Mastery: connect the pieces

1. Idea — Local models run on infrastructure you control. That changes data flow, ops, and hardware limits—not automatic superiority.…
2. Mechanism — Download/quantize weights, serve with a local runtime, apply same prompt/tool patterns, monitor CPU/GPU RAM, and handle updates yourself.…
3. Controls — model size, quantization, context length, GPU/CPU offload, update channel, disk encryption
4. Tradeoff — Locality helps privacy and offline use and shifts patching, capacity, and quality risk onto you. Stronger quantization saves RAM and can hurt medical nuance.…
5. Failures — Phantom privacy; Silent quality drop after quant
6. Metrics — PHI exfil test pass, note rubric vs cloud baseline, tokens/sec, crash rate, patch lag
7. Ship rule — Allow local 7B 4-bit only if gold suite severe misses≤baseline and no telemetry of note text is confirmed.

Neighboring layers (retrieval, serving, policy, human review) still own what local LLMs cannot on the offline clinical note assistant. Do not ask this chapter’s dial to replace permissions or source truth.

Mastery is transfer: take local LLMs into a second scenario that is not the offline clinical note assistant and rebuild the checklist without copying sentences. If you can only recite this chapter’s examples for local LLMs, you have memorized a story, not a model.

Also pin one numeric memory from this local LLMs chapter: A 7B 4-bit model may fit in ~5–6 GB RAM yet lose accuracy on rare clinical abbreviations—measure on your notes, not only tokens/sec. That number is not decoration; it is a template for how claims about local LLMs on the offline clinical note assistant should look in design docs. Scoped specifically to local LLMs / offline clinical note assistant / mastery-connect.

People confuse local LLMs with neighboring buzzwords when debugging the offline clinical note assistant. Before changing prompts, ask whether the broken stage was evidence gathering, the local LLMs judgment itself, validation, or the product action. Fixing the wrong stage creates folklore (“we tried local LLMs and it failed”) that blocks the next team on the offline clinical note assistant. Scoped specifically to local LLMs / offline clinical note assistant / mastery-connect.

Write a five-line artifact for this page: goal, inputs, check, owner, stop rule. Invent one fluent failure that the check would catch. Keep details specific to local llms rather than generic AI advice.