Warning: Grad strides do not match bucket view strides pytorch利用DDP报错

遇到报错: [W reducer.cpp:362] Warning: Grad strides do not match bucket view strides. This may indicate grad was not created according to the gradient layout contract, or that the param’s strides changed since DDP was constructed. This is not an error, but may impair performance.

机翻:警告。梯度与桶状视图的梯度不一致。这可能表明grad没有按照梯度布局合同创建,或者参数的步长在DDP构建后发生了变化。 这不是一个错误,但可能影响性能。

在这里插入图片描述

方法1:
Github的issue:Grad strides do not match bucket view strides.#47163
pytorch利用DDP进行加速的报错问题
在transpose或者permute后面加 .contiguous()
但是加了之后还是有警告
Warning: Grad strides do not match bucket view strides pytorch利用DDP报错_第1张图片

方法2:
类似于x = x[:, :H, :W, :]这种切片的操作
改成:

x = x[:, :H, :W, :].contiguous()

你可能感兴趣的:(神经网络,pytorch,人工智能,python)