Getting started

Install

To use GWAI, we strongly recommend using the release of GWAI container with GPU nodes.

You can launch an instance of the astropre container and mount GWAI as well as your dataset with the following Docker commands.

$ docker pull zzhopezhou/astropre:gwda
$ docker run --gpus all -itd -v /path/to/GWAI:/workspace/GWAI -v /path/to/dataset:/workspace/dataset zzhopezhou/astropre:gwda

In the container, two environments of different python version are provided. Specifically, the base environment is mainly used for model training and the waveform environment is for data generation.

(base) root@93b17a314f9d:/workspace# which python
/opt/conda/bin/python
(base) root@93b17a314f9d:/workspace# conda activate waveform
(waveform) root@93b17a314f9d:/workspace# which python
/opt/conda/envs/waveform/bin/python

If you can’t use this for some reason, use the latest pytorch, cuda, nccl, NVIDIA APEX and make sure that the following required python packages are successfully installed.

astropy
corner
deepspeed
esbonio
fastdtw
few
fftw
gwdatafind
gwpy
gwsurrogate
hydra-core
imbalanced-learn==0.11.0
lalsimulation
lalsuite
librosa
ligotimegps
lisaorbits
nflows
numpy
matplotlib
omegaconf
pandas
pillow
pybind11
PyCBC
rich
scikit-learn
scipy
speechbrain
statsmodels
tensorrt
torch
torchsummary
torchtext
torchvision
transformers
wandb