Build metrics (tail=1,2,3) distribution plotter for runs=[ ] #41
Description
To document in-depth hyper parameter searches, build a metrics distribution plotter similar to smooth.
Want to show that hyper-parameter value selection is done statistically robustly.
While only 1 or 2 will be discussed in detail in the documentation, could be done for multiple if time persists. If so, needs multiple runs (more than 4) for each parameter search (exhaustive). Seems like a task for the first week of May.