Reading Probabilities and Confidence
Confidence is derived from the distribution, not reported separately. Drag the sliders until that clicks.
Confidence is not a separate thing the model reports. It is a statistic computed from the probability distribution you already received. All the mass on one option gives you 1.0. A flat spread gives you close to 0.
Choice · three options
How the distribution shapes confidence
confidence
Act automatically
This demo approximates confidence as (n × peak − 1) / (n − 1), the formula TypeSafe uses in its own docs demo. The API computes the real value for you and returns it on every Choice and Score answer. Noul answers carry no confidence field — a two-outcome distribution is fully described by the single noul value.
Full lesson
Keep reading Reading Probabilities and Confidence
The rest of this lesson — including the interactive exercises and the worked project — is part of Jev Foundations.
- ✓Every lesson in all three courses
- ✓Interactive builders, graders, and playgrounds
- ✓Project source you can run against your own API key
- ✓Free updates as Jev ships new versions
Already bought them? Sign in to unlock.