Access pre-trained MACE models
Project description
MACE models
Effortlessly integrate pre-trained MACE models into your projects with the user-friendly mace-models
package.
pip install mace-models
Loading models is simple with a few lines of code:
import mace_models
# Load a default MACE model
model = mace_models.load()
torch_model = model.get_model()
ase_calculator = model.get_calculator()
# get the model file
model.get_file()
Customize your model selection by loading models from various repositories or specific revisions:
model = mace_models.load(
"<model-name>",
rev="<branch-or-sha>",
remote="https://github.com/<user>/<repo>"
)
Ensure a seamless integration by first installing the MACE library and upgrading your PyTorch installation:
pip install --upgrade torch --extra-index-url https://download.pytorch.org/whl/cu116
pip install git+https://github.com/ACEsuit/mace.git
Models
ani500k_small
training data: https://www.nature.com/articles/sdata2017193hydromace
https://github.com/RokasEl/hydromacemedium_spice
https://arxiv.org/abs/2312.15211small_spice
https://arxiv.org/abs/2312.15211
Example usages with ASE
import mace_models
from ase.build import molecule
model = mace_models.load("medium_spice")
water = molecule("H2O")
water.calc = model.get_calculator(dtype="float64")
print(water.get_potential_energy())
>>> -14.159366
References
Models are taken from https://github.com/ACEsuit/mace/blob/docs/docs/examples/ANI_trained_MACE.zip and from
@misc{kovacsMACEOFF23TransferableMachine2023,
title = {MACE-OFF23: Transferable Machine Learning Force Fields for Organic Molecules},
author = {Kov{\'a}cs, D{\'a}vid P{\'e}ter and Moore, J. Harry and Browning, Nicholas J. and Batatia, Ilyes and Horton, Joshua T. and Kapil, Venkat and Witt, William C. and Magd{\u a}u, Ioan-Bogdan and Cole, Daniel J. and Cs{\'a}nyi, G{\'a}bor},
year = {2023},
number = {arXiv:2312.15211},
}
MACE is described in
@misc{batatiaMACEHigherOrder2022,
title = {{{MACE}}: {{Higher Order Equivariant Message Passing Neural Networks}} for {{Fast}} and {{Accurate Force Fields}}},
shorttitle = {{{MACE}}},
author = {Batatia, Ilyes and Kov{\'a}cs, D{\'a}vid P{\'e}ter and Simm, Gregor N. C. and Ortner, Christoph and Cs{\'a}nyi, G{\'a}bor},
year = {2022},
number = {arXiv:2206.07697},
eprint = {2206.07697},
primaryclass = {cond-mat, physics:physics, stat},
publisher = {{arXiv}},
urldate = {2022-06-19},
archiveprefix = {arxiv},
langid = {english}
}
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
mace_models-0.1.1.tar.gz
(3.6 kB
view hashes)
Built Distribution
Close
Hashes for mace_models-0.1.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 9cf7a962846643b2d04f614bc3db68df39573475c67c78f390e5864325972d37 |
|
MD5 | d5d603bbc5668d8b0958131fff8b3e20 |
|
BLAKE2b-256 | 244a84de04a28f418a2dcb210758123030e4f5d1264c814aa42b26d16af21895 |