run

Definition

To execute something, and as a noun, one complete execution. In training, a run is a single end-to-end experiment under one configuration, and results are properly reported across several runs with different random seeds because seed-to-seed variance can be substantial. A small improvement backed by a single run is weak evidence, which is why reviewers ask for variance and why ablations should share the same seed protocol.

Explainers using this term

• Paper walkthrough: StateM — 95.3% on Terminal-Bench 2.1 and a USD 15 run, without touching a single weightStateM: Reaching 95.3% Raw Accuracy, or a \$15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling