.. _arch-dpa3:

DPA3 (experimental)
======================

This is an interface to the DPA3 architecture described in https://arxiv.org/abs/2506.01686
and implemented in deepmd-kit (https://github.com/deepmodeling/deepmd-kit).

.. _arch-dpa3_installation:

Installation
------------

To install this architecture along with the ``metatrain`` package, run:

.. code-block:: bash

    pip install metatrain[dpa3]

where the square brackets indicate that you want to install the optional
dependencies required for ``dpa3``.

.. _arch-dpa3_default_hypers:

Default Hyperparameters
-----------------------

The description of all the hyperparameters used in ``dpa3`` is provided
further down this page. However, here we provide you with a yaml file containing all
the default hyperparameters, which might be convenient as a starting point to
create your own hyperparameter files:

.. literalinclude:: ../default_hypers/dpa3-default-hypers.yaml
   :language: yaml

.. _arch-dpa3_model_hypers:

Model hyperparameters
------------------------

The parameters that go under the ``architecture.model`` section of the config file
are the following:

.. container:: mtt-hypers-remove-classname

    ..

        .. autoattribute:: metatrain.experimental.dpa3.documentation.ModelHypers.type_map

        .. autoattribute:: metatrain.experimental.dpa3.documentation.ModelHypers.descriptor

        .. autoattribute:: metatrain.experimental.dpa3.documentation.ModelHypers.fitting_net



.. _arch-dpa3_trainer_hypers:

Trainer hyperparameters
-------------------------

The parameters that go under the ``architecture.trainer`` section of the config file
are the following:

.. container:: mtt-hypers-remove-classname

    ..

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.distributed

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.distributed_port

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.batch_size

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.num_epochs

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.learning_rate

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.scheduler_patience

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.scheduler_factor

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.log_interval

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.checkpoint_interval

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.scale_targets

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.fixed_composition_weights

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.per_structure_targets

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.log_mae

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.log_separate_blocks

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.best_model_metric

        .. autoattribute:: metatrain.experimental.dpa3.documentation.TrainerHypers.loss



.. _arch-dpa3_references:

References
----------

.. footbibliography::