xfflzl/CIM-CIL

The fixed_memory value

zwwaaavv opened this issue · 0 comments

Sorry to bother you. Does setting fixed_memory = true represent the "growing" setting an experimental setting rather than the "fixed" setting with a fixed memory size? Because I noticed that when I set fixed_memory = false, the number of exemplars for each class varies with each task training , while the sum of the exemplars for all classes approaches the number of memory_size.Thank you!