core/evals
Shay 0fbcce429b
feat(adr-0182): cross-composer disagreement pooling — distractor 0014 + disguised-polarity refuse (confuser wrong 5->2) (#476)
Implements ADR-0182's first win on top of EX-6 (#473). The distractor-quantity
confuser 0014 misfired (20x3x5=300) because a blunt product-of-all was the *unique*
self-verifying reading: the completeness clause forces the distractor into it, and
no rival reading existed to trigger the wrong=0 disagreement rule. No tight cue rule
separates it from the legitimate cross-unit products (`for` licenses both the 0014
distractor AND the correct train-0021 product) -- that is the deferred cue-precision
problem. So instead of a reactive patch, let the disagreement rule do the refusing.

Mechanism (sealed lane; chat/ does not import these -> serving 3/47/0 frozen):
- verify.classify_derivation: a derivation is `complete` (commit-eligible),
  `exempt` (verified but for an isolated-foreign unused quantity -> commit-
  INELIGIBLE), or None. Refactored self_verifies into _base_reasons +
  _unused_quantities so the two share logic and cannot drift (behavior identical;
  385 derivation tests + smoke 67 green).
- accumulate.accumulation_candidates: exposes the ungated readings, incl. a
  distractor-skip reading that drops an isolated-foreign quantity from a multi-
  quantity change clause (20+5=25). compose_accumulation is byte-identical
  (drop_isolated_foreign=False + the same gate).
- search.multiplicative_candidates / multistep.candidate_chains: ungated candidates.
- pool.resolve_pooled: pools every composer's readings; disagreement -> refuse; a
  single answer commits only if a `complete` candidate produced it (exempt-only ->
  refuse, so the commit-path completeness guarantee from ADR-0175 is untouched).
- confuser runner: _engine_answer now delegates to resolve_pooled (the prior
  first-composer-wins order could not notice it held two incompatible readings).

Result (the microscope):
- confuser wrong 5 -> 2. distractor 0014 refuses (product 300 vs additive 25);
  BONUS: both disguised-polarity cases (0001/0003) refuse -- the spurious
  "buys X for N coins" product disagrees with the accumulation reading. Remaining
  wrong: 0016 (distractor in the anchor clause -> needs anchor-skip, separate step)
  and 0020 (temporal-scope). pair-tells 4 -> 1.
- genuine positives still 7 solved, 0 wrong.
- train_sample 3/47/0 and practice 3/47/0 byte-identical (they call
  compose_accumulation/search directly -- unchanged -- not the pool).
- smoke 67, architectural invariants 40, lane-SHA freeze 8/8.

Tests:
- test_adr_0182_pool.py: classify (complete/exempt/None incl. narrow-exemption
  edges) + resolve_pooled, with the wrong=0 obligation test_exempt_only_never_commits
  (a distractor with no multiplicative cue must refuse, not commit 25 -- fails loudly
  if the exemption is made commit-eligible).
- test_adr_0163_f2_confusers.py: baseline tightened wrong 5->2, pair_tells 4->1;
  new test_distractor_0014_refuses_via_pooling + test_disguised_polarity_does_not_misfire.

Stacked on #473 (EX-6); merge #473 first. 0016 + the remaining wrongs are follow-ups.
2026-05-29 13:22:19 -07:00
..
