Getting Started

• Quick Start — End-to-End in 5 Steps
  • Step 1 — Pull the container from NGC
  • Step 2 — Train the Edge Prediction (EP) model
  • Step 3 — Serve the EP model
  • Step 4 — Install client dependencies
  • Step 5 — Run the EP inference client
  • Step 5b — (Optional) Train EP with saved node embeddings
  • What to check at each step
• Prerequisites
  • Hardware Requirements
  • Software Requirements
  • Technology Stack
  • NGC Authentication
  • Inference Client Requirements
  • LLM Explainability (Optional)
• Support Matrix
  • Container
  • Inference Server
  • Supported GPU Architectures
  • Supported Data Formats
  • Supported Model Kinds
  • Supported GNN Encoders
  • Supported LLM Providers (Explainability)
• Run Training Using Financial Fraud Training
  • 1. Authenticate with NVIDIA GPU Cloud (NGC) and Pull the Docker Image
  • 2. Train Models