Rank competing hypotheses with online Bradley-Terry updates and LUCB intervals. Replaces elo-select. Use whenever there are 3 or more candidate hypotheses competing for the next experiment, or when the user wants to understand which branch is currently leading.
Receba novas skills de Pesquisa e Web toda segunda
1 email curto, só com as skills novas de Pesquisa e Web. 4 minutos de leitura, sem spam, cancela com 1 clique.
Você confirma o email no primeiro envio. Sem spam. Cancela com 1 clique.
Ler descrição completaRecolher descrição
BT Tournament
This skill is the V3.0 successor to elo-select. It runs an online Bradley-Terry tournament instead of one-shot Elo updates and exposes 95% confidence intervals so "low-confidence ties" stay visible.
When to invoke
Researcher subagent has just emitted >= 3 hypothesis nodes in one turn.
The user explicitly typed /bt-tournament.
The cockpit shows two candidates whose 95% intervals overlap and the user asks which to push first.
Workflow
Gather the candidate hy
[Description truncada. Veja o README completo no GitHub.]