salesforce
How to specify a control tag in the sample script?
In the paper it was mentioned that one can sample sequences based on a control tag. Could you provide an example for using a control tag with the sample.py script?
I'm very curious about that too. Based on the paper I'm assuming they introduced the control tags during fine tuning on a sample protein family (lysozymes in the case of the paper).
For my use case unfortunately there isn't a well known family with enough sequence data to train on. I'm wondering if we could prompt the model with an existing sequence to generate new ones based on distance.
@okaris Apparently this repo is for ProGen2. In ProGen2 the control tags were removed AFAIK. The Nature Biotechenology paper refers to ProGen1 and that's what made the confusion.
Will be happy for a confirmation from one of the authors
That's really confusing. (And I do NOT understand why they'd remove the control tags :( )
