Improvements on dynamic GPU handling and TorchDrug unpatching: - dynamic inference for RegressionTransformer - finer control on torch versions in the unpatching module
0.44.2
Supporting flexible device allocation for RegressionTransformer inference.
0.44.1
Updating MoLeR dependency to version v0.2.0.
0.44.0
Updating target handling in PaccMannGP to ensure deepcopy in-between repeated runs.