1. Running TensorBoard remotely

• When working on a remote server, you can use SSH tunneling to forward the port of the remote server to your local machine at port (port 6006 in this example)

  • containner on remote server

ssh -L 6006:localhost:6006 containner_name 

or

• remote server_name (eg. puffer)

ssh -L 6006:localhost:6006 puffer 

• Then launch on remote server with 2.1 or 2.2 command

2. On local machine

• 2.1 Normal tensorboard

tensorboard --logdir=runs

• 2.2 Dev Tensorboard (able to share link)

tensorboard dev upload --logdir runs

• write log to runs file

python3 demo2.py