https://github.com/songlab-cal/mogwai
https://github.com/songlab-cal/mogwai
Last synced: 22 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/songlab-cal/mogwai
- Owner: songlab-cal
- License: bsd-3-clause
- Created: 2020-10-12T19:31:21.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2022-05-09T17:24:14.000Z (about 3 years ago)
- Last Synced: 2023-08-15T13:47:30.259Z (almost 2 years ago)
- Language: Python
- Size: 2.58 MB
- Stars: 24
- Watchers: 5
- Forks: 6
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# mogwai: Probabilistic Models of Protein Families
A library of tested models, metrics, and data loading for protein families. Implemented in PyTorch and PyTorch Lightning.
Under active development, feedback welcome.
## Getting Started
For now, we support cloning and installing in developer mode.
```bash
pip install -e .
```
You will also need to install `apex` (needed to use DDP and FusedLamb training):
```
source venv/bin/activate
git clone [email protected]:NVIDIA/apex.git
cd apex
```
Modify `setup.py`
Find:
```
if (bare_metal_major != torch_binary_major) or (bare_metal_minor != torch_binary_minor):
```
And replace it with
```
if (bare_metal_major != torch_binary_major):
```
To remove the minor version check. This will allow apex to install.
Install apex
```
pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" ./
```
## Examples
* [Potts Model with Pseudolikelihood](https://github.com/nickbhat/mogwai/blob/main/examples/gremlin_train.ipynb)