This small PR implements the changes we found to be necessary when trying to run on the HMS cluster. Ideally, @mauriciogtec can try training in this branch and confirm that everything works as expected on cannon as well. Then we can merge.
Changes:
ee0860d specify cuda version in enviornment.yaml
f6aee4d fix the wandb.init() call
bd2cfac fix a small bug in --post_pool_filter
c84f888 add new connectivity self
564c14f create training script with EGNN-like hyperparams
This small PR implements the changes we found to be necessary when trying to run on the HMS cluster. Ideally, @mauriciogtec can try training in this branch and confirm that everything works as expected on cannon as well. Then we can merge.
Changes:
enviornment.yaml
wandb.init()
call--post_pool_filter
self