User Guide

This page details how to use mmml.

Overview

High-level concepts, capabilities, and typical workflows with mmml.

Command-line Interface

How to run common tasks from the CLI (training, ESP generation, conversions).

PhysNet Models

Charge-Spin Conditioned Model

For multi-state predictions (ionization energies, singlet-triplet gaps, etc.), see Charge-Spin Conditioned PhysNet.

Data Loading

Packed Memmap Loader

For efficient training on large datasets (>10GB), see Packed Memmap Data Loader.

Saving Results from PySCF GPU

Use the CLI flag --save_option to control how results are persisted when running mmml.pyscf4gpuInterface.calcs. Supported values are pkl, npz, and hdf5. See PySCF GPU Interface API for details and examples.

Python API

Programmatic usage patterns and key modules to import.

Configuration

How to configure runs, set seeds, and control resources.

Troubleshooting

Common issues and tips (imports, GPU setup, missing optional dependencies).