Assign color to scalars when logging to tensorboard logger

I’ve recently begun to convert my models over to pytorch-lightning and am trying to take advantage of the logger (default: tensorboard).

The logger seems to randomly assign colors to the scalars for every run which becomes awful messy when comparing various metrics and runs. I can go through and set them by hand after they show up in tensorboard, but I would really like to assign them at the time I’m logging the value.

I would like to be able to assign all train metrics one color and all validation metrics a different color, for example. So it’s clear from chart to chart which one I’m looking at. Then I could alter those for different model runs so that even with several runs showing in one chart I could clearly see each model’s train/validation metric curves.

is there a way to do this? I can’t seem to find any indication in the documents.