Audits an ML training experiment for reproducibility, evidence integrity, leakage, checkpoint identity, evaluation validity, and overclaimed conclusions. Use before accepting results, spending on a longer run, publishing artifacts, or handing work to another agent.
Audits an ML training experiment for reproducibility, evidence integrity, leakage, checkpoint identity, evaluation validity, and overclaimed conclusions. Use before accepting results, spending on a longer run, publishing artifacts, or handing work to another agent.
This entry publishes no npm package, so Forge has no dependency tree for it. That is a gap in coverage — not a statement that it has no dependencies.