How To Run Jupyter Notebooks And Tensorboard On A Remote Server At The Same Time

This works for me. First you need to ssh into your remote machine with:

ssh -L 8001:localhost:8888 -L 16006:127.0.0.1:6006 leo@my_server_ip

The command -L 8001:localhost:8888 transfers the port 8888 of the remote server into the port 8001 of the local machine you’re using. The command -L 16006:127.0.0.1:6006 transfers the port 6006 of the remote server into the port 16006 of the local machine you’re using.

Then, you can launch tensorboard on your remote server with

--logdir <logs_directory> --port 6006

and on your local machine go to http://127.0.0.1:16006 in your browser. Now you can launch jupyter lab (for example) on your remote server by navigating to the top directory you want open and typing:

jupyter lab

and then on your local machine go to http://localhost:8001/ in your browser.