Generating loguniform distribution of hyperparameters and saving pictures

Hi all!

How can I make use of a loguniform distribution of hyperparameters?
Now I’m defining, for example the learning rate like:

How I could make a log uniform distribution? hp.loguniform does not work. I’m applying this strategy now:

Also, how would I upload images into tensorboard? I can see them all when I’m applying in Jupyter!

%tensorboard --logdir logs/train_data

However, when I’m uploading my images it is not recognizing my files; my files are jpg values, just as it has been done here: ‘Logging arbitrary data’ : Displaying image data in TensorBoard  |  TensorFlow

My function :

I log it in my code with:

Generating my fig by

After running experiments, I upload it by this code:


Which is false, since there are files in there:


Each map has:


I’d be happy to hear from you!

Kind regards,