Compute allocation recipe: 32 tuning runs at smallest scale, 68.7% budget at largest
karen_ullrich · x · 2026-10-29
Thread sub-post with the actionable recipe: small models are cheap to train, so allocate compute across sizes—run 32 tuning configs at the cheapest scale (vs 1 at the next), but spend 68.7% of the compute budget on the largest scale and only 8.6% on the smallest. Also shows endpoint errors move predictions far more than mid-curve errors. Full recommendations in the paper.
Related event: Meta/NYU study shows scaling laws hold down to 4M-parameter models(5 posts)→
More from Infra
- Why Macs lose to DGX Spark at batched inference: bandwidth wins solo, FP4 compute wins crowds — QuixiAI · 2026-10-29
- ComfyUI reproducibility: the workflow JSON alone isn't enough — Im_bored_rn69 · 2026-10-29
- Akamai launches Akamai Inference and Agent Sandbox for enterprise AI deployment — lucaolivari · 2026-10-29
- 512GB M5 Ultra Priced at $22,300, Earliest Pickup December 3 — rez0__ · 2026-10-29
- Lumentum CEO says copper-to-optics upgrades will drive growth even if AI buildout cools — pstAsiatech · 2026-10-29
- Training speedrun record smashed: 30M model in 22.9s, 21.8x faster — RobertTLange · 2026-10-29