drsad / icarl Goto Github PK

请问保存旧样本的列表太大了总是报错有什么办法解决吗，或者说您用的GPU的内存是多少G的呢？

您好，阅读了您的代码，写得很好，有个疑问，这里的损失您好像只计算了蒸馏的损失，对于论文中的第一项损失似乎没有计算？

感谢您的工作，代码写的很优秀；
这边想请教您为什么最后出来的结果比原文好这么多，我跑完平均也有65+；
具体是哪几个部分在原文上做了优化呢，感谢回复。

想问一下，为什么我复现您的实验结果时，正确率总是会偏低一些

你好，我最近想学习一下增量学习的相关知识点，感谢您写的关于icarl的代码，我运行了一下，除了一个小问题：显示iCIFAR100.py文件中的57行，AttributeError：‘iCIFAR100’ object has no attribute 'data'。希望您能为我指点一下，谢谢。我感觉应该是版本的我问题，我用的torch=0.4.1

你好，麻烦登一下QQ，有问题想请教

Hi, I am a student who learning incremental learning. I have use this code to train cifar-100 dataset which you provide, now I want to train my own dataset by this model. I meet some problem, can you tell me how to change this code to make it to be suitable for my dataset which is .mat form and it is necessary to divide the dataset to train and test file? Thanks for you reading, and I will expect you reply.

what different mode 0 and mode 1 accuracy ...called from train and after train with testloader

在iCaRL.py第187行应保存模型的OrderedDict参数，即使用torch.save(self.model.state_dict(), 'file_name')

Please. why the number of test sets is more than that of training sets in the training process?
for instance
the size of train set is (6980, 32, 32, 3)
the size of train label is (6980,)
the size of test set is (10000, 32, 32, 3)
the size of test label is (10000,)
please.