Is there or will there be support for fine-tuning based on the property prediction task (i.e., weight transfer functionality) ? about torchdrug

[KiddoZhu]

Hi! I think this is already easy to implement in TorchDrug.

Because models and tasks are separated in TorchDrug, you can define a representation learning model, wrap it in a pretraining task and a finetuning task. You first train the pretraining task and then train the finetuning task.

I just checked the tutorial of pretrained molecular representation, and it misses the steps of finetuning. We will update it later.

[KiddoZhu]

Finetuning is now in the tutorial of pretrained molecular representations. Please refer to that if you need it.