Modal trains RL models on mixed-accelerator clusters, matching homogeneous setups at lower cost
charles_irl · x · 2026-10-14
Modal challenges the norm that RL training requires a homogeneous cluster of the latest accelerators, calling big homogeneous clusters "a sledgehammer for an inference-shaped nail." Using fully asynchronous RL with mixed-precision, mixed-accelerator samplers plus off-policy correction on agentic coding tasks, training stayed stable through convergence — on par with homogeneous samplers at lower cost. A reason to rethink RL compute procurement.
More from Infra
- Only 7% of frontier lab compute to pretraining by Q4 2026, but the viral chart is being misread — repligate · 2026-10-14
- AI agents hit monorepo scale: you need Google-style copy-on-write primitives, not Git — rickasaurus · 2026-10-14
- A backyard "distributed data center" ADU startup for local LLM enthusiasts — dee_hw · 2026-10-14
- Dots Studio's DoSA speeds sparse attention indexer up to 2.74x over SGLang, open-sourced — dots-studio · 2026-10-14
- ShellElastic trains one supernet yielding all sizes, with provable hyperparameter transfer — Can Jin · 2026-10-14
- Report: Microsoft strains under data center spending and AI talent war, mulls cost cuts — SumitGup · 2026-10-14