Generate Predictions

Step 1 - Generate Predictions

In this step, we will configure the Inference section, which allows you to get specific predictions from the Fine-tuned Model.

1. From the screen where we left off, proceed to the Settings tab of your newly created inference.
2. First, select a Fine-tuned model by clicking the Select fine-tuned model button. This will be the model used to generate predictions. Click Apply once selected.
3. We will skip the Scoring Audience configuration in this tutorial, as we want to calculate propensity scores for the entire population.
4. In the Scoring Schedule section, choose One-time scoring, and set it to Start immediately. This configuration mirrors what we did in the training stage for the Foundation Model.

The final screen at this stage should look like this:

Step 2 - Save Predictions to Snowflake Table

The final setup step is to define where the prediction results will be stored.

1. In the Scoring Output section, click the Define button (or Create New if available).
2. Select the HM_KAGGLE database.
3. Choose the PUBLIC schema.
4. Enter a name for your results table, e.g., results_propensity.
5. Click Apply to save your output configuration.

Your configuration screen should now look like this:

Final Step – Run All

Once everything is configured:

• Click the Activate button in the upper right-hand corner.
• Then click Run now to execute the scenario and generate predictions.

You’re done! 🎉 Your model will score the selected population and write results to the specified Snowflake table.