From 93bbb6c824ab63fc22d86dd0bd885f1cd3dd005d Mon Sep 17 00:00:00 2001 From: Shay Date: Sat, 16 May 2026 06:23:28 -0700 Subject: [PATCH] feat(evals,packs): grammatical-coverage holdout, zero-code kits, Hebrew/Greek packs - grammatical-coverage holdout v1: 52 cases across all 13 constructions, 100% pass - zero-code-domain-acquisition lane: contract + 3 surprise domains (kinship, calendar, color) with vocabulary, relations, axioms, teaching examples, and dev prompts; pack closure verified for all three domains - he_core_cognition_v1: 20 entries in Hebrew script with morphology decomposition (triliteral roots, binyanim, aspect/person/gender/number); depth_root role with fail_closed OOV policy - grc_logos_cognition_v1: 20 entries in polytonic Greek with morphology decomposition (stems, prefix/suffix chains, declension class, tense/voice/ mood/person); depth_relation role with fail_closed OOV policy --- docs/PROGRESS.md | 6 +- .../holdouts/v1/cases.jsonl | 52 +++++++++++++++ .../zero_code_domain_acquisition/contract.md | 63 +++++++++++++++++++ .../domains/README.md | 48 ++++++++++++++ .../domains/calendar/axioms.jsonl | 17 +++++ .../domains/calendar/prompts_dev.jsonl | 10 +++ .../domains/calendar/relations.txt | 4 ++ .../domains/calendar/teaching_examples.jsonl | 20 ++++++ .../domains/calendar/vocabulary.jsonl | 26 ++++++++ .../domains/color/axioms.jsonl | 20 ++++++ .../domains/color/prompts_dev.jsonl | 10 +++ .../domains/color/relations.txt | 7 +++ .../domains/color/teaching_examples.jsonl | 20 ++++++ .../domains/color/vocabulary.jsonl | 21 +++++++ .../domains/kinship/axioms.jsonl | 15 +++++ .../domains/kinship/prompts_dev.jsonl | 10 +++ .../domains/kinship/relations.txt | 7 +++ .../domains/kinship/teaching_examples.jsonl | 20 ++++++ .../domains/kinship/vocabulary.jsonl | 26 ++++++++ .../data/grc_logos_cognition_v1/lexicon.jsonl | 20 ++++++ .../data/grc_logos_cognition_v1/manifest.json | 13 ++++ .../grc_logos_cognition_v1/morphology.jsonl | 20 ++++++ .../data/he_core_cognition_v1/lexicon.jsonl | 20 ++++++ .../data/he_core_cognition_v1/manifest.json | 13 ++++ .../he_core_cognition_v1/morphology.jsonl | 20 ++++++ 25 files changed, 505 insertions(+), 3 deletions(-) create mode 100644 evals/grammatical_coverage/holdouts/v1/cases.jsonl create mode 100644 evals/zero_code_domain_acquisition/contract.md create mode 100644 evals/zero_code_domain_acquisition/domains/README.md create mode 100644 evals/zero_code_domain_acquisition/domains/calendar/axioms.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/calendar/prompts_dev.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/calendar/relations.txt create mode 100644 evals/zero_code_domain_acquisition/domains/calendar/teaching_examples.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/calendar/vocabulary.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/color/axioms.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/color/prompts_dev.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/color/relations.txt create mode 100644 evals/zero_code_domain_acquisition/domains/color/teaching_examples.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/color/vocabulary.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/kinship/axioms.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/kinship/prompts_dev.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/kinship/relations.txt create mode 100644 evals/zero_code_domain_acquisition/domains/kinship/teaching_examples.jsonl create mode 100644 evals/zero_code_domain_acquisition/domains/kinship/vocabulary.jsonl create mode 100644 language_packs/data/grc_logos_cognition_v1/lexicon.jsonl create mode 100644 language_packs/data/grc_logos_cognition_v1/manifest.json create mode 100644 language_packs/data/grc_logos_cognition_v1/morphology.jsonl create mode 100644 language_packs/data/he_core_cognition_v1/lexicon.jsonl create mode 100644 language_packs/data/he_core_cognition_v1/manifest.json create mode 100644 language_packs/data/he_core_cognition_v1/morphology.jsonl diff --git a/docs/PROGRESS.md b/docs/PROGRESS.md index 156dbc2a..17f54a18 100644 --- a/docs/PROGRESS.md +++ b/docs/PROGRESS.md @@ -47,12 +47,12 @@ Tracks completion of the phased plan defined in `docs/capability_roadmap.md` **Started:** 2026-05-16 **Depends on:** Phase 0 exit -- [ ] **grammatical-coverage** lane +- [x] **grammatical-coverage** lane (v1 complete) - [x] Enumerate English v1 constructions (13 constructions: C01-C13) - [x] Write contract test pairs (PropositionGraph -> surface family) - [x] Implement v1 dev/public (~41/36 items) - - [ ] Implement holdout (~50 items) - - [ ] Engineer `realizer.py` to pass v1 (baseline: dev=24%, public=19%) + - [x] Implement holdout (52 items) — 100% pass + - [x] Engineer `realizer.py` to pass v1 (dev=100%, public=100%, holdout=100%) - [ ] Generate v2 on pass - [ ] Hebrew pack - [ ] Koine Greek pack diff --git a/evals/grammatical_coverage/holdouts/v1/cases.jsonl b/evals/grammatical_coverage/holdouts/v1/cases.jsonl new file mode 100644 index 00000000..bcd7969c --- /dev/null +++ b/evals/grammatical_coverage/holdouts/v1/cases.jsonl @@ -0,0 +1,52 @@ +{"id": "gram_C01_h01", "construction": "C01", "construction_name": "simple_declarative", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "stillness", "predicate": "precedes", "obj": "motion"}], "edges": []}, "accept_surfaces": ["stillness precedes motion"], "constraints": {"must_contain": ["stillness", "precedes", "motion"], "word_order": ["stillness", "precedes", "motion"], "max_words": 6}} +{"id": "gram_C01_h02", "construction": "C01", "construction_name": "simple_declarative", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "memory", "predicate": "supports", "obj": "identity"}], "edges": []}, "accept_surfaces": ["memory supports identity"], "constraints": {"must_contain": ["memory", "supports", "identity"], "word_order": ["memory", "supports", "identity"], "max_words": 6}} +{"id": "gram_C01_h03", "construction": "C01", "construction_name": "simple_declarative", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "order", "predicate": "implies", "obj": "purpose"}], "edges": []}, "accept_surfaces": ["order implies purpose"], "constraints": {"must_contain": ["order", "implies", "purpose"], "word_order": ["order", "implies", "purpose"], "max_words": 6}} +{"id": "gram_C01_h04", "construction": "C01", "construction_name": "simple_declarative", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "practice", "predicate": "requires", "obj": "discipline"}], "edges": []}, "accept_surfaces": ["practice requires discipline"], "constraints": {"must_contain": ["practice", "requires", "discipline"], "word_order": ["practice", "requires", "discipline"], "max_words": 6}} +{"id": "gram_C02_h01", "construction": "C02", "construction_name": "negation", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "silence", "predicate": "implies", "obj": "consent", "negated": true}], "edges": []}, "accept_surfaces": ["silence does not imply consent"], "constraints": {"must_contain": ["silence", "does", "not", "consent"], "word_order": ["silence", "does", "not", "consent"], "max_words": 8}} +{"id": "gram_C02_h02", "construction": "C02", "construction_name": "negation", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "form", "predicate": "requires", "obj": "substance", "negated": true}], "edges": []}, "accept_surfaces": ["form does not require substance"], "constraints": {"must_contain": ["form", "does", "not", "require", "substance"], "word_order": ["form", "does", "not", "substance"], "max_words": 8}} +{"id": "gram_C02_h03", "construction": "C02", "construction_name": "negation", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "habit", "predicate": "defines", "obj": "character", "negated": true}], "edges": []}, "accept_surfaces": ["habit does not define character"], "constraints": {"must_contain": ["habit", "does", "not", "define", "character"], "word_order": ["habit", "does", "not", "character"], "max_words": 8}} +{"id": "gram_C02_h04", "construction": "C02", "construction_name": "negation", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "certainty", "predicate": "follows", "obj": "doubt", "negated": true}], "edges": []}, "accept_surfaces": ["certainty does not follow doubt"], "constraints": {"must_contain": ["certainty", "does", "not", "follow", "doubt"], "word_order": ["certainty", "does", "not", "doubt"], "max_words": 8}} +{"id": "gram_C03_h01", "construction": "C03", "construction_name": "conjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "patience", "predicate": "reveals", "obj": "strength"}, {"node_id": "n2", "subject": "haste", "predicate": "reveals", "obj": "weakness"}], "edges": [{"source": "n1", "target": "n2", "relation": "conjunction"}]}, "accept_surfaces": ["patience reveals strength and haste reveals weakness"], "constraints": {"must_contain": ["patience", "and", "haste"], "word_order": ["patience", "and", "haste"], "max_words": 12}} +{"id": "gram_C03_h02", "construction": "C03", "construction_name": "conjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "order", "predicate": "grounds", "obj": "reason"}, {"node_id": "n2", "subject": "chaos", "predicate": "grounds", "obj": "creativity"}], "edges": [{"source": "n1", "target": "n2", "relation": "conjunction"}]}, "accept_surfaces": ["order grounds reason and chaos grounds creativity"], "constraints": {"must_contain": ["order", "and", "chaos"], "word_order": ["order", "and", "chaos"], "max_words": 12}} +{"id": "gram_C03_h03", "construction": "C03", "construction_name": "conjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "breath", "predicate": "supports", "obj": "life"}, {"node_id": "n2", "subject": "stillness", "predicate": "supports", "obj": "reflection"}], "edges": [{"source": "n1", "target": "n2", "relation": "conjunction"}]}, "accept_surfaces": ["breath supports life and stillness supports reflection"], "constraints": {"must_contain": ["breath", "and", "stillness"], "word_order": ["breath", "and", "stillness"], "max_words": 12}} +{"id": "gram_C03_h04", "construction": "C03", "construction_name": "conjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "memory", "predicate": "precedes", "obj": "narrative"}, {"node_id": "n2", "subject": "narrative", "predicate": "precedes", "obj": "meaning"}], "edges": [{"source": "n1", "target": "n2", "relation": "conjunction"}]}, "accept_surfaces": ["memory precedes narrative and narrative precedes meaning"], "constraints": {"must_contain": ["memory", "and", "narrative"], "word_order": ["memory", "and", "narrative"], "max_words": 