Gordon Wetzstein: coding agent in the loop boosts optical NN accuracy 35% to 96.8%
GordonWetzstein · x · 2026-11-20
Gordon Wetzstein's group introduces "Optics as code," a new paradigm in physical AI: a coding agent with the optical hardware in the loop rapidly tunes the optics.
- Problem: optical neural networks are notoriously hard to calibrate and suffer a sim2real gap
- Method: a coding agent iterates directly against the physical hardware
- Result: image classification accuracy jumps from 35% to 96.8%
Presented as a 6-tweet thread; a notable advance for optical computing in physical AI.
More from coding & agent
- Your old CLAUDE.md is hurting new Claude models; author trimmed it to 36 rules-free lines — PrajwalTomar_ · 2026-11-20
- Open-sourced Claude AI team: PM, CTO, UX roles that challenge your product ideas — marilynika · 2026-11-20
- Open-source Intern harness: Opus 5.5 hands off to Hy4 Flash, 92.4% at 1/3 the cost — m4rkGS · 2026-11-20
- Dev shares MCP lessons: human-only publishing enforced by omitting the tool entirely — mariobalca · 2026-11-20
- Optics as Code: Stanford agent tunes optical neural nets from 35% to 96.8% via hardware-in-loop experiments — GordonWetzstein · 2026-11-20
- Post-acquisition: Cognition's Poke agent gets its own computer and a chat interface — marvinvonhagen · 2026-11-20