![](https://img-blog.csdnimg.cn/20201014180756780.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
bugs
平凡搬砖
这个作者很懒,什么都没留下…
展开
-
tensorboard常见报错
tensorboard可能出现的bugs原创 2022-11-30 17:05:22 · 1869 阅读 · 2 评论 -
One of the differentiated Tensors appears to not have been used in the graph. Set allow_unused=True
One of the differentiated Tensors appears to not have been used in the graph. Set allow_unused=True if this is the desired behavior.原创 2022-08-09 15:44:30 · 3748 阅读 · 0 评论 -
tensorflow.python.framework.errors_impl.CancelledError: [_Derived_]RecvAsync is cancelled.
问题描述:使用tensorflow-gpu时,跑lstm,报了以下错误。具体原因不明,有知道的朋友帮我解释一下吧,谢谢! 参考:link环境python == 3.6tensorflow-gpu == 2.4.0报错复现tensorflow.python.framework.errors_impl.CancelledError: [_Derived_]RecvAsync is cancelled. [[{{node Adam/Adam/group_deps/NoOp/_95}}]] [原创 2022-02-16 16:15:03 · 1375 阅读 · 0 评论 -
关于pytorch tensor/params没有放在同一个设备上的经验
前言用pytorch做深度学习时候,全部tensor和变量都要放到同一个设备上, i.e. GPU or CPU。否则会报一下错误:RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument mat2 in method wrapper_mm)出现以上bug,目前我遇到的原创 2022-01-23 11:30:23 · 2631 阅读 · 0 评论 -
拓展vmware虚拟机的硬盘
记录这篇文章是因为碰到了以下这个bug,docker pull的时候设备空间不足:write /var/lib/docker/tmp/getimageblob415296441: no space left on device解决方法基本上可以follow这篇文章但是我也遇到一些问题:(1) 开机后进入不了gaprted的主界面,这里需要勾选Connect at power on,使他随开机一并读入 (如图Fig.1)。如果开机后依旧没有见到gaprted主界面,则重启,按Esc键进入如下界面,选原创 2022-01-03 18:40:10 · 616 阅读 · 0 评论 -
OSError: [WinError 127] The specified procedure could not be found.
报错复现:Traceback (most recent call last): File "./infer.py", line 27, in <module> import torch File "D:\Apps\Anaconda\envs\tse_2021_bases\lib\site-packages\torch\__init__.py", line 117, in <module> raise errOSError: [WinError 127] T原创 2021-11-18 21:32:18 · 2753 阅读 · 2 评论 -
TypeError: can‘t convert CUDA tensor to numpy.
报错复现及解决:TypeError: can't convert CUDA tensor to numpy. Use Tensor.cpu() to copy the tensor to host memory first.有点时候需要将tensor转化为numpy.ndarray,直接使用.numpy()可能出现以上问题。这是因为tensor在GPU上,需要先放到CPU上才能转。随后改用.cpu().numpy(),可能出现如下报错:RuntimeError: Can't call numpy()原创 2021-10-17 12:39:42 · 785 阅读 · 0 评论 -
Pytorch错误:Torch not compiled with CUDA enabled
报错复现:Traceback (most recent call last): File "../train.py", line 388, in <module> main() File "../train.py", line 384, in main train(args) File "../train.py", line 378, in train trainer.train() File "../train.py", line 322, in tr原创 2021-10-07 14:58:49 · 708 阅读 · 0 评论