task-free online continual learning (TF-CL) is a challenging problem where
the model incrementally learns tasks without explicit task information.
Although training with entire data from the past, present as well as future is
considered as the gold standard, naive approaches in TF-CL w