PyTorch finally ships stateless RNG API after years of userland workarounds
ezyang · x · 2026-11-17
PyTorch core dev Edward Yang reveals the long-standing RNG reproducibility problem was never fixed upstream, leaving users with bodgy userland workarounds. Inspired by giffmana's push for stateless RNG and JAX's API design, jbschlosser finally implemented a stateless RNG API.
The thread revisits why earlier fixes stalled: they required carefully reconstructing the exact sampling process, and even a one-time BC break would still ideally require user code changes.
More from coding & agent
- Dev runs Grok bots from his car via Tailscale, while rival touts $200 Claude 3-agent setup — haydendevs · 2026-11-17
- YC launch CoverAgents turns broker emails into carrier quotes in 3 minutes, 90+ agencies live — ycombinator · 2026-11-17
- AI agent turned 18 growth ideas into 4 winning tests, best lift 26% — aryanXmahajan · 2026-11-17
- Open-source REQL builds code graphs to cut coding agents' token waste — cl0wnfire · 2026-11-17
- Using Figma MCP to let Claude create designs and prototypes straight into Figma — Tegadesigns · 2026-11-17
- Arkenos v0.6.12 ships set-and-forget autonomous research agents with free tier — _AustinCalvert_ · 2026-11-17