Basics and Overview

What's the difference between Artifacts and Attributes, since both add information to Trials?

Artifacts store the actual files, and only the file path is linked to the Trial. Attributes are stored directly in the Trial as string information.

With so many optimization methods, which one should I choose?

Since v1.0, Tunny includes AutoSampler, which automatically selects an appropriate method for the problem. We recommend trying this first.

Where are the optimization results stored?

By default, results are stored in the fish.log file on the desktop, and artifacts in the artifacts folder. You can see which log file path is used to save the results at the top of the UI.

As results accumulate, the log file takes longer to load when TunnyUI launches. We recommend re-creating the log file as a new one at regular intervals.

From the Quick Access at the top of the UI, you can create a new results file or re-reference previous files.

Manage your results files so they do not grow too large. Overly large files can affect optimization performance.

Can I open the dashboard quickly, just to check the results?

Click Tunny in the menu bar at the top of the window, and you will see "Run Optuna Dashboard..." there. You can launch the dashboard from there, or use the TunnyRunDashboard command in Rhino.

What is AUTO Sampler?

The AUTO sampler automatically selects a sampling method that performs above a certain level, depending on the problem. It gives a certain level of performance, but note that it is not necessarily the best sampler.

For more information, see the following:

AutoSampler: Automatic Selection of Optimization Algorithms in Optuna