Pause a low-prevalence Label during Signal AI Studio training to unblock the system’s search for higher-prevalence labels.When training a model in Signal AI Studio, the system automatically surfaces calls to help you find examples for your labels. However, if a label has very low prevalence — meaning the specific behavior or phrase rarely occurs in your 5,000-call sample — training can get stuck in a continual loop of no calls found. Pausing these labels lets the system focus on your higher-prevalence data and unblocks the training flow.
Why you should pause a label
Consider pausing a label if you notice:- Stuck training: the system can’t surface new calls for you to label.
- Low prevalence: a label shows 0 or 1 found examples at the bottom of your training list.
- Too many labels: you’ve created multiple small labels competing for a limited number of call examples.
- Stop surfacing calls for that label during labeling sessions.
- Ignore current true/false counts for that label.
- Ignore current true/false call counts toward the “Verified Calls” total.
- Pause accuracy tracking for that label.
- Preserve its labeled call history.
How to pause a label
- Log in to the Invoca platform and click the Signal tab, then Signal AI Studio.
- Select the model you’re currently training.
- Locate the label list in the training interface and select Verify Labels. The Transcript and Active Labels appear.
- Click the three-dot menu next to the label you want to stop, in the Active Labels section.
- Select Pause Label.
- The label grays out and moves into a new “Paused Labels” section below the list of labels.
- Repeat this for other low-prevalence labels until the system begins surfacing calls again.
How to unpause a label
- Navigate to your model training screen.
- Locate the grayed-out paused label.
- Click the three-dot menu next to the label name.
- Select Unpause Label.
- Click Save to refresh the training logic.