RuntimeError：梯度计算所需的变量之一已被就地操作修改：PyTorch 错误

Posted 2023-03-25

【中文标题】RuntimeError：梯度计算所需的变量之一已被就地操作修改：PyTorch 错误

【英文标题】：RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: PyTorch error

【发布时间】：2021-03-04 02:00:03

【问题描述】：

我正在尝试在 PyTorch 中运行一些代码，但此时我被堆积了：

在第一次迭代中，判别器和生成器的后向操作都运行良好

....

self.G_loss.backward(retain_graph=True)

self.D_loss.backward()

...

在第二次迭代中，当self.G_loss.backward(retain_graph=True) 执行时，我收到此错误：

RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [8192, 512]] is at version 2; expected version 1 instead. Hint: the backtrace further above shows the operation that failed to compute its gradient. The variable in question was changed in there or anywhere later. Good luck!

根据torch.autograd.set_detect_anomaly，判别器网络中的以下最后一行负责此：

    bottleneck = bottleneck[:-1]
    self.embedding = x.view(x.size(0), -1)
    self.logit = self.layers[-1](self.embedding)

奇怪的是，我在其他正常工作的代码中使用了该网络架构。有什么建议吗？

完整的错误：

    site-packages\torch\autograd\__init__.py", line 127, in backward
    allow_unreachable=True)  # allow_unreachable flag
RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [8192, 512]] is at version 2; expected version 1 instead. Hint: the backtrace further above shows the operation that failed to compute its gradient. The variable in question was changed in there or anywhere later. Good luck!

【参考方案1】：

通过删除带有loss += loss_val 行的代码解决