User Guide

Use these pages after the quickstart when you need workflow-level guidance for real data preparation, training, and optional integrations.

Workflow pages

Tree input

Start from in-memory ete3.Tree objects or optional file readers.

Feature engineering

Attach deterministic numeric attributes to tree nodes.

Graph conversion

Convert feature-bearing trees into PyTorch Geometric Data objects.

Datasets and splits

Package graph samples, labels, and deterministic train/validation/test partitions.

Training

Run the trainer lifecycle with PyG datasets, loaders, checkpoints, and predictions.

Training configuration

Use local TOML files for repeatable model, trainer, loss, metrics, and tracking settings.

Metrics and tracking

Use built-in metrics and optional Weights & Biases logging.

• Tree Input
• Feature Engineering
• Graph Conversion
• Datasets And Splits
• Training
• Training Configuration
• Metrics And Tracking

How the pages fit

Start with tree input, attach features, convert graphs, prepare datasets, and train with local checkpoints. Optional pages explain tracking and file formats that require extras. API details live in Reference.