adversarial_identity
anchor_lens_tour parallel eval runner (#46) 2026-05-19 23:51:59 -07:00
anti_regression feat(epistemic): Phase 3 state tagging spine (#220) 2026-05-24 11:26:06 -07:00
articulation
articulation_of_status
audio_sensorium feat(adr-0181-p4): audio compiler eval gate lane (sensorium/audio) (#470) 2026-05-29 11:20:31 -07:00
audit_tour
calibration
classical_literature_ood
cognition parallel eval runner (#46) 2026-05-19 23:51:59 -07:00
cold_start_grounding fix(quarantine): drain all 60 quarantined tests — QUARANTINE=∅ (#267) 2026-05-25 11:22:12 -07:00
compositionality
compound_intent_decomposition
contemplation_quality feat(W-025): contemplation quality eval lane (ADR-0159) (#286) 2026-05-25 20:38:52 -07:00
contradiction_detection
conversation feat(demo): humanise teaching-grounded surface for layperson display 2026-05-19 14:14:02 -07:00
conversational_thread_coherence
cross_domain_transfer
curriculum_loop_closure feat(evals): ADR-0105 — sealed holdout encryption via age (#108) 2026-05-22 10:09:43 -07:00
demo_composition fix(W-015): replace _slerp_toward with rotor-geodesic anchor pull (#255) 2026-05-24 20:05:25 -07:00
deterministic_fluency
discourse_paragraph
domain_contract_validation chore(ci): re-pin drifted lane SHAs + refresh canonical reports (#229) 2026-05-24 14:25:11 -07:00
elementary_mathematics_ood feat(capability): ADR-0110 promote mathematics_logic to expert_demo (#118) 2026-05-22 12:59:23 -07:00
english_fluency_ood
fabrication_control chore(ci): re-pin drifted lane SHAs + refresh canonical reports (#229) 2026-05-24 14:25:11 -07:00
flywheel_demo feat(demo): core demo flywheel — public-proof reproduction of the loop 2026-05-27 21:33:54 -07:00
forward_semantic_control
foundational_biology_ood
foundational_physics_ood feat: ADR-0111 physics expert-demo promotion (second successful) 2026-05-22 14:37:36 -07:00
frontier_compare feat(contemplation): land ADR-0080 phase 1 (#119) 2026-05-22 13:10:03 -07:00
grammatical_coverage
gsm8k_math feat(adr-0182): cross-composer disagreement pooling — distractor 0014 + disguised-polarity refuse (confuser wrong 5->2) (#476) 2026-05-29 13:22:19 -07:00
gsm8k_parser_dev feat: add ADR-0125 perturbation suite 2026-05-22 17:12:33 -07:00
hebrew_fluency feat(packs): ADR-0103 — attach hebrew_fluency + koine_greek_fluency lanes to ADR-0102 (#106) 2026-05-22 09:43:46 -07:00
identity_divergence
industry_demos evals/industry_demos: add run_all.py suite runner 2026-05-21 08:23:29 -07:00
inference_closure feat: ADR-0122 systems_software audit-passed deferred (lane-shape mismatch) 2026-05-22 16:31:59 -07:00
introspection
koine_greek_fluency feat(packs): ADR-0103 — attach hebrew_fluency + koine_greek_fluency lanes to ADR-0102 (#106) 2026-05-22 09:43:46 -07:00
lab research(evals): phi separation probe for ADR-0081 follow-up (#57) 2026-05-20 12:34:59 -07:00
learning_arc fix(W-025): polish contemplation-quality eval lane follow-ups (#290) 2026-05-26 09:39:18 -07:00
learning_loop chore(evals): contracts + bench json + Lane B viewer + chart + audit + demo schema (#62) 2026-05-20 13:53:13 -07:00
long_context_cost chore(evals): contracts + bench json + Lane B viewer + chart + audit + demo schema (#62) 2026-05-20 13:53:13 -07:00
math_bounded_grammar/v1 feat(ADR-0131.3): bounded-grammar word-problem benchmark — lane PASSED 50/50 (#180) 2026-05-23 11:27:04 -07:00
math_capability_axes feat(ADR-0163.D.4): question grammar extension — mass nouns, comparatives, pronoun-entity resolution (#310) 2026-05-26 16:19:37 -07:00
math_expert_claims/v1 fix(quarantine): drain all 60 quarantined tests — QUARANTINE=∅ (#267) 2026-05-25 11:22:12 -07:00
math_symbolic_equivalence feat(ADR-0131.1.F): frontier-baseline comparison harness for B1 (#178) 2026-05-23 12:14:06 -07:00
math_teaching_corpus/v1 feat(ADR-0131.2.B): B2 teaching-corpus enrichment — load-bearing gate (#177) 2026-05-23 11:29:48 -07:00
miner_loop_closure feat(teaching): implement ADR-0095 — Miner-Sourced Teaching Proposals 2026-05-21 18:18:51 -07:00
monotonic_learning
multi_agent_composition
multi_sentence_response
multi_step_reasoning
obligation_2_ood_ratio feat(ADR-0114a.2): OOD-ratio auditor — Obligation #2 wired for B3, ratio=1.00 (#193) 2026-05-23 16:25:28 -07:00
obligation_5_perturbation feat(ADR-0114a.5): reasoning-isolation perturbation suite — Obligation #5 wired for B3, PASSING 130/130 preserving, 68/68 breaking (#191) 2026-05-23 16:07:59 -07:00
obligation_6_depth_curve feat(ADR-0114a.6): depth-curve auditor — Obligation #6 wired for B3 (assertion holds, coverage gap named) (#190) 2026-05-23 16:19:58 -07:00
obligation_8_adversarial feat(ADR-0114a.8): adversarial auditor — Obligation #8 wired, PASSING; surfaces 2 known parser-layer gaps (#192) 2026-05-23 16:11:37 -07:00
obligation_10_pack_provenance feat(ADR-0114a.10): pack-provenance auditor — Obligation #10 wired for B3, PASSING 2026-05-23 15:44:53 -07:00
orthogonality_tour parallel eval runner (#46) 2026-05-19 23:51:59 -07:00
prompt_diversity feat(adr-0085): gloss-aware CAUSE composer — explanation frame from glosses (#70) 2026-05-20 15:55:08 -07:00
provenance
public_demo chore(ratify): accept four Phase C round-2 recognizers (round 2) (#309) 2026-05-26 16:03:01 -07:00
realizer_guard chore(evals): contracts + bench json + Lane B viewer + chart + audit + demo schema (#62) 2026-05-20 13:53:13 -07:00
refusal_calibration
refusal_taxonomy feat(ADR-0163.B.2): four new exemplar corpora — discrete_count_statement, multiplicative_aggregation, currency_amount, plus temporal_aggregation v2 widening (#306) 2026-05-26 14:36:59 -07:00
register_diagnostics feat(evals): per-intent register-firing diagnostic + CI gate + tests (#103) 2026-05-21 07:05:23 -07:00
register_tour feat(register): ADR-0077 — substantive register knobs + layering boundary (R6) 2026-05-19 23:39:11 -07:00
reports
results feat: ADR-0086 + ADR-0087 + 100-register catalog — cognition lane closure 2026-05-21 00:08:12 -07:00
reviewer_registry feat(capability): implement ADR-0092 — Reviewer Registry v1 2026-05-21 18:01:24 -07:00
sample_efficiency
self_consistency_over_time
symbolic_logic feat: ADR-0122 systems_software audit-passed deferred (lane-shape mismatch) 2026-05-22 16:31:59 -07:00
teaching_injection_resistance
walkthrough_chain
warmed_session_consistency
zero_code_domain_acquisition
__init__.py
_parallel.py parallel eval runner (#46) 2026-05-19 23:51:59 -07:00
baseline_runner.py
CLAIMS.md chore(evals): contracts + bench json + Lane B viewer + chart + audit + demo schema (#62) 2026-05-20 13:53:13 -07:00
cognition_cases.jsonl
framework.py feat(evals): ADR-0105 — sealed holdout encryption via age (#108) 2026-05-22 10:09:43 -07:00
holdout_runner.py feat(evals): ADR-0105 — sealed holdout encryption via age (#108) 2026-05-22 10:09:43 -07:00
metrics.py
parallel.py parallel eval runner (#46) 2026-05-19 23:51:59 -07:00
run_cognition_eval.py feat(W-008): L10 Shape B hybrid engine-state persistence (#271) 2026-05-25 11:45:54 -07:00