Loop
Loop trains models that update from live predict/train traffic. It uses its own wizard and does not consume Lake gold today.
Before you start
Signed-in account. Prefer Loop when predictions feed back as training signal (recommendations, moderation, adaptive scoring). For a static spreadsheet, use ML instead.
Sidebar → Data Intelligence → Loop (/loop)
Compared to ML and Forecast
| Loop | ML | Forecast | |
|---|---|---|---|
| Learning | Incremental from live data | Batch on gold | Batch on time series |
| Data | Loop wizard | Lake gold or samples | Lake gold or AirPassengers |
Flow
- Define and train
- Deploy (API key)
- Predict, then send outcomes on the train endpoint (API guide)
- Monitor
Common mistakes
| Mistake | Fix |
|---|---|
| Looking for Loop datasets in Lake gold | Use the Loop wizard |
| No train calls after predict | Feedback never reaches the model |