Use when packaging the artifacts of a COLM paper — model weights, training data, prompts, evaluation sets, and cached model outputs — for anonymous review and public post-acceptance release, navigating licenses, API terms-of-service limits, and the absence of a formal COLM artifact track.
