log directory execution in tensorboard

Question

I am working on a keras tutorial (Using Anaconda spyder 3.6 environment for TF on windows 10) for Mnist Convolution nets. to recognize digits, given at : www.blog.keras.io/building-autoencoders-in-keras.html

the problem i am facing is to monitor tensorboard log

in my code i use,

tensorboard=TensorBoard(log_dir="logs/{}".format(time)))

I do add the tensorboard in my keras model.fit( ,callbacks=[tensorboard])

this create the logs folder in my anaconda spyder working directory (which i can see).

However the next step is to open a terminal and run :

tensorboard --logdir=logs/

Q) where or what should be the terminal? i tried console in spyder and anaocnda prompt but none works. From spyder i get errors like invalid syntax or no operator assigned.

Please guide me considering a newbie here.

Much appreciated :)

Answer 1

The terminal is the console, where you can execute commands. As you have installed anaconda on windows, you can open the anaconda prompt (windows start menu). First make sure to navigate to your project directory. e.g. cd C:/projects/myProject

tensorboard --logdir=logs should work then. If it says something like "unrecognized command", tensorboard is not installed. In this case type conda install tensorboard. Make sure you use the same virtual environment as your project.