Skip to content
posttrainllm docs
Esc
navigateopen⌘Jpreview
On this page

posttrainllm docs

Mac-local LLM specialist factory — training, inference, evals, systems notes, and learning paths.

posttrainllm documentation

posttrainllm is a Mac-local specialist factory: target → data → post-training → eval → package → report. These docs are the committed source of truth; this site is the presentation + search layer over them.

Prefer the markdown mirrors (append .md to any URL) and llms.txt for agents. The full corpus is also at /llms-full.txt after build.

Start here

Surface Use it for
/factory Run schema, eval protocol, packaging, public artifacts, enforcement
/techniques Method-vs-recipe registry and target-specific technique backlogs
/attempt-ledger What worked, failed, regressed, or remains untried
/external-products-reviewed Products, papers, startups reviewed and what we stole or rejected
/recipes Closed-loop recipes (traces → SFT → specialist, distillation, eval-gate)
/parked Paused lanes and why they are paused
/doc-status Status label (active/evidence/reference/learning/parked/superseded/archive) for every major doc

For the maintainer golden path through the docs (the order a new reader should read them in), see README.md. When docs disagree, the conflict rule lives in doc-status.md.

Was this page helpful?