import torch def AddDataToTorch(oriData, newData, addDim): if (oriData.numel() != 0): if (len(oriData.shape) > addDim) and (len(newData.shape) > addDim): return torch.cat((oriData.view(-1, 1), newData.view(-1, 1)), dim = addDim) else: return torch.cat((oriData.view(-1, 1), newData.view(-1, 1)), dim = 0) mergeData = torch.Tensor() tmpData = torch.randn(3) for i in range(1, 10): mergeData = AddDataToTorch(mergeData, tmpData, 1) 这段代码为什么提示AttributeError: 'NoneType' object has no attribute 'numel'

这是因为在第一次调用`AddDataToTorch`时，`oriData`是空的，即为`NoneType`，没有`numel()`属性。你可以在调用`AddDataToTorch`之前添加一个判断语句，在`oriData`为空时进行特殊处理，如返回`newData`本身或创建一个大小为`newData`的新张量。例如：

import torch

def AddDataToTorch(oriData, newData, addDim):
    if oriData is None:
        return newData
    elif (len(oriData.shape) > addDim) and (len(newData.shape) > addDim):
        return torch.cat((oriData.view(-1, 1), newData.view(-1, 1)), dim=addDim)
    else:
        return torch.cat((oriData.view(-1, 1), newData.view(-1, 1)), dim=0)

mergeData = None
tmpData = torch.randn(3)
for i in range(1, 10):
    mergeData = AddDataToTorch(mergeData, tmpData, 1)

这样就可以避免该错误。