Onboarding

On the first launch of warpdrv, an onboarding flow walks you through the essential setup in a few quick steps. You can re-run it at any time from Settings → Re-run Onboarding.

Step 1: Welcome

A brief intro screen. Click Next to continue.

Step 2: Model Folders

Tell warpdrv where your GGUF models live. Models should follow a user/model folder structure (e.g. LiquidAI/LFM2.5-2.6B-GGUF).

1. Type a folder path, or click the folder icon to browse.
2. Click + to add the folder.
3. Click Save & Scan.

warpdrv scans the folder and reports how many models it found. If you don’t have any models yet, you can Skip this step and download some later from the Hub.

You can add multiple folders. warpdrv indexes all of them.

Step 3: Backends

A backend is a llama.cpp binary that runs your models. This step shows your detected hardware (OS, architecture, and any GPUs) and lists the available backends for your system.

1. Select the backend that matches your hardware:
   • CUDA for NVIDIA GPUs
   • ROCm for AMD GPUs
   • Vulkan for any GPU (Intel, AMD, NVIDIA)
   • CPU if you have no GPU (pre-selected by default)
2. Optionally check whisper.cpp backends (for voice dictation) and Kokoro TTS (for text-to-speech).
3. Click Install & continue.

If you’re not sure which backend to pick, see the hardware table in the Downloading & running your first AI model guide.

Step 4: Getting Started Guide

A short 3-slide carousel recaps the core workflow:

• Download Models from the Hub — search and download GGUF models from HuggingFace.
• Add a Backend — register llama.cpp builds so warpdrv can run them.
• Launch a Server — pick a model and backend, then start inference.

Click Next when you’re ready.

Step 5: All Set

You’re done. Click Start Using warpdrv to enter the main app.

Next steps

• Downloading & running your first AI model — get a model running and chatting.