david

@vilin777 @zhangjiantianyasmile 我也遇到了这个问题， 按照icml 2007上面tradaboost的权重更新方法复现了只有数值型特征的mushroom数据集上的实验，是无法收敛的，加权错误率始终在0.5左右。而按照作者更改之前的tradaboost更新方法（即beta_t是1，beta是-1），实验反而收敛了。。很奇怪

@siavash-khodadadeh I have the same question. And I notice that the paper said ``` If we are doing K-shot, N-way classification, then we sample tasks by selecting N classes from...

@Vampire-Vx @yinxiaojian I have also tried to set ```retain_graph = True, create_graph=True```, but for mini-Imagenet, the performance is weaker than before. Besides, the hidden dimensions I used for mini-imagent is...