Fair-Chem Environment Installation Guide - Based on Conda Env
2024-10-28
fairchem
is the FAIR Chemistry's centralized repository of all its data, models, demos, and application efforts for materials science and quantum chemistry.
Background
In the official tutorial of Fair-chem, while installing certain dependencies like torch-scatter
, torch-sparse
, torch-spline-conv
, torch-cluster
, and torch-geometric
, you might encounter some errors. However, there is a feasible way to set up the Fair-chem environment.
Environment Installation Process (Using Conda)
Create a new Conda environment:
conda create -n fair-chem python=3.10
Enter the fair-chem conda environment:
conda init
source ~/.bashrc
conda activate fair-chem
Install PyTorch with CUDA support in conda env:
pip install https://download.pytorch.org/whl/cu121/torch-2.4.0%2Bcu121-cp310-cp310-linux_x86_64.whl
Other versions are theoretically possible as well., Refer to the link at the end of this article.
Install the required dependencies:
pip install torch-scatter torch-sparse torch-spline-conv torch-cluster torch-geometric -f https://pytorch-geometric.com/whl/torch-2.4.0%2Bcu121.html
Install additional Python libraries(don't need):pip install numpy scipy ase matplotlib pandas
Installing Fair-chem
Clone the Fair-chem repository:
git clone https://github.com/FAIR-Chem/fairchem.git
Navigate to the Fair-chem directory:
cd fairchem
Install the core package:
pip install -e packages/fairchem-core[dev]
Install the data processing package:
pip install -e packages/fairchem-data-oc
Install the application package (optional):
pip install -e packages/fairchem-applications-cattsunami
If need other fair-chem packages, we can check them under the/packages
floor, and there are also usage demos available.check Installation:
>>python,Check installation status:import torch
import torch_geometric
import torch_scatter
import torch_sparse
import torch_cluster
import torch_spline_conv
import fairchem
print("=== Version Info ===")
print(f"PyTorch: {torch.__version__}")
print(f"CUDA: {torch.version.cuda}")
print(f"PyG: {torch_geometric.__version__}")
print(f"torch_scatter: {torch_scatter.__version__}")
print(f"torch_sparse: {torch_sparse.__version__}")
print(f"torch_cluster: {torch_cluster.__version__}")
print(f"torch_spline_conv: {torch_spline_conv.__version__}")
print(f"FAIR-Chem: {fairchem.__version__}")
>>python,List available models:from fairchem.core.models.model_registry import available_pretrained_models
print(available_pretrained_models)