https://github.com/ulissigroup/finetuna

Active Learning for Machine Learning Potentials
https://github.com/ulissigroup/finetuna

Last synced: 3 months ago
JSON representation

Active Learning for Machine Learning Potentials

• Host: GitHub
• URL: https://github.com/ulissigroup/finetuna
• Owner: ulissigroup
• License: mit
• Created: 2020-09-22T14:39:52.000Z (about 4 years ago)
• Default Branch: main
• Last Pushed: 2024-05-15T17:26:24.000Z (6 months ago)
• Last Synced: 2024-05-16T05:51:01.672Z (6 months ago)
• Language: Python
• Size: 8.02 MB
• Stars: 41
• Watchers: 3
• Forks: 11
• Open Issues: 9
• Metadata Files:
  • Readme: README.md
  • License: LICENSE.md

Awesome Lists containing this project

• best-of-atomistic-machine-learning - GitHub - 25% open · ⏱️ 15.05.2024): (Active learning)

README

        

[![Lint](https://github.com/ulissigroup/finetuna/actions/workflows/black.yml/badge.svg)](https://github.com/ulissigroup/finetuna/actions/workflows/black.yml)

[![Test](https://github.com/ulissigroup/finetuna/actions/workflows/unittests.yml/badge.svg)](https://github.com/ulissigroup/finetuna/actions/workflows/unittests.yml)

## *FINETUNA*: Fine-Tuning Accelerated Molecular Simulations



Are you doing structural optimizations with DFT or other electronic structure codes?? Try :monocle_face::fish: FINETUNA for accurate but 90% faster relaxation!

FINETUNA accelerates atomistic simulations by fine-tuning a pre-trained graph model in an active learning framework.

Installation is easy:

```

conda env create -f env.cpu.yml

conda activate finetuna

cd finetuna

pip install -e .

git clone https://github.com/Open-Catalyst-Project/ocp.git

cd ocp

pip install -e .

pip install git+https://github.com/ulissigroup/vasp-interactive.git

```

All pre-trained machine learning model checkpoint can be found [here](https://github.com/Open-Catalyst-Project/ocp/blob/main/MODELS.md). We recommend to download the GemNet-dT all model. [click here to download](https://dl.fbaipublicfiles.com/opencatalystproject/models/2021_08/s2ef/gemnet_t_direct_h512_all.pt).

You are all set! Now in your VASP input folder, run the calculation by: `finetuna_wrap.py -c /path/to/the/checkpoint`.



### Usage

If you have an ASE atoms object, see example [1](https://github.com/ulissigroup/finetuna/blob/main/examples/online_al/online_al_example.py) and [2](https://github.com/ulissigroup/finetuna/blob/main/examples/beef/online_al_beef_example.py).

If you have VASP input files (INCAR, KPOINTS, POTCAR, and POSCAR), see example [3](https://github.com/ulissigroup/finetuna/tree/main/finetuna/vasp_wrapper).