错误:
使用Cross_entropy
损失函数时出现 RuntimeError: multi-target not supported at
原因:
其参数说明:
input has to be a 2D Tensor of size batch x n.
This criterion expects a class index (0 to nClasses-1) as the target for each value of a 1D tensor of size n
其标签必须为0~n-1,而且必须为1维的,如果设置标签为[nx1]的,则也会出现以上错误。
以上文字来源于https://blog.csdn.net/u011276025/article/details/73826562 BUG4
出错代码为:
# print(outputs.size()) # (128L, 2L)
# print(trg) # (128L, 1L)
loss = criterion(outputs, trg)
更改为:
# print(outputs.size()) # (128L, 2L)
# print(trg.squeeze()) # (128L, ) 或 [128]
loss = criterion(outputs, trg.squeeze())
错误解决。