Tensorboard 无法在端口 6006 上正常运行 [英] Tensorboard Not Running Properly on port 6006

问题描述

我在使用 tensorboard 时遇到问题，可能是由于 windows10 中的退出不干净.我使用以下代码在 Jupyter 中启动它:

I've been having problems with tensorboard probably due to a unclean exit in windows10. I use the below code to launch it in Jupyter:

%load_ext tensorboard
%tensorboard --logdir={dir}

这是我得到的:

'ERROR: Timed out waiting for TensorBoard to start. It may still be running as pid 24472.'

在 cmd 窗口中使用以下代码在 Chrome 中启动 TB:

and below code in cmd window to launch TB in Chrome:

$ tensorboard --logdir {dir}'

这是我得到的:tensorboard_error

我做了一些研究并尝试了以下方法:

I've done some research and tried the following things:

1. -"taskkill/im tensorboard.exe/f" 杀死所有活动的pid命令
2. -删除了%TMP%.tensorboard-info"目录中的所有 pid-xxxx.info 文件.
3. -直接删除整个%TMP%.tensorboard-info"

1. -"taskkill /im tensorboard.exe /f" to kill all live pids in command
2. -deleted all the pid-xxxx.info files in the "%TMP%.tensorboard-info" directory.
3. -deleted the whole "%TMP%.tensorboard-info" directly

那没有帮助.

然后我尝试通过在命令窗口中添加 "--port 6005" 将端口更改为 6005 并且 TB 可以在 Chrome 下的localhost:6005.但它在 Jupyter 中不起作用.

Then I tried to change the port to 6005 by adding "--port 6005" in command window and TB could launch in Chrome under "localhost:6005. but it won't work in Jupyter.

知道如何让 TB 再次在 Jupyter 中运行吗?谢谢.

Any idea how I can get TB to run in Jupyter again? Thanks.

推荐答案

阅读此Github issue，您会发现在启动 Tensorboard 时手动指定主机显然可以解决问题.

Reading this Github issue, you can find that specifying the host manually when launching Tensorboard apparently does the trick.

代替

%tensorboard --logdir {logs_base_dir} 

运行

%tensorboard --logdir {logs_base_dir}  --host localhost

这篇关于Tensorboard 无法在端口 6006 上正常运行的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！