Warning
metatensor-models is still very early in the concept stage. You should not use it for anything important.
This is a repository for models using metatensor, in one shape or another. The only requirement is for these models to be able to take metatensor objects as inputs and outputs. The models do not need to live entirely in this repository: in the most extreme case, this repository can simply contain a wrapper to an external model.
The idea behind metatensor-models is to have a general hub
that provide an homogeneous enviroment and user interface to
train, export and evaluate ML models and to
connect those models with various MD engines (e.g. LAMMPS, i-PI, ASE ...).
metatensor-models is the tool that transforms every ML architecture in an
end-to-end model. Any custom ML architecture compatible with TorchScript
can be integrated in metatensor-models, gaining automatic access to
a training and evaluation interface, as well as compatibility with various MD engines.
Note: metatensor-models does not provide per se mathematical functionalities but
relies on external models that implement the various architectures.
- Custom ML Architecture: Integrate any TorchScriptable ML model to explore innovative architectures.
- MD Engine Compatibility: Supports various MD engines for diverse research and application needs.
- Streamlined Training: Automated process leveraging MD-generated data to
optimize ML models with minimal effort. It uses the
hydramodule to easy management of folder and files. - HPC Compatibility: Efficient in HPC environments for extensive simulations.
- Future-Proof: Extensible to accommodate advancements in ML and MD fields.
Currently metatensor-models supports the following architectures for building an atomistic model.
| Name | Description |
|---|---|
| SOAP BPNN | A Behler-Parrinello neural network with SOAP features |
| Alchemical Model | A Behler-Parrinello neural network with SOAP features and Alchemical Compression of the composition space |
For details, tutorials, and examples, please have a look at our documentation.
You can install metatensor-models with pip:
git clone https://github.com/lab-cosmo/metatensor-models
cd metatensor-models
pip install .In addition, specific models must be installed by specifying the model name. For example, to install the SOAP-BPNN model, you can run:
pip install .[soap-bpnn]metatensor-models comes with completion definitions for its commands for bash and
zsh. Since it is difficult to automatically configure shell completions in a robust
manner, you must manually configure your shell to enable its completion support.
To make the completions available, source the definitions as
part of your shell's startup. Add the following to your ~/.bash_profile,
~/.zshrc (or, if they don't exist, ~/.profile):
source $(metatensor-models --shell-completion)Having a problem with metatensor-models? Please let us know by submitting an issue.
Submit new features or bug fixes through a pull request.
Thanks goes to all people that make metatensor-models possible: