Ability to compare/overlay graphs between runs


When running multiple models simultaneously (or comparing a currently training model with a finished one), it is very useful to be able to compare their training graphs while it is still running.

While it is possible to overlay two channels/graphs from the same experiment, there doesn’t seem to be a way for me to directly compare the training graphs of two models - I am just having to put them side by side and compare values.

Being able to create a charts set which spans across experiments and allows me to compare models would be a huge improvement to my workflow when iterating through different models.


Hi Mikel, I agree and plan to add this feature. Will get back to you soon with ETA.