Article source | turbine cloud community (ai/deep learning cloud GPU server address training platform, official experience url: gpushare.com/)

Visdom is a PyTorch data visualization tool developed by Facebook

How do you install it on a cloud server? This tutorial will briefly cover ~

【 for 】

Before installation, please be reminded that the hengyuan cloud platform instance does not provide public IP address. The services are connected to the public network access point through port mapping

Therefore, when using Visdom, you need to stop the TensorBoard service, configure the Visdom port to be the same as port 6006 of TensorBoard, and use 0.0.0.0 for the listening address.

“Installation”

After completing the above steps, the installation is ready

Enter the following command on the terminal:

PIP install visdom # Prevent the download sed -i "s/HTTPS: / / / / \ \ (cdnjs.cloudflare.com *)/HTTP: / / / / \ 1 /" "$(PIP show visdom | grep Location | awk '{print $2}')/visdom/server.py" is the small container handling the TensorBoard container. Use the following command to configure the grep -e "autostart"/etc/supervisor/conf. D/tensorboard. Conf | | echo "autostart = false" > > / etc/supervisor/conf. D/tensorboard. Conf supervisorctl update # start Visdom service Visdom -- the hostname 0.0.0.0 - port 6006Copy the code

Once installed, click on the TensorBoard link to enter Visdom

In theory, other Web services could connect in this way, and replacing Visdom with other tools would work

[use]

Note that the Visdom service is started locally on port 6006. When the Python client is initialized, you need to add port=6006 as follows:

import visdom import numpy as np vis = visdom.Visdom(port=6006) vis.text('Hello, world! ')Copy the code