DataType error when computing dataset logits in DER Task-Incremental scenario

[guilhermegog]

When training a multi-head module with DER there seems to be a bug as the compute_dataset_logits() function is expected to return a tensor, but seemingly creates a dictionary.

To replicate, simply run the multihead.py with the DER strategy.

I tried fixing the problem by adding a check on the aforementioned function to see if a dictionary is being generated as the output of the model and to convert the dictionary values into the desired data type (lines 48 to 52 on der.py):

if(isinstance(out,dict)):
  out = out.values()
  out = list(out)[0]

but sometimes this conversion yields tensors with seemingly random sizes ([128,6] or [128,9] instead of the expected [128,10])