Part 1: LoRA With Your Eyes Open — mflux Dreambooth
Chapters 01–03: Learn what fine-tuning is for, train a visible image adapter, then sweep hyperparameters where failure modes are obvious.
TUTOR WITH THEFOCUS.AI
Copy this prompt into Claude, ChatGPT, or any external AI assistant. It points the assistant to the course instructions and links it to your student profile to track your progress and customize observations.
You are not enrolled yet. Enroll to generate a Student ID to track lesson completions and store learning notes.
Part 1: LoRA With Your Eyes Open
mflux Dreambooth adapter · ~1 day
How these chapters are taught: concept → concept check → install/smoke test → prepare artifacts (raw → 512) → train/sweep. Tutors stop at each gate; do not paste the whole module as one command list.
| Chapter | What You’ll Do |
|---|---|
| 01 · Why Fine-Tune | What fine-tuning changes, what it shouldn’t, and how adapters work at a high level |
| 02 · Train an Adapter | Dreambooth concept, mflux install + generate smoke test, raw→512 dataset, first train, what drives wall-clock tuning time |
| 03 · The Experiments | Knob intuition, then one-at-a-time sweeps into a contact sheet |
What You’ll Have After Part 1
- A clear sentence for what “success” means before any training run
- A working mflux training setup on Apple Silicon
- A trained Dreambooth LoRA of a subject you know well
- A contact sheet: one grid image, same prompt, every adapter variant labeled
- The ability to look at a generation and say “that’s overfit” or “that’s underbaked” without checking any numbers
This module is deliberately throwaway. The lasting artifact is judgment — and a contact sheet you’ll refer back to.