Tutorials

A hands-on walkthrough of MolCraftDiffusion, ordered as a learning path: prepare data → train → generate → evaluate. Each tutorial is self-contained, but they build on one another, so first-time users should follow them in order.

Tip

New here? Start with Installation, then work through Data Preparation → Training a Diffusion Model → Generation Overview. The rest are optional deep-dives you can reach for when you need them.

1 · Prepare

0 · Data Preparation

Compile raw .xyz files into ASE databases, featurize, augment, and wire them into the DataModule.

Tutorial 0: Data Preparation & Management

2 · Train

1 · Diffusion Model

Train a 3D molecular diffusion model from scratch with the override-only config workflow.

Tutorial 1: Training a Diffusion Model
2 · Regressor

Train a property predictor — a standalone model and the basis for gradient guidance.

Tutorial 2: Training a Regressor Model
3 · Guidance Model

Train a time-aware regressor on noisy data to steer generation toward target properties.

Tutorial 3: Training a Time-Aware Guidance Model
4 · Fine-Tuning

Adapt a pretrained model to a new chemical space, add conditions, or specialize for outpainting.

Tutorial 4: Fine-Tuning a Diffusion Model

3 · Generate

5 · Generation Overview

Unconditional sampling and the three generation modes at a glance.

Tutorial 5: Molecule Generation Overview
6 · Structure-Guided

Inpainting and outpainting with 3D geometric constraints, plus a full parameter-tuning guide.

Tutorial 6: Structure-Guided Generation
7 · Property-Directed

Steer generation with Classifier-Free Guidance, Gradient Guidance, or a hybrid of both.

Tutorial 7: Property-Directed Generation

4 · Evaluate & Scale

8 · Predict & Evaluate

Predict properties for new molecules and benchmark a model against a labeled set.

Tutorial 8: Prediction and Evaluation Modes
9 · Analyze

Post-generation analysis: xTB optimization, validity metrics, RMSD, electronic properties, featurization.

Tutorial 9: Analyze Module - 3D Molecular Structure Analysis
10 · Generation Sweeps

Grid and Bayesian sweeps over controlled-generation parameters with automatic metric collection.

Tutorial 10: Generation Parameter Sweeps