Eval result dataframe is missing confidence scores of model predictions #3887
Labels
1.x
P2
Medium priority, add to the next sprint if no P1 available
topic:eval
topic:pipeline
topic:retriever
type:feature
New feature or request
wontfix
This will not be worked on
Is your feature request related to a problem? Please describe.
When evaluating I want to see scores (confidence scores) associated with each model prediction.
With these scores, you can determine thresholds for only showing good model predictions. This is useful if you configure a search system for non technical endusers and want to filter predictions below a certain threshold.
Unfortunately, pipeline.eval results do not contain those scores.
Describe the solution you'd like
Please add the model prediction scores to the pipeline.eval results.
Tagging @sjrl since he experienced the same issue, as well as @tstadel because he might know a quick fix for this. :D
The text was updated successfully, but these errors were encountered: