Run camp-full_map-ultra-002/adversarial_belief

Run camp-full_map-ultra-002/adversarial_belief: passed essentially everything (99%) -- this test did not challenge the model.

Of 128 scored trials (belief query kind), the model passed 99%. That's saturation: the difficulty range probed here isn't finding this model's limits. Consider a harder preset (breaking_point or adversarial_belief) or a wider dimension range.

2 additional trial(s) ended in ERROR (infra failures, cost cap, or cancellation) and are excluded from the pass rate above.

1 failures recorded; all 1 survived verification.

1 belief failure(s) were classified against known shortcuts, but none of them dominate -- the failures don't share one obvious wrong-answer pattern.

Notable failures

• record 7b713cba — asked "Where does Jamal believe the dusty folder is? Answer with only the location name (or 'unknown' if they have no belief)."
  expected: the photo studio
  actual: unknown

  the model's answer matches no known shortcut -- an unclassified miss