I need a way to merge multiple databunches into one. For example lets say there are 3 databunches D1, D2 and D3. All 3 have batches which produce tensors of same size(same image size and batchsize, and same target shape). I want to construct databunch D whose train_dl has all train_dl batches of D1 then D2 then D3. D’s valid _dl also needs to be constructed similarly.
Is there any way to do this?