RuntimeError: one of the variables needed for gradient computation has been modified by an inplace

【Pytorch】RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [16, 512, 8, 8]], which is output 0 of ReluBackward1, is at version 3; expected version 2 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradient, with torch.autograd.set_detect_anomaly(True).

这个报错问题是在运行vgg16时出现的，将代码nn.ReLU(inplace=True)中的inplace=True去掉后，代码运行正常，inplace默认为False。但在运行AlexNet时inplace=True代码未报错，以前运行vgg16时也未报错，还不知道这个问题出现的根源在哪，先记录下这个问题，便于以后能快速查找解决问题的方法。

将

nn.ReLU(inplace=True),

改为

nn.ReLU(),

即可

参考博客：https://www.cnblogs.com/yunshangyue71/p/13294999.html