Modal Details How to Run RL Samplers on Mixed-Precision, Mixed-Accelerator Hardware
lmoroney · x · 2026-10-15
Modal researcher Nan Jiang's blog 'Toward RL on mixed-precision, mixed-accelerator samplers' explores decoupling samplers from the trainer: since sampling takes 80–85% of GPUs in large RL runs, they run samplers on six accelerator types with their own kernels and precisions (FP8, MXFP4, NVFP4) feeding one BF16 B200 trainer, addressing the learning and stability problems introduced by training–inference mismatch.
More from Infra
- a16z chart: old GPUs are appreciating assets as B200 rental hits $5.69/hour amid compute scarcity — a16z · 2026-10-15
- 1GW of Compute Matches the Gross Profit of 1M Cybercabs — SpaceX to Add 3-5GW Next Year — JOBhakdi · 2026-10-15
- Cloudflare data: Yandex outage didn't kill malicious bots, they just moved to other Russian hosts — Cloudflare Blog · 2026-10-15
- AI cost falling 54x faster than electricity ever did, five years in — LinusEkenstam · 2026-10-15
- Inference has no yardstick: 4.5-quadrillion token run rate, but no proof of what buyers get — AccBalanced · 2026-10-15
- Booming business, falling stocks: scarcity caps ASML and Applied Materials upside — TiernanRayTech · 2026-10-15