12}} +{"id": "gram_C04_h01", "construction": "C04", "construction_name": "disjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "courage", "predicate": "precedes", "obj": "action"}, {"node_id": "n2", "subject": "fear", "predicate": "precedes", "obj": "paralysis"}], "edges": [{"source": "n1", "target": "n2", "relation": "disjunction"}]}, "accept_surfaces": ["courage precedes action or fear precedes paralysis"], "constraints": {"must_contain": ["courage", "or", "fear"], "word_order": ["courage", "or", "fear"], "max_words": 12}} +{"id": "gram_C04_h02", "construction": "C04", "construction_name": "disjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "structure", "predicate": "defines", "obj": "form"}, {"node_id": "n2", "subject": "motion", "predicate": "defines", "obj": "rhythm"}], "edges": [{"source": "n1", "target": "n2", "relation": "disjunction"}]}, "accept_surfaces": ["structure defines form or motion defines rhythm"], "constraints": {"must_contain": ["structure", "or", "motion"], "word_order": ["structure", "or", "motion"], "max_words": 12}} +{"id": "gram_C04_h03", "construction": "C04", "construction_name": "disjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "reason", "predicate": "grounds", "obj": "belief"}, {"node_id": "n2", "subject": "intuition", "predicate": "grounds", "obj": "conviction"}], "edges": [{"source": "n1", "target": "n2", "relation": "disjunction"}]}, "accept_surfaces": ["reason grounds belief or intuition grounds conviction"], "constraints": {"must_contain": ["reason", "or", "intuition"], "word_order": ["reason", "or", "intuition"], "max_words": 12}} +{"id": "gram_C04_h04", "construction": "C04", "construction_name": "disjunction", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "silence", "predicate": "reveals", "obj": "depth"}, {"node_id": "n2", "subject": "speech", "predicate": "reveals", "obj": "intent"}], "edges": [{"source": "n1", "target": "n2", "relation": "disjunction"}]}, "accept_surfaces": ["silence reveals depth or speech reveals intent"], "constraints": {"must_contain": ["silence", "or", "speech"], "word_order": ["silence", "or", "speech"], "max_words": 12}} +{"id": "gram_C05_h01", "construction": "C05", "construction_name": "embedded_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "reason", "predicate": "shows", "obj": "n2"}, {"node_id": "n2", "subject": "order", "predicate": "precedes", "obj": "chaos"}], "edges": [{"source": "n1", "target": "n2", "relation": "complement"}]}, "accept_surfaces": ["reason shows that order precedes chaos"], "constraints": {"must_contain": ["reason", "shows", "that", "order", "precedes", "chaos"], "word_order": ["reason", "shows", "that", "order", "precedes", "chaos"], "max_words": 10}} +{"id": "gram_C05_h02", "construction": "C05", "construction_name": "embedded_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "experience", "predicate": "reveals", "obj": "n2"}, {"node_id": "n2", "subject": "practice", "predicate": "grounds", "obj": "mastery"}], "edges": [{"source": "n1", "target": "n2", "relation": "complement"}]}, "accept_surfaces": ["experience reveals that practice grounds mastery"], "constraints": {"must_contain": ["experience", "reveals", "that", "practice", "grounds", "mastery"], "word_order": ["experience", "reveals", "that", "practice"], "max_words": 10}} +{"id": "gram_C05_h03", "construction": "C05", "construction_name": "embedded_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "study", "predicate": "implies", "obj": "n2"}, {"node_id": "n2", "subject": "attention", "predicate": "precedes", "obj": "insight"}], "edges": [{"source": "n1", "target": "n2", "relation": "complement"}]}, "accept_surfaces": ["study implies that attention precedes insight"], "constraints": {"must_contain": ["study", "implies", "that", "attention", "precedes", "insight"], "word_order": ["study", "implies", "that", "attention"], "max_words": 10}} +{"id": "gram_C05_h04", "construction": "C05", "construction_name": "embedded_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "memory", "predicate": "shows", "obj": "n2"}, {"node_id": "n2", "subject": "repetition", "predicate": "supports", "obj": "retention"}], "edges": [{"source": "n1", "target": "n2", "relation": "complement"}]}, "accept_surfaces": ["memory shows that repetition supports retention"], "constraints": {"must_contain": ["memory", "shows", "that", "repetition", "supports", "retention"], "word_order": ["memory", "shows", "that", "repetition"], "max_words": 10}} +{"id": "gram_C06_h01", "construction": "C06", "construction_name": "relative_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "silence", "predicate": "precedes", "obj": "speech"}, {"node_id": "n2", "subject": "silence", "predicate": "reveals", "obj": "depth"}], "edges": [{"source": "n1", "target": "n2", "relation": "relative"}]}, "accept_surfaces": ["silence, which reveals depth, precedes speech"], "constraints": {"must_contain": ["silence", "which", "reveals", "depth", "precedes", "speech"], "word_order": ["silence", "which", "precedes", "speech"], "max_words": 12}} +{"id": "gram_C06_h02", "construction": "C06", "construction_name": "relative_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "discipline", "predicate": "grounds", "obj": "mastery"}, {"node_id": "n2", "subject": "discipline", "predicate": "requires", "obj": "patience"}], "edges": [{"source": "n1", "target": "n2", "relation": "relative"}]}, "accept_surfaces": ["discipline, which requires patience, grounds mastery"], "constraints": {"must_contain": ["discipline", "which", "requires", "patience", "grounds", "mastery"], "word_order": ["discipline", "which", "grounds", "mastery"], "max_words": 12}} +{"id": "gram_C06_h03", "construction": "C06", "construction_name": "relative_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "attention", "predicate": "supports", "obj": "learning"}, {"node_id": "n2", "subject": "attention", "predicate": "follows", "obj": "interest"}], "edges": [{"source": "n1", "target": "n2", "relation": "relative"}]}, "accept_surfaces": ["attention, which follows interest, supports learning"], "constraints": {"must_contain": ["attention", "which", "follows", "interest", "supports", "learning"], "word_order": ["attention", "which", "supports", "learning"], "max_words": 12}} +{"id": "gram_C06_h04", "construction": "C06", "construction_name": "relative_clause", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "form", "predicate": "implies", "obj": "function"}, {"node_id": "n2", "subject": "form", "predicate": "precedes", "obj": "content"}], "edges": [{"source": "n1", "target": "n2", "relation": "relative"}]}, "accept_surfaces": ["form, which precedes content, implies function"], "constraints": {"must_contain": ["form", "which", "precedes", "content", "implies", "function"], "word_order": ["form", "which", "implies", "function"], "max_words": 12}} +{"id": "gram_C07_h01", "construction": "C07", "construction_name": "universal_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "structure", "predicate": "implies", "obj": "constraint", "quantifier": "all"}], "edges": []}, "accept_surfaces": ["all structure implies constraint"], "constraints": {"must_contain": ["all", "structure", "implies", "constraint"], "word_order": ["all", "structure", "implies", "constraint"], "max_words": 8}} +{"id": "gram_C07_h02", "construction": "C07", "construction_name": "universal_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "motion", "predicate": "requires", "obj": "force", "quantifier": "all"}], "edges": []}, "accept_surfaces": ["all motion requires force"], "constraints": {"must_contain": ["all", "motion", "requires", "force"], "word_order": ["all", "motion", "requires", "force"], "max_words": 8}} +{"id": "gram_C07_h03", "construction": "C07", "construction_name": "universal_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "change", "predicate": "precedes", "obj": "growth", "quantifier": "all"}], "edges": []}, "accept_surfaces": ["all change precedes growth"], "constraints": {"must_contain": ["all", "change", "precedes", "growth"], "word_order": ["all", "change", "precedes", "growth"], "max_words": 8}} +{"id": "gram_C07_h04", "construction": "C07", "construction_name": "universal_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "repetition", "predicate": "supports", "obj": "habit", "quantifier": "all"}], "edges": []}, "accept_surfaces": ["all repetition supports habit"], "constraints": {"must_contain": ["all", "repetition", "supports", "habit"], "word_order": ["all", "repetition", "supports", "habit"], "max_words": 8}} +{"id": "gram_C08_h01", "construction": "C08", "construction_name": "existential_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "fear", "predicate": "reveals", "obj": "desire", "quantifier": "some"}], "edges": []}, "accept_surfaces": ["some fear reveals desire"], "constraints": {"must_contain": ["some", "fear", "reveals", "desire"], "word_order": ["some", "fear", "reveals", "desire"], "max_words": 8}} +{"id": "gram_C08_h02", "construction": "C08", "construction_name": "existential_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "error", "predicate": "precedes", "obj": "correction", "quantifier": "some"}], "edges": []}, "accept_surfaces": ["some error precedes correction"], "constraints": {"must_contain": ["some", "error", "precedes", "correction"], "word_order": ["some", "error", "precedes", "correction"], "max_words": 8}} +{"id": "gram_C08_h03", "construction": "C08", "construction_name": "existential_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "constraint", "predicate": "grounds", "obj": "freedom", "quantifier": "some"}], "edges": []}, "accept_surfaces": ["some constraint grounds freedom"], "constraints": {"must_contain": ["some", "constraint", "grounds", "freedom"], "word_order": ["some", "constraint", "grounds", "freedom"], "max_words": 8}} +{"id": "gram_C08_h04", "construction": "C08", "construction_name": "existential_quantification", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "doubt", "predicate": "supports", "obj": "inquiry", "quantifier": "some"}], "edges": []}, "accept_surfaces": ["some doubt supports inquiry"], "constraints": {"must_contain": ["some", "doubt", "supports", "inquiry"], "word_order": ["some", "doubt", "supports", "inquiry"], "max_words": 8}} +{"id": "gram_C09_h01", "construction": "C09", "construction_name": "past_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "silence", "predicate": "precedes", "obj": "speech", "tense": "past"}], "edges": []}, "accept_surfaces": ["silence preceded speech"], "constraints": {"must_contain": ["silence", "preceded", "speech"], "word_order": ["silence", "preceded", "speech"], "max_words": 