arborist/bench
russell@unturf.com e561dfc599
bench/qa: refresh question set; live test updates Mars adversarial
Question-set churn (bench/qa_questions.txt):
- narrow factoid: drop 'who founded apple computer?'; add
  'a bridge between new london & groton?' (single-answer factoid
  with named-entity grounding via Gold Star Memorial Bridge)
- broad descriptive: drop 'tell me about the apollo program'
- entity list: drop 'list the founders of microsoft'; add
  'name simpsons family members including pets?',
  'list of obelisk in connecticut'
- leading/adversarial: add 'whales are endangered from over hunting'
  (declarative-shape, tests verifier under premise-seeking phrasing)
- out-of-corpus: rephrase Mars no-such-entity question to
  'who is a benevolent dictator for life for mars?' — sharper
  adversarial shape since BDFL is a real OSS-governance term
  the model is tempted to project onto Mars
- minor edits: 'GNU linux', 'C programming language', 'method man',
  'cold fusion breakthrough' phrasing

tests/test_qa_quality_live.py: rename test_mars_prime_minister_*
to test_mars_benevolent_dictator_*; refresh refutation markers
to match the new BDFL phrasing; expand to cover 'no one' /
'no person' negations the model uses on this shape.
2026-04-30 16:58:00 -04:00
..
qa_questions.txt bench/qa: refresh question set; live test updates Mars adversarial 2026-04-30 16:58:00 -04:00
qa_sweep.py qa: 3-mode bench, HTTP retry, trim-and-verify, doc snapshot 2026-04-30 14:40:57 -04:00
run.sh progress reporter + structured benchmark 2026-04-27 11:37:20 -04:00