Yipeng Li

> If the number of images that belong to label 0 is 5923, which is less than 6000, some clients may have samples of more than two labels. I think...

> > 你好，10.14中z=WTx，WTx是d维的向量，z是d'维的向量，这里为什么相等了？ > > 同样的问题出现在西瓜书中+const那行的公式，zTWTx两者的维度都不一样，不是很懂这里怎么相乘 > > 所以，我其实觉得书中的W=（w1,w2,...,wd)其实应该是W=（w1,w2,...,wd')才正确，那样整个推导就合理了。 > > 是不是西瓜书那里印错了，我看10.13那里的W就是d'，在10.14是d了 > > z_i 是 d'维的列向量，WT是d'*d的矩阵，x_i是d维列向量，这个乘式的维度是正确的 请问下，对上面引述，如果看10.14下面这行，西瓜书对W的定义，W = (w1, w2, ... , wd)，那么W的维度应该是(anydimension\*d)，WT为(d\*anydimension)，您说的WT是d'*d与这里就矛盾了。

 您好。对于南瓜书上这段话，。根据上下文，alpha的下标 i 应该表示的是样本，而 u 和 v 则是词汇稀疏表示。这样看来，“没有样本之间的交互”我认为有失偏颇，因为它们（alpha^u和alpha^v）都在同一个样本 i 上。

We agree with mahf93, tiangency. The code uses `Softmax()` (line 16 in ref [1]) as the output unit and `NLLLoss()` as the loss function (line 34 in ref [2]), which...

Thanks for your approval. I'm happy to contribute to this repo. Since I'm not familiar how to pull requests, it may cost some time. By the way, we find that...

\^o^/, I have added ExDir successfully. I have only added some codes, so it is safe to add this strategy to the original code. One example: MNIST, num_clients=10, num_classes=10, C=5...

Best of luck with your CVPR paper!

Thanks for your approval. I have updated the pull request, with Extended Dirichlet strategy added. Feel free to change the code to meet the style of PFLlib, and just call...

Thanks for your patience and kindness. I have checked it and have no further problems.

Thanks. We are glad to hear from you. The code is [ExDirPartition](https://github.com/bird-two/convergence/blob/master/sim/data/partition.py), and you can generate the map with the following command (changing the dataset location is required). ``` python...