Quick Start

Create an experiment and run your first model evaluation

This quick start shows the fastest path to run one evaluation and inspect results.

Step 1: Open Evaluations

1. Sign in to Bud AI Foundry.
2. Go to Evaluations in the sidebar.
3. Keep the Experiments tab open for run management.

Step 2: Create an Experiment

1. Click New experiment.
2. Enter:
   • Name: Baseline Reasoning Comparison
   • Description: Compare candidate models on reasoning traits
   • Tags: baseline, reasoning
3. Save the experiment.

Step 3: Start a Run

1. Open the experiment you created.
2. Click Run Evaluation.
3. Select:
   • A model (deployment or supported target)
   • One or more traits
   • Datasets mapped to those traits
4. Confirm and start the run.

Step 4: Monitor Progress

1. Watch run status in the experiment table.
2. Open run details to review:
   • Status and duration
   • Trait-level scores
   • Dataset-level benchmark summary

Step 5: Compare and Decide

1. Open the dataset in Evaluations Hub.
2. Use Leaderboard for cross-model ranking.
3. Use Evaluations Explorer for prompt/response-level inspection.

Use consistent experiment tags (for example: release-candidate, nightly) so filtering stays clean as runs scale.

Next Steps

Creating Your First Evaluation

Follow a production-style walkthrough

Troubleshooting

Resolve common run and filter issues quickly