Uploading and saving Adapter-Fusion layers

[Soumyadeepsar]

We can upload adapters using push_adapter_to_hub() in our huggingface hub. What is the way of uploading the Adapter fusion layer (fused layer) after training on a task? So for adapter-fusion, we create a fuse block and also add a classification head. Can you tell me how to upload these layers so that they can be used later on by other people?

[calpt]

Hey, this functionality will be added with this PR: #771

[Soumyadeepsar]

Ok, thank you for your reply. That's great news!!