Victor Liu
Victor Liu
> > Hi,thank you for sharing this work. If I want to use VOLO to finetune on my own dataset, how can I create my own token labeling? > >...
~~同样问题，刚重置了密码还是不行，美国和国内节点都试过了，请问有什么解决办法吗？~~ 已经解决，用了cookie登录方法[issues #478 ](https://github.com/LeetCode-OpenSource/vscode-leetcode/issues/478),通过点击`all`拿到cookie然后复制全部cookie的value即可
请问GANomaly的实际表现如何？
Remove code at the beginning when loading the pretrained model and it will start from scratch.
同问题。上传非常慢，有没有解决方案？
It is just a shortcut to avoid repeating such code:`self.to_q = nn.Linear(dim, inner_dim, bias = bias)` three times(for qkv you need three linear mappings). Of course there are other implementations...
Agree. Line#143 already has one negative sign.
For people who are not willing to install the dependency, go check this thread about [depth-wise conv](https://discuss.pytorch.org/t/pytorch-implementation-of-depthwise-convolution/126919). And from that thread, the equivalent native pytorch implementation should be: `self.dwconv =...
fail和字符串没有关系，调整版本即可