gforman44

This is pretty important. I'd like to see three criteria: 1. minimum number of epochs (e.g. 2), no matter what. 2. stop if the validation set doesn't improve for earlyStopPatience...

Here's a proposal for what I'd like to see: Parameters for flexible stopping criteria: ``` static int maxEpochs = 1000; static double objectiveSuccess = 0.5;// done if validation loss objective...

It is a Loss, not just an Evaluator. I put this reference in the java doc for further explanation. https://bibinmjose.github.io/2021/03/08/errorblog.html Yes, it deserves a test case. Harder with non-deterministic learning....

Oh. Thank you for that clarification. That’ll be easy. George On Fri, Jun 17, 2022 at 8:19 PM Zach Kimberg ***@***.***> wrote: > I'm not sure what you mean by...

One more thing before the framework gets too ossified: consider making Loss an *interface* that is a sub-interface of Evaluator *interface*. Then any Loss can be used as an additional...

It is a Loss, not just an Evaluator.** I felt some of the questions were already addressed in the java doc, such as when you want to train for the...

Jason, I just added a right-click context menu to my chart.  Unfortunately, when I drag the chart around with the right button, it ends the drag by opening the context...

Hi, Jason, >  If I recall, jfxutils doesn't consume the events in most places. > Righto.  I'm using a mouse event filter that uses left-drag for zooming and right-drag for...