Pre-training
tobyordoxford · x · 2026-07-03
Toby Ord notes that by standard AI scaling law metrics, current scaling performance is quite solid: halving the error rate in pre-training requires about 1,000,000x the compute. However, on math benchmarks, both inference scaling and RL training deliver logarithmic improvements. This is the most informative post within the same AISecurityInst discussion.
More from Infra
- WSJ: Nvidia is in talks to backstop about $250 billion of OpenAI's data center plan — KateClarkTweets · 2026-07-27
- YC talk on BCI x AI says infrastructure is what really determines speed — garrytan · 2026-07-27
- A 13B model ran on a no-GPU PC by paging weights from SSD via llama.cpp — ID_R_McGregor · 2026-07-27
- llama.cpp warns that GGUFs made before a recent change must be regenerated — EconomySerious · 2026-07-27
- RTX 5090 local tests show Qwen Q6 can drop to 15 tok/s at 80k context — LFAdvice7984 · 2026-07-27
- Surprising Ubuntu Setup: NVIDIA 5090 PC Becomes the Easiest AI Rig — _xjdr · 2026-07-27