Stanford paper: coding agents train optical neural networks in situ from measurements alone
StanfordUniversity · hf · 2026-11-21
Stanford and EPFL researchers introduce "optics as code": a coding agent trains an optical neural network from hardware measurements alone by reparametrizing it with interpretable optical elements. It beats gradient-free methods on classification and image encoding at lower cost, approaching backprop through a calibrated model.
More from coding & agent
- MEGA AI Engineering Course Wraps First Cohort with 2,000+ Developers Enrolled — johnlindquist · 2026-11-21
- OuterSSH open-sourced: SSH tunneling to run graphical agent apps from your phone — savvyRL · 2026-11-21
- You hit the kill switch but subagents keep running: Anaconda's agent incident tabletop — anacondainc · 2026-11-21
- Luke Wroblewski: why open an app to guide agents? Just use what's on screen — LukeW · 2026-11-21
- Gatana adds support for skills-over-MCP (SEP-2640) — Gatana_Official · 2026-11-21
- Team built 25 internal apps to talk to agents instead of dictating orders — willcb · 2026-11-21