Run camp-full_map-gpt-oss-120b-001/belief

Run camp-full_map-gpt-oss-120b-001/belief: 97% pass rate on belief queries.

Of 159 scored trials (belief query kind), the model passed 97% and failed 3%.

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

5 failures recorded; all 5 survived verification.

Diffuse weakness (below 90% but no outright flip) on: witness_gap (pass rate 80% at the high end).

In 3 of 5 classifiable belief failures the model reported where the object ACTUALLY was, not where the queried entity BELIEVED it was -- it collapsed their perspective into reality.

Notable failures

• record 89641036 — asked "Where does Dexter believe the faded box is? Answer with only the location name (or 'unknown' if they have no belief)."
  expected: the archive room
  actual: print room

  the model reported the object's TRUE final location, not what the queried entity actually believed -- it collapsed belief into reality

• record 52ea26a5 — asked "Where does Elliot believe the dusty wrench is? Answer with only the location name (or 'unknown' if they have no belief)."
  expected: the east wing
  actual: conference room

  the model reported the object's TRUE final location, not what the queried entity actually believed -- it collapsed belief into reality

• record 04f097b3 — asked "Where does Quinn believe the narrow wrench is? Answer with only the location name (or 'unknown' if they have no belief)."
  expected: the utility closet
  actual: library

  the model reported the object's TRUE final location, not what the queried entity actually believed -- it collapsed belief into reality