The Tag-AI Setup Wizard guides you through the initial configuration process to ensure all components are properly installed and configured.
The wizard runs automatically on first launch and handles five essential setup steps.
The first step detects your GPU and installs appropriate drivers and libraries for GPU acceleration.
For NVIDIA GPUs, the wizard installs CUDA 12.4, which enables AI acceleration. The CUDA installer will launch in a separate window and requires administrator privileges.
For AMD GPUs, the wizard installs ROCm 6.1+. This process varies by operating system and may require following on-screen instructions.
If you don't have a compatible GPU or prefer not to set up GPU acceleration, you can click "Skip GPU Setup and Continue" to proceed in CPU-only mode. Note that processing will be significantly slower without GPU acceleration.
GPU acceleration provides 5-10x faster performance. CPU-only mode is functional but much slower.
The wizard checks for and installs Ollama, which is required for local AI image processing.
Ollama installation is handled automatically for most operating systems. On Windows, an installer window may appear that requires your confirmation.
After Ollama is installed, the wizard downloads and sets up the LLaVA model, which provides AI vision capabilities.
The LLaVA model is approximately 4GB in size. Download time depends on your internet connection speed.
The wizard shows real-time progress during the model download and installation.
The wizard creates and initializes the SQLite database that Tag-AI uses to store image metadata and tags.
This includes:
In the final step, the wizard saves your configuration and marks the setup as complete.
Initial default settings include:
All these settings can be adjusted later from the Configuration Editor.
If GPU installation fails, you can choose to retry or skip to CPU-only mode. Ensure your GPU drivers are up to date.
If Ollama installation fails:
If the LLaVA model fails to download: