Getting Started¶
This section walks you from a bare machine to a running fine-tuning session in MLX Training Studio. There are four phases:
| Phase | What happens |
|---|---|
| 1. Validate | Confirm your hardware and software meet the requirements. |
| 2. Install CLI | Add the mlx-training-studio command via Homebrew (or manually). |
| 3. Install app | The CLI clones upstream source, builds with Xcode, and places the .app. |
| 4. Configure & use | Set up the Python environment inside the app and start training. |
| 5. Stay current | Use mlx-training-studio update whenever you want the latest upstream. |
End-to-end flow¶
sequenceDiagram
autonumber
actor User
participant HB as Homebrew
participant CLI as mlx-training-studio
participant GH as GitHub (stevenatkin/mlx-lm-gui)
participant XC as Xcode / xcodebuild
participant Apps as /Applications
User->>HB: brew install jsgermanaai/tap/mlx-training-studio
HB-->>User: CLI installed
User->>CLI: mlx-training-studio doctor
CLI-->>User: preflight checks pass
User->>CLI: mlx-training-studio install
CLI->>GH: git clone stevenatkin/mlx-lm-gui
GH-->>CLI: source cloned
CLI->>XC: xcodebuild -configuration Release
XC-->>CLI: MLX GUI.app built
CLI->>Apps: cp -R MLX GUI.app (atomic via .tmp)
Apps-->>CLI: installed
CLI-->>User: manifest written · open "/Applications/MLX GUI.app" Next steps¶
-
Requirements
Check hardware, macOS, Xcode, and Python before you install anything.
-
Quick install
Three install methods — Homebrew, one-liner curl, or manual clone.
-
First run
Open the app, configure your Python environment, and run a smoke test.