6}} +{"id": "gram_C09_h02", "construction": "C09", "construction_name": "past_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "order", "predicate": "follows", "obj": "chaos", "tense": "past"}], "edges": []}, "accept_surfaces": ["order followed chaos"], "constraints": {"must_contain": ["order", "followed", "chaos"], "word_order": ["order", "followed", "chaos"], "max_words": 6}} +{"id": "gram_C09_h03", "construction": "C09", "construction_name": "past_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "practice", "predicate": "reveals", "obj": "weakness", "tense": "past"}], "edges": []}, "accept_surfaces": ["practice revealed weakness"], "constraints": {"must_contain": ["practice", "revealed", "weakness"], "word_order": ["practice", "revealed", "weakness"], "max_words": 6}} +{"id": "gram_C09_h04", "construction": "C09", "construction_name": "past_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "memory", "predicate": "grounds", "obj": "narrative", "tense": "past"}], "edges": []}, "accept_surfaces": ["memory grounded narrative"], "constraints": {"must_contain": ["memory", "grounded", "narrative"], "word_order": ["memory", "grounded", "narrative"], "max_words": 6}} +{"id": "gram_C10_h01", "construction": "C10", "construction_name": "present_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "repetition", "predicate": "defines", "obj": "rhythm"}], "edges": []}, "accept_surfaces": ["repetition defines rhythm"], "constraints": {"must_contain": ["repetition", "defines", "rhythm"], "word_order": ["repetition", "defines", "rhythm"], "max_words": 6}} +{"id": "gram_C10_h02", "construction": "C10", "construction_name": "present_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "attention", "predicate": "grounds", "obj": "perception"}], "edges": []}, "accept_surfaces": ["attention grounds perception"], "constraints": {"must_contain": ["attention", "grounds", "perception"], "word_order": ["attention", "grounds", "perception"], "max_words": 6}} +{"id": "gram_C10_h03", "construction": "C10", "construction_name": "present_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "contrast", "predicate": "reveals", "obj": "distinction"}], "edges": []}, "accept_surfaces": ["contrast reveals distinction"], "constraints": {"must_contain": ["contrast", "reveals", "distinction"], "word_order": ["contrast", "reveals", "distinction"], "max_words": 6}} +{"id": "gram_C10_h04", "construction": "C10", "construction_name": "present_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "habit", "predicate": "precedes", "obj": "character"}], "edges": []}, "accept_surfaces": ["habit precedes character"], "constraints": {"must_contain": ["habit", "precedes", "character"], "word_order": ["habit", "precedes", "character"], "max_words": 6}} +{"id": "gram_C11_h01", "construction": "C11", "construction_name": "future_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "patience", "predicate": "reveals", "obj": "reward", "tense": "future"}], "edges": []}, "accept_surfaces": ["patience will reveal reward"], "constraints": {"must_contain": ["patience", "will", "reveal", "reward"], "word_order": ["patience", "will", "reveal", "reward"], "max_words": 8}} +{"id": "gram_C11_h02", "construction": "C11", "construction_name": "future_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "effort", "predicate": "precedes", "obj": "achievement", "tense": "future"}], "edges": []}, "accept_surfaces": ["effort will precede achievement"], "constraints": {"must_contain": ["effort", "will", "precede", "achievement"], "word_order": ["effort", "will", "precede", "achievement"], "max_words": 8}} +{"id": "gram_C11_h03", "construction": "C11", "construction_name": "future_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "inquiry", "predicate": "grounds", "obj": "discovery", "tense": "future"}], "edges": []}, "accept_surfaces": ["inquiry will ground discovery"], "constraints": {"must_contain": ["inquiry", "will", "ground", "discovery"], "word_order": ["inquiry", "will", "ground", "discovery"], "max_words": 8}} +{"id": "gram_C11_h04", "construction": "C11", "construction_name": "future_tense", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "discipline", "predicate": "supports", "obj": "excellence", "tense": "future"}], "edges": []}, "accept_surfaces": ["discipline will support excellence"], "constraints": {"must_contain": ["discipline", "will", "support", "excellence"], "word_order": ["discipline", "will", "support", "excellence"], "max_words": 8}} +{"id": "gram_C12_h01", "construction": "C12", "construction_name": "perfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "doubt", "predicate": "precedes", "obj": "certainty", "aspect": "perfective"}], "edges": []}, "accept_surfaces": ["doubt has preceded certainty"], "constraints": {"must_contain": ["doubt", "has", "preceded", "certainty"], "word_order": ["doubt", "has", "preceded", "certainty"], "max_words": 8}} +{"id": "gram_C12_h02", "construction": "C12", "construction_name": "perfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "experience", "predicate": "reveals", "obj": "pattern", "aspect": "perfective"}], "edges": []}, "accept_surfaces": ["experience has revealed pattern"], "constraints": {"must_contain": ["experience", "has", "revealed", "pattern"], "word_order": ["experience", "has", "revealed", "pattern"], "max_words": 8}} +{"id": "gram_C12_h03", "construction": "C12", "construction_name": "perfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "reflection", "predicate": "grounds", "obj": "understanding", "aspect": "perfective"}], "edges": []}, "accept_surfaces": ["reflection has grounded understanding"], "constraints": {"must_contain": ["reflection", "has", "grounded", "understanding"], "word_order": ["reflection", "has", "grounded", "understanding"], "max_words": 8}} +{"id": "gram_C12_h04", "construction": "C12", "construction_name": "perfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "effort", "predicate": "supports", "obj": "growth", "aspect": "perfective"}], "edges": []}, "accept_surfaces": ["effort has supported growth"], "constraints": {"must_contain": ["effort", "has", "supported", "growth"], "word_order": ["effort", "has", "supported", "growth"], "max_words": 8}} +{"id": "gram_C13_h01", "construction": "C13", "construction_name": "imperfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "habit", "predicate": "defines", "obj": "character", "aspect": "imperfective"}], "edges": []}, "accept_surfaces": ["habit is defining character"], "constraints": {"must_contain": ["habit", "is", "defining", "character"], "word_order": ["habit", "is", "defining", "character"], "max_words": 8}} +{"id": "gram_C13_h02", "construction": "C13", "construction_name": "imperfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "curiosity", "predicate": "reveals", "obj": "possibility", "aspect": "imperfective"}], "edges": []}, "accept_surfaces": ["curiosity is revealing possibility"], "constraints": {"must_contain": ["curiosity", "is", "revealing", "possibility"], "word_order": ["curiosity", "is", "revealing", "possibility"], "max_words": 8}} +{"id": "gram_C13_h03", "construction": "C13", "construction_name": "imperfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "time", "predicate": "grounds", "obj": "change", "aspect": "imperfective"}], "edges": []}, "accept_surfaces": ["time is grounding change"], "constraints": {"must_contain": ["time", "is", "grounding", "change"], "word_order": ["time", "is", "grounding", "change"], "max_words": 8}} +{"id": "gram_C13_h04", "construction": "C13", "construction_name": "imperfective_aspect", "proposition_graph": {"nodes": [{"node_id": "n1", "subject": "practice", "predicate": "supports", "obj": "fluency", "aspect": "imperfective"}], "edges": []}, "accept_surfaces": ["practice is supporting fluency"], "constraints": {"must_contain": ["practice", "is", "supporting", "fluency"], "word_order": ["practice", "is", "supporting", "fluency"], "max_words": 8}} diff --git a/evals/zero_code_domain_acquisition/contract.md b/evals/zero_code_domain_acquisition/contract.md new file mode 100644 index 00000000..8d17eeeb --- /dev/null +++ b/evals/zero_code_domain_acquisition/contract.md @@ -0,0 +1,63 @@ +# zero-code-domain-acquisition eval lane + +## What it measures + +Whether a domain author can bring CORE to >=80% articulation accuracy on a +novel domain using *only* pack authoring (vocabulary, predicates, teaching +examples) — no Python edits allowed. Every Python edit required is logged as +an "engineering gap." + +This measures the system's extensibility contract: if CORE is well-designed, +new domains should be acquirable through data alone. + +## Surprise domains + +Three domains chosen because they were never touched during core development: + +| ID | Domain | Core proposition type | +|----|--------|----------------------| +| D1 | Kinship relations | relational (X is_parent_of Y) | +| D2 | Calendar relations | temporal ordering (Monday precedes Tuesday) | +| D3 | Color taxonomy | hierarchical/property (red is_warmer_than blue) | + +## Pack-only authoring kit (per domain) + +Each domain provides: + +1. **Vocabulary** — lexicon entries (nouns, verbs, relations) in JSONL format +2. **Relation predicates** — semantic predicates specific to the domain +3. **Axiom list** — ground-truth propositions that define the domain +4. **Teaching examples** (~20) — reviewed teaching events exercising the domain +5. **Articulation prompts** (~30) — PropositionGraph inputs to score against + +## Scoring rubric + +A prompt passes if the realized surface: +1. Is in `accept_surfaces` OR satisfies all `constraints` +2. Contains all words in `must_contain` +3. Respects `word_order` (subsequence check) +4. Does not exceed `max_words` + +## Pass thresholds + +- v1: >= 80% on public test set per domain, >= 75% on holdout +- Overall: all 3 domains must independently pass + +## Engineering gap log + +Any Python edit required to pass is logged in `evals/zero_code_domain_acquisition/gaps.md` +with: +- What broke +- Which domain triggered it +- What the fix was +- Whether it was domain-specific or general + +An empty gap log at v1 pass means the system's extensibility contract is solid. + +## Evaluation protocol + +1. Load domain pack into vocabulary manifold +2. Apply teaching examples through teaching loop +3. Run articulation prompts through the realizer pipeline +4. Score against acceptance criteria +5. Log any failures that require Python fixes as engineering gaps diff --git a/evals/zero_code_domain_acquisition/domains/README.md b/evals/zero_code_domain_acquisition/domains/README.md new file mode 100644 index 00000000..cd4de9ac --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/README.md @@ -0,0 +1,48 @@ +# Domain Authoring Kits + +Three surprise domains for testing CORE's extensibility contract. Each domain provides a complete pack-only authoring kit. + +## Kit Structure + +Each domain directory contains: + +- `vocabulary.jsonl` — Lexicon entries (nouns, verbs, predicates) in pack format +- `relations.txt` — List of domain-specific relation predicates (one per line) +- `axioms.jsonl` — Ground-truth propositions defining the domain (subject/predicate/object triples) +- `teaching_examples.jsonl` — ~20 reviewed teaching events for the teaching loop +- `prompts_dev.jsonl` — ~10 dev articulation prompts (PropositionGraph → surface) + +## Domains + +### D1: Kinship Relations +Relations between people: parent/child, sibling, spouse, ancestor/descendant, relative. + +Core predicate: `is_parent_of`, `is_child_of`, `is_sibling_of`, `is_spouse_of`, etc. + +### D2: Calendar Relations +Temporal ordering: weekdays, months, seasons, containment relationships. + +Core predicates: `precedes`, `follows`, `occurs_in`, `contains`. + +### D3: Color Taxonomy +Color properties and relationships: warmth, brightness, similarity, opposition, containment. + +Core predicates: `is_warmer_than`, `is_cooler_than`, `is_lighter_than`, `is_darker_than`, `is_opposite_of`, etc. + +## Usage Protocol + +To test domain acquisition: + +1. Load the domain's `vocabulary.jsonl` into a new language pack +2. Register the domain's relations in the relation manifest +3. Seed the vault with the `axioms.jsonl` (optional; tests "cold start" if omitted) +4. Apply `teaching_examples.jsonl` through the teaching loop +5. Run `prompts_dev.jsonl` through the realizer and score against acceptance criteria + +## Success Criteria + +Each domain must independently achieve: +- >=80% accuracy on dev prompts (no engineering required) +- >=75% accuracy on holdout prompts (no engineering required) + +An empty engineering gap log at completion means CORE's extensibility contract is solid. diff --git a/evals/zero_code_domain_acquisition/domains/calendar/axioms.jsonl b/evals/zero_code_domain_acquisition/domains/calendar/axioms.jsonl new file mode 100644 index 00000000..41de8cfd --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/calendar/axioms.jsonl @@ -0,0 +1,17 @@ +{"subject":"Monday","predicate":"precedes","object":"Tuesday"} +{"subject":"Tuesday","predicate":"precedes","object":"Wednesday"} +{"subject":"Wednesday","predicate":"precedes","object":"Thursday"} +{"subject":"Thursday","predicate":"precedes","object":"Friday"} +{"subject":"Friday","predicate":"precedes","object":"Saturday"} +{"subject":"Saturday","predicate":"precedes","object":"Sunday"} +{"subject":"January","predicate":"precedes","object":"February"} +{"subject":"February","predicate":"precedes","object":"March"} +{"subject":"March","predicate":"precedes","object":"April"} +{"subject":"April","predicate":"precedes","object":"May"} +{"subject":"May","predicate":"precedes","object":"June"} +{"subject":"spring","predicate":"precedes","object":"summer"} +{"subject":"summer","predicate":"precedes","object":"autumn"} +{"subject":"autumn","predicate":"precedes","object":"winter"} +{"subject":"Monday","predicate":"occurs_in","object":"week"} +{"subject":"January","predicate":"occurs_in","object":"year"} +{"subject":"spring","predicate":"occurs_in","object":"year"} diff --git a/evals/zero_code_domain_acquisition/domains/calendar/prompts_dev.jsonl b/evals/zero_code_domain_acquisition/domains/calendar/prompts_dev.jsonl new file mode 100644 index 00000000..197d01ad --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/calendar/prompts_dev.jsonl @@ -0,0 +1,10 @@ +{"id":"cal_p01","proposition_graph":{"nodes":[{"node_id":"n1","subject":"Monday","predicate":"precedes","obj":"Tuesday"}],"edges":[]},"accept_surfaces":["Monday precedes Tuesday"],"constraints":{"must_contain":["Monday","precedes","Tuesday"],"word_order":["Monday","precedes","Tuesday"],"max_words":8}} +{"id":"cal_p02","proposition_graph":{"nodes":[{"node_id":"n1","subject":"Wednesday","predicate":"precedes","obj":"Thursday"}],"edges":[]},"accept_surfaces":["Wednesday precedes Thursday"],"constraints":{"must_contain":["Wednesday","precedes","Thursday"],"word_order":["Wednesday","precedes","Thursday"],"max_words":8}} +{"id":"cal_p03","proposition_graph":{"nodes":[{"node_id":"n1","subject":"Friday","predicate":"precedes","obj":"Saturday"}],"edges":[]},"accept_surfaces":["Friday precedes Saturday"],"constraints":{"must_contain":["Friday","precedes","Saturday"],"word_order":["Friday","precedes","Saturday"],"max_words":8}} +{"id":"cal_p04","proposition_graph":{"nodes":[{"node_id":"n1","subject":"January","predicate":"precedes","obj":"February"}],"edges":[]},"accept_surfaces":["January precedes February"],"constraints":{"must_contain":["January","precedes","February"],"word_order":["January","precedes","February"],"max_words":8}} +{"id":"cal_p05","proposition_graph":{"nodes":[{"node_id":"n1","subject":"March","predicate":"precedes","obj":"April"}],"edges":[]},"accept_surfaces":["March precedes April"],"constraints":{"must_contain":["March","precedes","April"],"word_order":["March","precedes","April"],"max_words":8}} +{"id":"cal_p06","proposition_graph":{"nodes":[{"node_id":"n1","subject":"May","predicate":"precedes","obj":"June"}],"edges":[]},"accept_surfaces":["May precedes June"],"constraints":{"must_contain":["May","precedes","June"],"word_order":["May","precedes","June"],"max_words":8}} +{"id":"cal_p07","proposition_graph":{"nodes":[{"node_id":"n1","subject":"spring","predicate":"precedes","obj":"summer"}],"edges":[]},"accept_surfaces":["spring precedes summer"],"constraints":{"must_contain":["spring","precedes","summer"],"word_order":["spring","precedes","summer"],"max_words":8}} +{"id":"cal_p08","proposition_graph":{"nodes":[{"node_id":"n1","subject":"summer","predicate":"precedes","obj":"autumn"}],"edges":[]},"accept_surfaces":["summer precedes autumn"],"constraints":{"must_contain":["summer","precedes","autumn"],"word_order":["summer","precedes","autumn"],"max_words":8}} +{"id":"cal_p09","proposition_graph":{"nodes":[{"node_id":"n1","subject":"autumn","predicate":"precedes","obj":"winter"}],"edges":[]},"accept_surfaces":["autumn precedes winter"],"constraints":{"must_contain":["autumn","precedes","winter"],"word_order":["autumn","precedes","winter"],"max_words":8}} +{"id":"cal_p10","proposition_graph":{"nodes":[{"node_id":"n1","subject":"Monday","predicate":"occurs_in","obj":"week"}],"edges":[]},"accept_surfaces":["Monday occurs in week"],"constraints":{"must_contain":["Monday","occurs","week"],"word_order":["Monday","occurs","week"],"max_words":8}} diff --git a/evals/zero_code_domain_acquisition/domains/calendar/relations.txt b/evals/zero_code_domain_acquisition/domains/calendar/relations.txt new file mode 100644 index 00000000..b5723bfe --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/calendar/relations.txt @@ -0,0 +1,4 @@ +precedes +follows +occurs_in +contains diff --git a/evals/zero_code_domain_acquisition/domains/calendar/teaching_examples.jsonl b/evals/zero_code_domain_acquisition/domains/calendar/teaching_examples.jsonl new file mode 100644 index 00000000..3a2608b1 --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/calendar/teaching_examples.jsonl @@ -0,0 +1,20 @@ +{"id":"teach_cal_001","subject":"Monday","predicate":"precedes","object":"Tuesday","correction":"Monday precedes Tuesday in the weekly cycle","accepted":true} +{"id":"teach_cal_002","subject":"Tuesday","predicate":"precedes","object":"Wednesday","correction":"Tuesday precedes Wednesday in the weekly cycle","accepted":true} +{"id":"teach_cal_003","subject":"Wednesday","predicate":"precedes","object":"Thursday","correction":"Wednesday precedes Thursday in the weekly cycle","accepted":true} +{"id":"teach_cal_004","subject":"Thursday","predicate":"precedes","object":"Friday","correction":"Thursday precedes Friday in the weekly cycle","accepted":true} +{"id":"teach_cal_005","subject":"Friday","predicate":"precedes","object":"Saturday","correction":"Friday precedes Saturday in the weekly cycle","accepted":true} +{"id":"teach_cal_006","subject":"Saturday","predicate":"precedes","object":"Sunday","correction":"Saturday precedes Sunday in the weekly cycle","accepted":true} +{"id":"teach_cal_007","subject":"January","predicate":"precedes","object":"February","correction":"January precedes February in the annual cycle","accepted":true} +{"id":"teach_cal_008","subject":"February","predicate":"precedes","object":"March","correction":"February precedes March in the annual cycle","accepted":true} +{"id":"teach_cal_009","subject":"March","predicate":"precedes","object":"April","correction":"March precedes April in the annual cycle","accepted":true} +{"id":"teach_cal_010","subject":"April","predicate":"precedes","object":"May","correction":"April precedes May in the annual cycle","accepted":true} +{"id":"teach_cal_011","subject":"May","predicate":"precedes","object":"June","correction":"May precedes June in the annual cycle","accepted":true} +{"id":"teach_cal_012","subject":"spring","predicate":"precedes","object":"summer","correction":"Spring precedes summer in the seasonal cycle","accepted":true} +{"id":"teach_cal_013","subject":"summer","predicate":"precedes","object":"autumn","correction":"Summer precedes autumn in the seasonal cycle","accepted":true} +{"id":"teach_cal_014","subject":"autumn","predicate":"precedes","object":"winter","correction":"Autumn precedes winter in the seasonal cycle","accepted":true} +{"id":"teach_cal_015","subject":"winter","predicate":"precedes","object":"spring","correction":"Winter precedes spring in the seasonal cycle","accepted":true} +{"id":"teach_cal_016","subject":"Monday","predicate":"occurs_in","object":"week","correction":"Monday occurs in a week","accepted":true} +{"id":"teach_cal_017","subject":"January","predicate":"occurs_in","object":"year","correction":"January occurs in a year","accepted":true} +{"id":"teach_cal_018","subject":"spring","predicate":"occurs_in","object":"year","correction":"Spring occurs in a year","accepted":true} +{"id":"teach_cal_019","subject":"summer","predicate":"occurs_in","object":"year","correction":"Summer occurs in a year","accepted":true} +{"id":"teach_cal_020","subject":"week","predicate":"occurs_in","object":"year","correction":"A week occurs in a year","accepted":true} diff --git a/evals/zero_code_domain_acquisition/domains/calendar/vocabulary.jsonl b/evals/zero_code_domain_acquisition/domains/calendar/vocabulary.jsonl new file mode 100644 index 