Zhipeng Zhou

@songdejia How to define large object in your code? Is that the ratio of the groundtruth to the image is greater than a certain threshold? Then how much the ratio...

@songdejia

@BigPuns @xujiao3628 难道不能用大一点的batch训练么 我很奇怪好多个代码都是按照batch_size=1来训练的，如果训练数据是ILSVRC和youtube-bb的话 那得训练到什么时候……

> Tasks in testing phase have the same structures as those in training phase. So, in testing phase, query_set should be generated as what in training does. The optimal theta...

I want to implement a normalized linear layer（e.g. [NormedLinear](https://github.com/kaidic/LDAM-DRW/blob/master/models/resnet_cifar.py)）that can be supported by KFAC in BackPack. What should I do? Are there any templates that I can refer to?

Thanks for your guidance! :)

If I leverage KFAC as above many times (e.g., repeat it each epoch), it would be possible to get the error "RuntimeError: The size of tensor a (64) must match...

If I fix the seed and test on different environments (Linux terminal, Docker container, etc.), this error happens at different time points; if I test on a single environment, then...

> The `create_batch` function is only called once when the `MiniImagenet` dataset object is created, which means the tasks sampled are the same in every epoch. > > I changed...