To trigger a test run on Nucleus, the new model will appear in the Nucleus Models tab. It should have a badge saying “Executable” on it.
After clicking on the model, a Run Model button will appear on the upper right side of the screen.
After clicking on this, a dataset selection pop-up will appear to run the model on. Upon clicking run, an inference job will be run.
The progress of the job can be tracked in the Jobs tab.
After the job is complete, a new model run should appear in the Runs section of the model page.
If the job fails, please see the error message attached to the Job and either resolve the problem in the code or ask Scale Support for debugging help. Note that these jobs could take between 10 mins to an hour to run, depending on how large the dataset is and the model's performance.
Updated 4 months ago