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 :doc:`physnet_charge_spin`.

Data Loading
------------

Packed Memmap Loader
^^^^^^^^^^^^^^^^^^^^

For efficient training on large datasets (>10GB), see :doc:`packed_memmap_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 :ref:`PySCF GPU Interface API <pyscf4gpu_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).