Thank you for your good paper and code. But I have a few questions.
There is no mention of "Gridsearch" in the paper, but there is "GridSearch" in your Git code. I wonder if you used "Gridsearch" in the paper experiment you wrote.
SH files in the posted script are set to use 200 exemplars for all methods. I wonder if you used the 200 exemplars for all CL methods from the experiment in your paper.
I tried to run the code as it was your Git code. In particular, the results were different from the Figure 2 of your paper, and when compared to the Table 2, they also were different.
I wonder if I missed something.
It seems your code uses ResNet-18 taken from Torchvision models. Did you use ResNet-18 without modification? Because MedMNIST datasets image size are (28 28), not (224 224).
Thank you for your good paper and code. But I have a few questions.
There is no mention of "Gridsearch" in the paper, but there is "GridSearch" in your Git code. I wonder if you used "Gridsearch" in the paper experiment you wrote.
SH files in the posted script are set to use 200 exemplars for all methods. I wonder if you used the 200 exemplars for all CL methods from the experiment in your paper.
I tried to run the code as it was your Git code. In particular, the results were different from the Figure 2 of your paper, and when compared to the Table 2, they also were different. I wonder if I missed something.
It seems your code uses ResNet-18 taken from Torchvision models. Did you use ResNet-18 without modification? Because MedMNIST datasets image size are (28 28), not (224 224).