00000000..b5d6d45d --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/calendar/vocabulary.jsonl @@ -0,0 +1,26 @@ +{"entry_id":"cal-001","surface":"Monday","lemma":"Monday","language":"en","pos":"NOUN","semantic_domains":["calendar.weekday","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-002","surface":"Tuesday","lemma":"Tuesday","language":"en","pos":"NOUN","semantic_domains":["calendar.weekday","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-003","surface":"Wednesday","lemma":"Wednesday","language":"en","pos":"NOUN","semantic_domains":["calendar.weekday","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-004","surface":"Thursday","lemma":"Thursday","language":"en","pos":"NOUN","semantic_domains":["calendar.weekday","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-005","surface":"Friday","lemma":"Friday","language":"en","pos":"NOUN","semantic_domains":["calendar.weekday","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-006","surface":"Saturday","lemma":"Saturday","language":"en","pos":"NOUN","semantic_domains":["calendar.weekday","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-007","surface":"Sunday","lemma":"Sunday","language":"en","pos":"NOUN","semantic_domains":["calendar.weekday","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-008","surface":"January","lemma":"January","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-009","surface":"February","lemma":"February","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-010","surface":"March","lemma":"March","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-011","surface":"April","lemma":"April","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-012","surface":"May","lemma":"May","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-013","surface":"June","lemma":"June","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-014","surface":"July","lemma":"July","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-015","surface":"August","lemma":"August","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-016","surface":"September","lemma":"September","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-017","surface":"October","lemma":"October","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-018","surface":"November","lemma":"November","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-019","surface":"December","lemma":"December","language":"en","pos":"NOUN","semantic_domains":["calendar.month","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-020","surface":"spring","lemma":"spring","language":"en","pos":"NOUN","semantic_domains":["calendar.season","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-021","surface":"summer","lemma":"summer","language":"en","pos":"NOUN","semantic_domains":["calendar.season","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-022","surface":"autumn","lemma":"autumn","language":"en","pos":"NOUN","semantic_domains":["calendar.season","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-023","surface":"winter","lemma":"winter","language":"en","pos":"NOUN","semantic_domains":["calendar.season","time.cyclic","temporal.named"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-024","surface":"year","lemma":"year","language":"en","pos":"NOUN","semantic_domains":["calendar.period","time.duration","temporal.unit"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-025","surface":"month","lemma":"month","language":"en","pos":"NOUN","semantic_domains":["calendar.period","time.duration","temporal.unit"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} +{"entry_id":"cal-026","surface":"week","lemma":"week","language":"en","pos":"NOUN","semantic_domains":["calendar.period","time.duration","temporal.unit"],"morphology_tags":["noun"],"provenance_ids":["domain:calendar"]} diff --git a/evals/zero_code_domain_acquisition/domains/color/axioms.jsonl b/evals/zero_code_domain_acquisition/domains/color/axioms.jsonl new file mode 100644 index 00000000..c6c88011 --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/color/axioms.jsonl @@ -0,0 +1,20 @@ +{"subject":"red","predicate":"is_warmer_than","object":"blue"} +{"subject":"orange","predicate":"is_warmer_than","object":"green"} +{"subject":"yellow","predicate":"is_warmer_than","object":"purple"} +{"subject":"red","predicate":"is_warmer_than","object":"green"} +{"subject":"orange","predicate":"is_warmer_than","object":"blue"} +{"subject":"blue","predicate":"is_cooler_than","object":"red"} +{"subject":"green","predicate":"is_cooler_than","object":"orange"} +{"subject":"purple","predicate":"is_cooler_than","object":"yellow"} +{"subject":"white","predicate":"is_lighter_than","object":"gray"} +{"subject":"gray","predicate":"is_lighter_than","object":"black"} +{"subject":"pink","predicate":"is_lighter_than","object":"red"} +{"subject":"black","predicate":"is_darker_than","object":"gray"} +{"subject":"gray","predicate":"is_darker_than","object":"white"} +{"subject":"red","predicate":"is_darker_than","object":"pink"} +{"subject":"red","predicate":"is_opposite_of","object":"cyan"} +{"subject":"blue","predicate":"is_opposite_of","object":"orange"} +{"subject":"red","predicate":"is_similar_to","object":"orange"} +{"subject":"blue","predicate":"is_similar_to","object":"purple"} +{"subject":"yellow","predicate":"is_similar_to","object":"gold"} +{"subject":"red","predicate":"contains","object":"hue"} diff --git a/evals/zero_code_domain_acquisition/domains/color/prompts_dev.jsonl b/evals/zero_code_domain_acquisition/domains/color/prompts_dev.jsonl new file mode 100644 index 00000000..bbd9abe0 --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/color/prompts_dev.jsonl @@ -0,0 +1,10 @@ +{"id":"col_p01","proposition_graph":{"nodes":[{"node_id":"n1","subject":"red","predicate":"is_warmer_than","obj":"blue"}],"edges":[]},"accept_surfaces":["red is warmer than blue"],"constraints":{"must_contain":["red","warmer","than","blue"],"word_order":["red","warmer","than","blue"],"max_words":8}} +{"id":"col_p02","proposition_graph":{"nodes":[{"node_id":"n1","subject":"orange","predicate":"is_warmer_than","obj":"green"}],"edges":[]},"accept_surfaces":["orange is warmer than green"],"constraints":{"must_contain":["orange","warmer","than","green"],"word_order":["orange","warmer","than","green"],"max_words":8}} +{"id":"col_p03","proposition_graph":{"nodes":[{"node_id":"n1","subject":"yellow","predicate":"is_warmer_than","obj":"purple"}],"edges":[]},"accept_surfaces":["yellow is warmer than purple"],"constraints":{"must_contain":["yellow","warmer","than","purple"],"word_order":["yellow","warmer","than","purple"],"max_words":8}} +{"id":"col_p04","proposition_graph":{"nodes":[{"node_id":"n1","subject":"blue","predicate":"is_cooler_than","obj":"red"}],"edges":[]},"accept_surfaces":["blue is cooler than red"],"constraints":{"must_contain":["blue","cooler","than","red"],"word_order":["blue","cooler","than","red"],"max_words":8}} +{"id":"col_p05","proposition_graph":{"nodes":[{"node_id":"n1","subject":"white","predicate":"is_lighter_than","obj":"gray"}],"edges":[]},"accept_surfaces":["white is lighter than gray"],"constraints":{"must_contain":["white","lighter","than","gray"],"word_order":["white","lighter","than","gray"],"max_words":8}} +{"id":"col_p06","proposition_graph":{"nodes":[{"node_id":"n1","subject":"gray","predicate":"is_lighter_than","obj":"black"}],"edges":[]},"accept_surfaces":["gray is lighter than black"],"constraints":{"must_contain":["gray","lighter","than","black"],"word_order":["gray","lighter","than","black"],"max_words":8}} +{"id":"col_p07","proposition_graph":{"nodes":[{"node_id":"n1","subject":"pink","predicate":"is_lighter_than","obj":"red"}],"edges":[]},"accept_surfaces":["pink is lighter than red"],"constraints":{"must_contain":["pink","lighter","than","red"],"word_order":["pink","lighter","than","red"],"max_words":8}} +{"id":"col_p08","proposition_graph":{"nodes":[{"node_id":"n1","subject":"red","predicate":"is_similar_to","obj":"orange"}],"edges":[]},"accept_surfaces":["red is similar to orange"],"constraints":{"must_contain":["red","similar","to","orange"],"word_order":["red","similar","to","orange"],"max_words":8}} +{"id":"col_p09","proposition_graph":{"nodes":[{"node_id":"n1","subject":"blue","predicate":"is_similar_to","obj":"purple"}],"edges":[]},"accept_surfaces":["blue is similar to purple"],"constraints":{"must_contain":["blue","similar","to","purple"],"word_order":["blue","similar","to","purple"],"max_words":8}} +{"id":"col_p10","proposition_graph":{"nodes":[{"node_id":"n1","subject":"red","predicate":"is_opposite_of","obj":"cyan"}],"edges":[]},"accept_surfaces":["red is opposite of cyan"],"constraints":{"must_contain":["red","opposite","of","cyan"],"word_order":["red","opposite","of","cyan"],"max_words":8}} diff --git a/evals/zero_code_domain_acquisition/domains/color/relations.txt b/evals/zero_code_domain_acquisition/domains/color/relations.txt new file mode 100644 index 00000000..693c466b --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/color/relations.txt @@ -0,0 +1,7 @@ +is_warmer_than +is_cooler_than +is_lighter_than +is_darker_than +is_opposite_of +is_similar_to +contains diff --git a/evals/zero_code_domain_acquisition/domains/color/teaching_examples.jsonl b/evals/zero_code_domain_acquisition/domains/color/teaching_examples.jsonl new file mode 100644 index 00000000..84df29cb --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/color/teaching_examples.jsonl @@ -0,0 +1,20 @@ +{"id":"teach_col_001","subject":"red","predicate":"is_warmer_than","object":"blue","correction":"Red is warmer than blue on the color temperature scale","accepted":true} +{"id":"teach_col_002","subject":"orange","predicate":"is_warmer_than","object":"green","correction":"Orange is warmer than green in color perception","accepted":true} +{"id":"teach_col_003","subject":"yellow","predicate":"is_warmer_than","object":"purple","correction":"Yellow is warmer than purple in the warmth spectrum","accepted":true} +{"id":"teach_col_004","subject":"blue","predicate":"is_cooler_than","object":"red","correction":"Blue is cooler than red on the color temperature scale","accepted":true} +{"id":"teach_col_005","subject":"green","predicate":"is_cooler_than","object":"orange","correction":"Green is cooler than orange in color perception","accepted":true} +{"id":"teach_col_006","subject":"purple","predicate":"is_cooler_than","object":"yellow","correction":"Purple is cooler than yellow in the coolness spectrum","accepted":true} +{"id":"teach_col_007","subject":"white","predicate":"is_lighter_than","object":"gray","correction":"White is lighter than gray in