Skip to content

README.md

Latest commit

 

History

History
26 lines (20 loc) · 1.32 KB

README.md

File metadata and controls

26 lines (20 loc) · 1.32 KB
Raw

LoopEval docs

Guides for the two main workflows and the simulator internals. Everything here renders on GitHub — no build step, no server.

  • Simulator guide — how the closed-loop counterfactual simulator works: the substrate, the physiological counterfactual, the sensitivity-fidelity model, disruption handling, candidate levers, and scoring. The deep reference.
  • Frontier experiments — "does this change actually help?" The lift method: sweep a candidate against the insulin-needs reference and score TIR vs t<54.
  • Case studies — "is the simulator (and the candidate) behaving correctly?" Multi-panel plots of one window to root-cause a surprising outcome number.
  • Loop algorithm classes — deployed-Loop vs LoopAlgorithm-package behavior classes and the emulation flags that reproduce each.

Local preview

These render on GitHub as-is. To preview locally (or share on your LAN) with the same GitHub rendering — Mermaid diagrams, alerts, images — run:

python3 scripts/docserve.py        # http://localhost:8080/docs/  (+ the LAN URL it prints)

Stdlib only; Markdown is rendered in-browser via marked + mermaid from a CDN (the viewing machine needs internet), everything else is served as static files.