Skip to content

Weighted random selections#10

Open
agusmba wants to merge 5 commits intoRoman-:masterfrom
agusmba:weighted-random-selections
Open

Weighted random selections#10
agusmba wants to merge 5 commits intoRoman-:masterfrom
agusmba:weighted-random-selections

Conversation

@agusmba
Copy link

@agusmba agusmba commented Jan 11, 2026

Each selected case starts with equal probability of being randomly selected (1). After each solve, if the time is faster than the goal, the probability of choosing the same case is halved. Otherwise it's doubled. This way you can train on a few cases and the weighted random selection will gravitate towards those that are still not fully learned (fast enough).

The default now is 0 (deactivated). Also, there is an explanation on hover:
image

This PR is based on my previous one so it also incorporates the oll times order fix.

Let me know if you want additional changes.

fix oll number order in times box
Assign same probability to all selected cases.
If selected case is solved faster than goal ms,
divide its probability in half. Otherwise double it.
Deactivate weights as a default and explain how to use the feature.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant