Closed davidegraff closed 2 years ago
this PR updates the distributed training code to bring it up-to-date with ray 1.11.
Summary:
mpnn.sgd
mpnn.ray
ray.sgd.v2
ray.train
MoleculeDataset
DataLoader
mask = torch.tensor([[bool(y) for y in ys] for ys in targets], device=device) # <-- necessary when targets was List[List[Optional[float]]
to
mask = ~torch.isnan(targets) # <-- possible because targets is now torch.Tensor
molpal.models
Description
this PR updates the distributed training code to bring it up-to-date with ray 1.11.
Summary:
mpnn.sgd
module tompnn.ray
ray.sgd.v2
to the newray.train
interfaceMoleculeDataset
andDataLoader
code to return arrays/tensors rather than lists (to work better with training loops overall)to
molpal.models
moduleChecklist