Skip to content

Getting Started

Jump to bottom
The Duskfall Portal Crew edited this page Mar 3, 2026 · 4 revisions

Getting Started

⚠️ Alpha Status — March 2026
Cross-platform AI Model training interface (Next.js + Python)

This guide walks you through setting up the Ktiseos Nyx Trainer locally or on a cloud GPU instance.

1. Prerequisites

Core Requirements (All Users)

  • Python 3.10+ (Recommended: 3.10.12)
  • Node.js 18+ (LTS recommended)
  • Git
  • GPU Drivers: CUDA-compatible drivers installed (if running locally)

Optional (Cloud Deployment)

  • Cloud Account: VAST.ai, RunPod, Lambda Labs, etc. (if not running locally)
  • SSH Client: Configured for remote instance access
  • VAST.ai Account: Only required if using VAST.ai specifically

2. Installation

Clone the Repository

git clone https://github.com/Ktiseos-Nyx/Ktiseos-Nyx-Trainer.git
cd Ktiseos-Nyx-Trainer

Backend Setup (Python)

  1. Create a virtual environment: 
    python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
  2. Install dependencies: 
    pip install -r requirements.txt
  3. Path Configuration: Ensure the custom_scheduler directory is added to PYTHONPATH if using advanced optimizers like CAME (usually handled automatically, but verify if ModuleNotFoundError occurs).

Frontend Setup (Next.js)

  1. Install dependencies: 
    npm install
  2. Create environment file: 
    cp .env.example .env.local
  3. Configure variables (see 02-Configuration#Environment-Variables).

3. Deployment Modes

Option A: Local Deployment

Best for: Developers with local GPUs, debugging, testing.

  1. Start the backend API: 
    python main.py
  2. Start the frontend dev server: 
    npm run dev
  3. Open http://localhost:3000.

Option B: Cloud Deployment (VAST.ai, RunPod, etc.)

Best for: Users without local GPU hardware.

  1. Rent an Instance: Choose a GPU with sufficient VRAM for your model size.
  2. Deploy Code: Clone the repository onto the instance.
  3. Expose Ports: Ensure port 3000 (frontend) and backend ports are accessible via HTTP/HTTPS.
  4. Access UI: Navigate to the instance's public IP/url.

💡 Note: While this project is tested on VAST.ai, it is compatible with any cloud GPU provider that supports Python 3.10+ and Node.js 18+.

4. Your First Training Run

  1. Navigate to the Form: The default view is the training configuration form.
  2. Fill Required Fields:
    • Model Path: HuggingFace path (e.g., runwayml/stable-diffusion-v1-5) or local path.
    • Dataset Path: Path to your training dataset (local or mounted cloud storage).
    • Optimizer: Select AdamW8bit for standard runs (see 02-Configuration#Optimizers).
  3. Save Config: Click "Save" to persist settings to localStorage (prevents data loss on refresh).
  4. Start Training: Click "Start Training".
  5. Monitor: Watch the console output or log files for progress.

5. Verification

To confirm everything is working:

  • ✅ No Pydantic validation errors in the UI alert box.
  • ✅ Training logs appear in the console/output window.
  • ✅ No ModuleNotFoundError for schedulers (e.g., CAME).
  • ✅ GPU utilization shows activity (via nvidia-smi or cloud monitor).

6. Next Steps

❓ Issues?

If you encounter errors during setup:

  1. Check 03-Troubleshooting first.
  2. Ensure your Python/Node versions match the prerequisites.
  3. Open a GitHub Issue with your error logs and deployment mode (Local/Cloud).

Clone this wiki locally