Training overview and run history
Find saved runs and understand which result and settings belong together.
Before you start
Select the project and the output roots that contain its runs.
Quick start
- Open Overview or Training & Run History and choose a saved run.
- Review run state, data identity, settings and result evidence. Continue to Evaluation for the chosen backend.
Functions and choices
Check run status
Newest first is the default date order. A failed or interrupted run is distinct from a completed model.
Fix missing paths
If folders or evidence cannot be read, use Fix paths to open Settings → Storage and inspect the saved training locations.
Model Journal
Model Journal in the header opens the current project’s model records.
Read the timing
Total elapsed time covers the whole run, including preparation and evaluation, and is restored from the run history. The remaining-time estimate covers training iterations.
Judge model quality
Loss describes the run, not performance on unseen images.
Result
A selected run with identifiable provenance.
Next step
Evaluate its predictions and compare with earlier runs.