Running a model bake-off
Business
Running a model bake-off
- How to Run a Model Bake-Off: Script, Score and Sign-Off
- When to Accept a Live Call in Your Test Suite
- Building a Baseline Test Set from Production Traces
- How to Keep a Bake-Off Repo a Reviewer Can Rerun
- How to Sample Production Traffic for a Test Set
- How to Set a Cost Ceiling for a Model Evaluation Run
- Logging Two Providers Side by Side: A Cross-Run Comparison Schema
- Mocking an Agent's Tool Calls Without Mocking the Model
- What a Flaky Test Tells You About a Nondeterministic Model