molprop.utils.transfer_learning.freeze

molprop.utils.transfer_learning.freeze(model, args)

Freezes MPNN part of GNN. Processing on fingerprint is still updated.

Parameters:

• model (torch.nn.Module) – Model used in training process.

• args – Parameters defined below Transfer_learning in defaultconfig.yaml.

Returns:

Model used in training process.

Return type:

model (torch.nn.Module)