PyTorch | UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to

UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.

 

源程序大概是这样的：

import torch.nn as nn

.......

class gan(object):

  def __init__(self,opt,dataloader=None):

    ......

    self.sf = nn.Softmax

    ......

  def test(self):

    self.x = self.sf(self.y)

运行，报错：

/home/liuming/weilanlan/DL/lib/model.py:420: UserWarning: Implicit dimension choice for softmax has been deprecated. Change the call to include dim=X as an argument.
  self.gt2_pre_labels_tmp[i*self.opt.batchsize : i*self.opt.batchsize+error.size(0)] = torch.argmax(self.smax(self.latent_i_rs), dim = 1)
Segmentation fault (core dumped)

根据报错，添加了dim的信息，self.x = self.sf(self.y,dim=1),运行依然报错

    self.gt2_pre_labels_tmp[i*self.opt.batchsize : i*self.opt.batchsize+error.size(0)] = torch.argmax(self.smax(self.latent_i_rs, dim=1), dim = 1)
  File "/home/liuming/.local/lib/python2.7/site-packages/torch/nn/modules/module.py", line 477, in __call__
    result = self.forward(*input, **kwargs)
TypeError: forward() got an unexpected keyword argument 'dim'

 

excuse me????我以为是什么内部问题，后来看到这篇文章https://discuss.pytorch.org/t/hypocritical-error-messages-re-softmax/27206，才发现是我把dim位置放错了

于是按照提示修改了dim的位置，

self.sf =nn.Softmax(dim=1)

成功的解决了这个问题！

解决这个问题花费了我两个小时的时间，希望以后有人在遇到类似的问题可以快速解决这个问题！！

 

欢迎各位不吝赐教。