Practical walk-through for someone landing on the platform without
context: pick a base model, prepare data, run a fine-tune, test the
result. Grounded in what the platform actually does today.
Covers the four main pages (/models, /datasets/upload, /jobs/create,
/jobs/view) and flags current limitations (NFP-52 chat-test gap for
platform-trained models, etc.). Includes an iteration table mapping
common failure modes to fixes, and a "first-week mistakes" list.
Linked from README alongside the developer MANUAL.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
