部分模块在使用GPU计算时报错(Most modules report errors when using GPU)

[ImcwjHere]

报错内容类似于：
RuntimeError: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same
或者
Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument index in method wrapper__index_select)

修复方案：
将list修改为modulelist
如model/mlp/mlp_mixer.py文件中(49行)：
将self.mlp_blocks=[]修改为self.mlp_blocks=nn.ModuleList([])

[zeng-cy]

改了还是不行

[daviscsuft]

请问还有别的方法吗，ccnet里没有list