
Reputation
Badges 1
11 × Eureka!ugh too bad. I changed the name of one of my metrics and now It's difficult to compare. It would help if it was possible (even by renaming metrics)
I need to group images in a similiar way to what is done by iteration
, but based on a string value which will be visible in the UI
because series
in already used. Think of something like showing debug images based on 2 independent parameters for the model
The iteration
in my use case doesn't make much sense (I'm evaluation a model)
thanks! so basically for experiments that are already finished I have no way to compare ATM, right?
I doesn't really matter to me. One solution I had in mind is that this can be done by the web client on demand, meaning you can manually (or using the Task
object) specify how many iteration constitute a single epoch, and instead of scaling the plots will just be subsampled (or interpolated)
right now the situation is problematic, because as I mentioned, I can't compare the training process between different batch sizes (or effective batch size, if I use a different number of GPUs)
Thanks for the awesome work!!!
Is the trains-server release backwards-compatible with earlier versions of trains?