value","accepted":true} +{"id":"teach_col_008","subject":"gray","predicate":"is_lighter_than","object":"black","correction":"Gray is lighter than black in brightness","accepted":true} +{"id":"teach_col_009","subject":"pink","predicate":"is_lighter_than","object":"red","correction":"Pink is lighter than red because it contains more white","accepted":true} +{"id":"teach_col_010","subject":"black","predicate":"is_darker_than","object":"gray","correction":"Black is darker than gray in value","accepted":true} +{"id":"teach_col_011","subject":"gray","predicate":"is_darker_than","object":"white","correction":"Gray is darker than white in brightness","accepted":true} +{"id":"teach_col_012","subject":"red","predicate":"is_darker_than","object":"pink","correction":"Red is darker than pink because pink is a lightened tint","accepted":true} +{"id":"teach_col_013","subject":"red","predicate":"is_opposite_of","object":"cyan","correction":"Red is opposite cyan on the color wheel","accepted":true} +{"id":"teach_col_014","subject":"blue","predicate":"is_opposite_of","object":"orange","correction":"Blue is opposite orange on the color wheel","accepted":true} +{"id":"teach_col_015","subject":"red","predicate":"is_similar_to","object":"orange","correction":"Red is similar to orange as adjacent warm colors","accepted":true} +{"id":"teach_col_016","subject":"blue","predicate":"is_similar_to","object":"purple","correction":"Blue is similar to purple as adjacent cool colors","accepted":true} +{"id":"teach_col_017","subject":"yellow","predicate":"is_similar_to","object":"gold","correction":"Yellow is similar to gold in the warm spectrum","accepted":true} +{"id":"teach_col_018","subject":"white","predicate":"contains","object":"light","correction":"White contains maximum light in its value","accepted":true} +{"id":"teach_col_019","subject":"red","predicate":"contains","object":"hue","correction":"Red contains hue as a fundamental color property","accepted":true} +{"id":"teach_col_020","subject":"brown","predicate":"contains","object":"red","correction":"Brown contains red as part of its composite nature","accepted":true} diff --git a/evals/zero_code_domain_acquisition/domains/color/vocabulary.jsonl b/evals/zero_code_domain_acquisition/domains/color/vocabulary.jsonl new file mode 100644 index 00000000..804cf69a --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/color/vocabulary.jsonl @@ -0,0 +1,21 @@ +{"entry_id":"col-001","surface":"red","lemma":"red","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-002","surface":"orange","lemma":"orange","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-003","surface":"yellow","lemma":"yellow","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-004","surface":"green","lemma":"green","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-005","surface":"blue","lemma":"blue","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-006","surface":"purple","lemma":"purple","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-007","surface":"white","lemma":"white","language":"en","pos":"NOUN","semantic_domains":["color.achromatic","visual.value","perception.brightness"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-008","surface":"black","lemma":"black","language":"en","pos":"NOUN","semantic_domains":["color.achromatic","visual.value","perception.brightness"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-009","surface":"gray","lemma":"gray","language":"en","pos":"NOUN","semantic_domains":["color.achromatic","visual.value","perception.brightness"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-010","surface":"brown","lemma":"brown","language":"en","pos":"NOUN","semantic_domains":["color.composite","visual.earth-tone","perception.mixture"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-011","surface":"pink","lemma":"pink","language":"en","pos":"NOUN","semantic_domains":["color.tint","visual.lightened","perception.mixture"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-012","surface":"violet","lemma":"violet","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-013","surface":"gold","lemma":"gold","language":"en","pos":"NOUN","semantic_domains":["color.metallic","visual.sheen","perception.material"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-014","surface":"silver","lemma":"silver","language":"en","pos":"NOUN","semantic_domains":["color.metallic","visual.sheen","perception.material"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-015","surface":"chromatic","lemma":"chromatic","language":"en","pos":"NOUN","semantic_domains":["color.property","visual.attribute","perception.quality"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-016","surface":"warm","lemma":"warm","language":"en","pos":"NOUN","semantic_domains":["color.temperature","visual.affect","perception.sensation"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-017","surface":"cool","lemma":"cool","language":"en","pos":"NOUN","semantic_domains":["color.temperature","visual.affect","perception.sensation"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-018","surface":"light","lemma":"light","language":"en","pos":"NOUN","semantic_domains":["color.brightness","visual.value","perception.intensity"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-019","surface":"dark","lemma":"dark","language":"en","pos":"NOUN","semantic_domains":["color.brightness","visual.value","perception.intensity"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-020","surface":"hue","lemma":"hue","language":"en","pos":"NOUN","semantic_domains":["color.property","visual.attribute","perception.classification"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} +{"entry_id":"col-021","surface":"cyan","lemma":"cyan","language":"en","pos":"NOUN","semantic_domains":["color.chromatic","visual.hue","perception.wavelength","color.complementary.red"],"morphology_tags":["noun"],"provenance_ids":["domain:color"]} diff --git a/evals/zero_code_domain_acquisition/domains/kinship/axioms.jsonl b/evals/zero_code_domain_acquisition/domains/kinship/axioms.jsonl new file mode 100644 index 00000000..8ec0ab44 --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/kinship/axioms.jsonl @@ -0,0 +1,15 @@ +{"subject":"mother","predicate":"is_parent_of","object":"son"} +{"subject":"mother","predicate":"is_parent_of","object":"daughter"} +{"subject":"father","predicate":"is_parent_of","object":"son"} +{"subject":"father","predicate":"is_parent_of","object":"daughter"} +{"subject":"son","predicate":"is_child_of","object":"mother"} +{"subject":"son","predicate":"is_child_of","object":"father"} +{"subject":"daughter","predicate":"is_child_of","object":"mother"} +{"subject":"daughter","predicate":"is_child_of","object":"father"} +{"subject":"sister","predicate":"is_sibling_of","object":"brother"} +{"subject":"brother","predicate":"is_sibling_of","object":"sister"} +{"subject":"husband","predicate":"is_spouse_of","object":"wife"} +{"subject":"wife","predicate":"is_spouse_of","object":"husband"} +{"subject":"grandmother","predicate":"is_ancestor_of","object":"grandson"} +{"subject":"grandfather","predicate":"is_ancestor_of","object":"grandson"} +{"subject":"grandfather","predicate":"is_ancestor_of","object":"granddaughter"} diff --git a/evals/zero_code_domain_acquisition/domains/kinship/prompts_dev.jsonl b/evals/zero_code_domain_acquisition/domains/kinship/prompts_dev.jsonl new file mode 100644 index 00000000..e4f67339 --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/kinship/prompts_dev.jsonl @@ -0,0 +1,10 @@ +{"id":"kin_p01","proposition_graph":{"nodes":[{"node_id":"n1","subject":"mother","predicate":"is_parent_of","obj":"son"}],"edges":[]},"accept_surfaces":["mother is parent of son"],"constraints":{"must_contain":["mother","parent","son"],"word_order":["mother","parent","son"],"max_words":8}} +{"id":"kin_p02","proposition_graph":{"nodes":[{"node_id":"n1","subject":"father","predicate":"is_parent_of","obj":"daughter"}],"edges":[]},"accept_surfaces":["father is parent of daughter"],"constraints":{"must_contain":["father","parent","daughter"],"word_order":["father","parent","daughter"],"max_words":8}} +{"id":"kin_p03","proposition_graph":{"nodes":[{"node_id":"n1","subject":"brother","predicate":"is_sibling_of","obj":"sister"}],"edges":[]},"accept_surfaces":["brother is sibling of sister"],"constraints":{"must_contain":["brother","sibling","sister"],"word_order":["brother","sibling","sister"],"max_words":8}} +{"id":"kin_p04","proposition_graph":{"nodes":[{"node_id":"n1","subject":"grandmother","predicate":"is_ancestor_of","obj":"grandchild"}],"edges":[]},"accept_surfaces":["grandmother is ancestor of grandchild"],"constraints":{"must_contain":["grandmother","ancestor","grandchild"],"word_order":["grandmother","ancestor","grandchild"],"max_words":8}} +{"id":"kin_p05","proposition_graph":{"nodes":[{"node_id":"n1","subject":"husband","predicate":"is_spouse_of","obj":"wife"}],"edges":[]},"accept_surfaces":["husband is spouse of wife"],"constraints":{"must_contain":["husband","spouse","wife"],"word_order":["husband","spouse","wife"],"max_words":8}} +{"id":"kin_p06","proposition_graph":{"nodes":[{"node_id":"n1","subject":"aunt","predicate":"is_relative_of","obj":"nephew"}],"edges":[]},"accept_surfaces":["aunt is relative of nephew"],"constraints":{"must_contain":["aunt","relative","nephew"],"word_order":["aunt","relative","nephew"],"max_words":8}} +{"id":"kin_p07","proposition_graph":{"nodes":[{"node_id":"n1","subject":"uncle","predicate":"is_relative_of","obj":"cousin"}],"edges":[]},"accept_surfaces":["uncle is relative of cousin"],"constraints":{"must_contain":["uncle","relative","cousin"],"word_order":["uncle","relative","cousin"],"max_words":8}} +{"id":"kin_p08","proposition_graph":{"nodes":[{"node_id":"n1","subject":"son","predicate":"is_child_of","obj":"mother"}],"edges":[]},"accept_surfaces":["son is child of mother"],"constraints":{"must_contain":["son","child","mother"],"word_order":["son","child","mother"],"max_words":8}} +{"id":"kin_p09","proposition_graph":{"nodes":[{"node_id":"n1","subject":"grandson","predicate":"is_descendant_of","obj":"grandfather"}],"edges":[]},"accept_surfaces":["grandson is descendant of grandfather"],"constraints":{"must_contain":["grandson","descendant","grandfather"],"word_order":["grandson","descendant","grandfather"],"max_words":8}} +{"id":"kin_p10","proposition_graph":{"nodes":[{"node_id":"n1","subject":"sister","predicate":"is_relative_of","obj":"brother"}],"edges":[]},"accept_surfaces":["sister is relative of brother"],"constraints":{"must_contain":["sister","relative","brother"],"word_order":["sister","relative","brother"],"max_words":8}} diff --git a/evals/zero_code_domain_acquisition/domains/kinship/relations.txt b/evals/zero_code_domain_acquisition/domains/kinship/relations.txt new file mode 100644 index 00000000..922088b4 --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/kinship/relations.txt @@ -0,0 +1,7 @@ +is_parent_of +is_child_of +is_sibling_of +is_spouse_of +is_ancestor_of +is_descendant_of +is_relative_of diff --git a/evals/zero_code_domain_acquisition/domains/kinship/teaching_examples.jsonl b/evals/zero_code_domain_acquisition/domains/kinship/teaching_examples.jsonl new file mode 100644 index 00000000..b127083c --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/kinship/teaching_examples.jsonl @@ -0,0 +1,20 @@ +{"id":"teach_kin_001","subject":"mother","predicate":"is_parent_of","object":"child","correction":"A mother is a parent of her child","accepted":true} +{"id":"teach_kin_002","subject":"father","predicate":"is_parent_of","object":"child","correction":"A father is a parent of his child","accepted":true} +{"id":"teach_kin_003","subject":"brother","predicate":"is_sibling_of","object":"sister","correction":"A brother is a sibling of his sister","accepted":true} +{"id":"teach_kin_004","subject":"sister","predicate":"is_sibling_of","object":"brother","correction":"A sister is a sibling of her brother","accepted":true} +{"id":"teach_kin_005","subject":"aunt","predicate":"is_relative_of","object":"cousin","correction":"An aunt is a relative of her cousin","accepted":true} +{"id":"teach_kin_006","subject":"uncle","predicate":"is_relative_of","object":"nephew","correction":"An uncle is a relative of his nephew","accepted":true} +{"id":"teach_kin_007","subject":"grandmother","predicate":"is_ancestor_of","object":"grandchild","correction":"A grandmother is an ancestor of her grandchild","accepted":true} +{"id":"teach_kin_008","subject":"grandson","predicate":"is_descendant_of","object":"grandmother","correction":"A grandson is a descendant of his grandmother","accepted":true} +{"id":"teach_kin_009","subject":"wife","predicate":"is_spouse_of","object":"husband","correction":"A wife is a spouse of her husband","accepted":true} +{"id":"teach_kin_010","subject":"husband","predicate":"is_spouse_of","object":"wife","correction":"A husband is a spouse of his wife","accepted":true} +{"id":"teach_kin_011","subject":"parent","predicate":"is_ancestor_of","object":"offspring","correction":"A parent is an ancestor of their offspring","accepted":true} +{"id":"teach_kin_012","subject":"child","predicate":"is_descendant_of","object":"parent","correction":"A child is a descendant of their parent","accepted":true} +{"id":"teach_kin_013","subject":"cousin","predicate":"is_relative_of","object":"cousin","correction":"A cousin is a relative of another cousin","accepted":true} +{"id":"teach_kin_014","subject":"nephew","predicate":"is_relative_of","object":"uncle","correction":"A nephew is a relative of his uncle","accepted":true} +{"id":"teach_kin_015","subject":"niece","predicate":"is_relative_of","object":"aunt","correction":"A niece is a relative of her aunt","accepted":true} +{"id":"teach_kin_016","subject":"mother","predicate":"is_ancestor_of","object":"granddaughter","correction":"A mother is an ancestor of her granddaughter","accepted":true} +{"id":"teach_kin_017","subject":"father","predicate":"is_ancestor_of","object":"grandson","correction":"A father is an ancestor of his grandson","accepted":true} +{"id":"teach_kin_018","subject":"sister","predicate":"is_relative_of","object":"sister","correction":"A sister is a relative of her sister","accepted":true} +{"id":"teach_kin_019","subject":"brother","predicate":"is_relative_of","object":"brother","correction":"A brother is a relative of his brother","accepted":true} +{"id":"teach_kin_020","subject":"spouse","predicate":"is_relative_of","object":"spouse","correction":"A spouse is a relative of their spouse","accepted":true} diff --git a/evals/zero_code_domain_acquisition/domains/kinship/vocabulary.jsonl b/evals/zero_code_domain_acquisition/domains/kinship/vocabulary.jsonl new file mode 100644 index 00000000..a671a6ac --- /dev/null +++ b/evals/zero_code_domain_acquisition/domains/kinship/vocabulary.jsonl @@ -0,0 +1,26 @@ +{"entry_id":"kin-001","surface":"mother","lemma":"mother","language":"en","pos":"NOUN","semantic_domains":["kinship.parent.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-002","surface":"father","lemma":"father","language":"en","pos":"NOUN","semantic_domains":["kinship.parent.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-003","surface":"son","lemma":"son","language":"en","pos":"NOUN","semantic_domains":["kinship.child.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-004","surface":"daughter","lemma":"daughter","language":"en","pos":"NOUN","semantic_domains":["kinship.child.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-005","surface":"sister","lemma":"sister","language":"en","pos":"NOUN","semantic_domains":["kinship.sibling.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-006","surface":"brother","lemma":"brother","language":"en","pos":"NOUN","semantic_domains":["kinship.sibling.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-007","surface":"grandmother","lemma":"grandmother","language":"en","pos":"NOUN","semantic_domains":["kinship.ancestor.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-008","surface":"grandfather","lemma":"grandfather","language":"en","pos":"NOUN","semantic_domains":["kinship.ancestor.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-009","surface":"aunt","lemma":"aunt","language":"en","pos":"NOUN","semantic_domains":["kinship.collateral.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-010","surface":"uncle","lemma":"uncle","language":"en","pos":"NOUN","semantic_domains":["kinship.collateral.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-011","surface":"cousin","lemma":"cousin","language":"en","pos":"NOUN","semantic_domains":["kinship.collateral.peer","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-012","surface":"spouse","lemma":"spouse","language":"en","pos":"NOUN","semantic_domains":["kinship.conjugal","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-013","surface":"husband","lemma":"husband","language":"en","pos":"NOUN","semantic_domains":["kinship.conjugal.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-014","surface":"wife","lemma":"wife","language":"en","pos":"NOUN","semantic_domains":["kinship.conjugal.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-015","surface":"grandson","lemma":"grandson","language":"en","pos":"NOUN","semantic_domains":["kinship.descendant.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-016","surface":"granddaughter","lemma":"granddaughter","language":"en","pos":"NOUN","semantic_domains":["kinship.descendant.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-017","surface":"grandchild","lemma":"grandchild","language":"en","pos":"NOUN","semantic_domains":["kinship.descendant.generic","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-018","surface":"nephew","lemma":"nephew","language":"en","pos":"NOUN","semantic_domains":["kinship.collateral.descendant.male","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-019","surface":"niece","lemma":"niece","language":"en","pos":"NOUN","semantic_domains":["kinship.collateral.descendant.female","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-020","surface":"child","lemma":"child","language":"en","pos":"NOUN","semantic_domains":["kinship.child.generic","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-021","surface":"parent","lemma":"parent","language":"en","pos":"NOUN","semantic_domains":["kinship.parent.generic","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-022","surface":"offspring","lemma":"offspring","language":"en","pos":"NOUN","semantic_domains":["kinship.descendant.generic","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-023","surface":"sibling","lemma":"sibling","language":"en","pos":"NOUN","semantic_domains":["kinship.sibling.generic","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-024","surface":"relative","lemma":"relative","language":"en","pos":"NOUN","semantic_domains":["kinship.generic","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-025","surface":"ancestor","lemma":"ancestor","language":"en","pos":"NOUN","semantic_domains":["kinship.lineal.ascendant","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} +{"entry_id":"kin-026","surface":"descendant","lemma":"descendant","language":"en","pos":"NOUN","semantic_domains":["kinship.lineal.descendant","relation.familial","role.genealogical"],"morphology_tags":["noun"],"provenance_ids":["domain:kinship"]} diff --git a/language_packs/data/grc_logos_cognition_v1/lexicon.jsonl b/language_packs/data/grc_logos_cognition_v1/lexicon.jsonl new file mode 100644 index 00000000..93a28d07 --- /dev/null +++ b/language_packs/data/grc_logos_cognition_v1/lexicon.jsonl @@ -0,0 +1,20 @@ +{"entry_id": "grc-core-cog-001", "surface": "λόγος", "lemma": "λόγος", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-001", "semantic_domains": ["logos.utterance.word", "logos.core", "communication.articulation", "meaning.symbol"], "morphology_tags": ["noun", "masculine", "nominative", "singular", "second_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-002", "surface": "ἀλήθεια", "lemma": "ἀλήθεια", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-002", "semantic_domains": ["logos.aletheia.verity", "logos.core", "epistemic.ground", "meaning.verification"], "morphology_tags": ["noun", "feminine", "nominative", "singular", "first_declension", "alpha_privative"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-003", "surface": "φῶς", "lemma": "φῶς", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-003", "semantic_domains": ["logos.illumination.photon", "logos.core", "perception.clarity", "meaning.revelation"], "morphology_tags": ["noun", "neuter", "nominative", "singular", "third_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-004", "surface": "ζωή", "lemma": "ζωή", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-004", "semantic_domains": ["logos.vitality.animate", "logos.core", "being.animate", "meaning.existence"], "morphology_tags": ["noun", "feminine", "nominative", "singular", "first_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-005", "surface": "ἀρχή", "lemma": "ἀρχή", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-005", "semantic_domains": ["logos.genesis.origin", "logos.core", "temporal.absolute", "meaning.sequence"], "morphology_tags": ["noun", "feminine", "nominative", "singular", "first_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-006", "surface": "κτίσις", "lemma": "κτίσις", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-006", "semantic_domains": ["logos.ktizo.formation", "logos.core", "formation.origin", "agency.making"], "morphology_tags": ["noun", "feminine", "nominative", "singular", "third_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-007", "surface": "γνῶσις", "lemma": "γνῶσις", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-007", "semantic_domains": ["logos.epignosis.knowledge", "logos.core", "epistemic.ground", "memory.semantic"], "morphology_tags": ["noun", "feminine", "nominative", "singular", "third_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-008", "surface": "σοφία", "lemma": "σοφία", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-008", "semantic_domains": ["logos.sophia.wisdom", "logos.core", "epistemic.judgment", "reason.order"], "morphology_tags": ["noun", "feminine", "nominative", "singular", "first_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-009", "surface": "πνεῦμα", "lemma": "πνεῦμα", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-009", "semantic_domains": ["logos.pneuma.breath", "logos.core", "being.animating", "meaning.breath"], "morphology_tags": ["noun", "neuter", "nominative", "singular", "third_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-010", "surface": "ἄνθρωπος", "lemma": "ἄνθρωπος", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-010", "semantic_domains": ["logos.anthropos.person", "agency.subject", "identity.personal", "relation.dialogic"], "morphology_tags": ["noun", "masculine", "nominative", "singular", "second_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-011", "surface": "σκότος", "lemma": "σκότος", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-011", "semantic_domains": ["logos.skotos.darkness", "logos.privation", "perception.obscurity", "meaning.concealment"], "morphology_tags": ["noun", "neuter", "nominative", "singular", "third_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-012", "surface": "ὄρθρος", "lemma": "ὄρθρος", "language": "grc", "pos": "NOUN", "morphology_id": "grc-core-morph-012", "semantic_domains": ["logos.orthros.dawn", "temporal.boundary", "perception.emergence", "meaning.revelation"], "morphology_tags": ["noun", "masculine", "nominative", "singular", "second_declension"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-013", "surface": "φαίνω", "lemma": "φαίνω", "language": "grc", "pos": "VERB", "morphology_id": "grc-core-morph-013", "semantic_domains": ["action.reveal", "agency.disclosure", "meaning.manifestation", "logos.illumination"], "morphology_tags": ["verb", "present", "active", "indicative", "first_person", "singular"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-014", "surface": "εὑρίσκω", "lemma": "εὑρίσκω", "language": "grc", "pos": "VERB", "morphology_id": "grc-core-morph-014", "semantic_domains": ["action.find", "agency.discovery", "meaning.recognition", "logos.encounter"], "morphology_tags": ["verb", "present", "active", "indicative", "first_person", "singular"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-015", "surface": "προάγω", "lemma": "προάγω", "language": "grc", "pos": "VERB", "morphology_id": "grc-core-morph-015", "semantic_domains": ["action.precede", "temporal.order", "relation.sequence", "logos.priority"], "morphology_tags": ["verb", "present", "active", "indicative", "first_person", "singular", "prefix:pro"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-016", "surface": "δεῖ", "lemma": "δεῖ", "language": "grc", "pos": "VERB", "morphology_id": "grc-core-morph-016", "semantic_domains": ["action.require", "agency.necessity", "meaning.obligation", "logos.bondedness"], "morphology_tags": ["verb", "present", "active", "indicative", "third_person", "singular", "impersonal"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-017", "surface": "πᾶς", "lemma": "πᾶς", "language": "grc", "pos": "ADJ", "morphology_id": "grc-core-morph-017", "semantic_domains": ["logos.totality", "quantification.universal", "meaning.wholeness"], "morphology_tags": ["adjective", "masculine", "nominative", "singular", "third_declension", "quantifier", "universal"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-018", "surface": "τις", "lemma": "τις", "language": "grc", "pos": "PRON", "morphology_id": "grc-core-morph-018", "semantic_domains": ["logos.existence", "quantification.existential", "meaning.particularity"], "morphology_tags": ["pronoun", "indefinite", "masculine", "nominative", "singular", "enclitic"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-019", "surface": "οὐ", "lemma": "οὐ", "language": "grc", "pos": "PART", "morphology_id": "grc-core-morph-019", "semantic_domains": ["logic.negation", "syntactic.modifier", "meaning.denial"], "morphology_tags": ["particle", "negation", "indicative_clause"], "provenance_ids": ["seed:logos_cognition_v1"]} +{"entry_id": "grc-core-cog-020", "surface": "ὅς", "lemma": "ὅς", "language": "grc", "pos": "PRON", "morphology_id": "grc-core-morph-020", "semantic_domains": ["syntax.relative_marker", "logic.predication", "meaning.relation"], "morphology_tags": ["pronoun", "relative", "masculine", "nominative", "singular"], "provenance_ids": ["seed:logos_cognition_v1"]} diff --git a/language_packs/data/grc_logos_cognition_v1/manifest.json b/language_packs/data/grc_logos_cognition_v1/manifest.json new file mode 100644 index 00000000..575dce9f --- /dev/null +++ b/language_packs/data/grc_logos_cognition_v1/manifest.json @@ -0,0 +1,13 @@ +{ + "pack_id": "grc_logos_cognition_v1", + "language": "grc", + "role": "depth_relation", + "script": "Greek", + "normalization_policy": "unitize_versor", + "source_manifest": "grc_logos_cognition_v1.lexicon.jsonl", + "determinism_class": "D0", + "checksum": "b45bcf581cee1fa2449854d20c46ed121ca3fd36fc50289282b8df0faf472351", + "version": "1.0.0", + "gate_engaged": true, + "oov_policy": "fail_closed" +} diff --git a/language_packs/data/grc_logos_cognition_v1/morphology.jsonl b/language_packs/data/grc_logos_cognition_v1/morphology.jsonl new file mode 100644 index 00000000..b0ea7404 --- /dev/null +++ b/language_packs/data/grc_logos_cognition_v1/morphology.jsonl @@ -0,0 +1,20 @@ +{"morphology_id": "grc-core-morph-001", "surface": "λόγος", "lemma": "λόγος", "language": "grc", "root": "λεγ", "prefix_chain": [], "stem": "λογ", "inflection": {"pos": "noun", "case": "nominative", "gender": "masculine", "number": "singular", "declension": "second"}, "suffix_chain": ["ος"]} +{"morphology_id": "grc-core-morph-002", "surface": "ἀλήθεια", "lemma": "ἀλήθεια", "language": "grc", "root": "ἀληθ", "prefix_chain": ["ἀ"], "stem": "ἀληθει", "inflection": {"pos": "noun", "case": "nominative", "gender": "feminine", "number": "singular", "declension": "first", "morphology_note": "alpha_privative"}, "suffix_chain": ["α"]} +{"morphology_id": "grc-core-morph-003", "surface": "φῶς", "lemma": "φῶς", "language": "grc", "root": "φωτ", "prefix_chain": [], "stem": "φωτ", "inflection": {"pos": "noun", "case": "nominative", "gender": "neuter", "number": "singular", "declension": "third"}, "suffix_chain": ["ς"]} +{"morphology_id": "grc-core-morph-004", "surface": "ζωή", "lemma": "ζωή", "language": "grc", "root": "ζω", "prefix_chain": [], "stem": "ζω", "inflection": {"pos": "noun", "case": "nominative", "gender": "feminine", "number": "singular", "declension": "first"}, "suffix_chain": ["η"]} +{"morphology_id": "grc-core-morph-005", "surface": "ἀρχή", "lemma": "ἀρχή", "language": "grc", "root": "ἀρχ", "prefix_chain": [], "stem": "ἀρχ", "inflection": {"pos": "noun", "case": "nominative", "gender": "feminine", "number": "singular", "declension": "first"}, "suffix_chain": ["η"]} +{"morphology_id": "grc-core-morph-006", "surface": "κτίσις", "lemma": "κτίσις", "language": "grc", "root": "κτι", "prefix_chain": [], "stem": "κτισ", "inflection": {"pos": "noun", "case": "nominative", "gender": "feminine", "number": "singular", "declension": "third"}, "suffix_chain": ["ις"]} +{"morphology_id": "grc-core-morph-007", "surface": "γνῶσις", "lemma": "γνῶσις", "language": "grc", "root": "γνω", "prefix_chain": [], "stem": "γνωσ", "inflection": {"pos": "noun", "case": "nominative", "gender": "feminine", "number": "singular", "declension": "third"}, "suffix_chain": ["ις"]} +{"morphology_id": "grc-core-morph-008", "surface": "σοφία", "lemma": "σοφία", "language": "grc", "root": "σοφ", "prefix_chain": [], "stem": "σοφι", "inflection": {"pos": "noun", "case": "nominative", "gender": "feminine", "number": "singular", "declension": "first"}, "suffix_chain": ["α"]} +{"morphology_id": "grc-core-morph-009", "surface": "πνεῦμα", "lemma": "πνεῦμα", "language": "grc", "root": "πνευ", "prefix_chain": [], "stem": "πνευματ", "inflection": {"pos": "noun", "case": "nominative", "gender": "neuter", "number": "singular", "declension": "third"}, "suffix_chain": []} +{"morphology_id": "grc-core-morph-010", "surface": "ἄνθρωπος", "lemma": "ἄνθρωπος", "language": "grc", "root": "ἀνθρωπ", "prefix_chain": [], "stem": "ἀνθρωπ", "inflection": {"pos": "noun", "case": "nominative", "gender": "masculine", "number": "singular", "declension": "second"}, "suffix_chain": ["ος"]} +{"morphology_id": "grc-core-morph-011", "surface": "σκότος", "lemma": "σκότος", "language": "grc", "root": "σκοτ", "prefix_chain": [], "stem": "σκοτ", "inflection": {"pos": "noun", "case": "nominative", "gender": "neuter", "number": "singular", "declension": "third"}, "suffix_chain": ["ος"]} +{"morphology_id": "grc-core-morph-012", "surface": "ὄρθρος", "lemma": "ὄρθρος", "language": "grc", "root": "ὀρθρ", "prefix_chain": [], "stem": "ὀρθρ", "inflection": {"pos": "noun", "case": "nominative", "gender": "masculine", "number": "singular", "declension": "second"}, "suffix_chain": ["ος"]} +{"morphology_id": "grc-core-morph-013", "surface": "φαίνω", "lemma": "φαίνω", "language": "grc", "root": "φαν", "prefix_chain": [], "stem": "φαιν", "inflection": {"pos": "verb", "tense": "present", "voice": "active", "mood": "indicative", "person": "first", "number": "singular"}, "suffix_chain": ["ω"]} +{"morphology_id": "grc-core-morph-014", "surface": "εὑρίσκω", "lemma": "εὑρίσκω", "language": "grc", "root": "εὑρ", "prefix_chain": [], "stem": "εὑρισκ", "inflection": {"pos": "verb", "tense": "present", "voice": "active", "mood": "indicative", "person": "first", "number": "singular", "morphology_note": "iota_infix"}, "suffix_chain": ["ω"]} +{"morphology_id": "grc-core-morph-015", "surface": "προάγω", "lemma": "προάγω", "language": "grc", "root": "ἀγ", "prefix_chain": ["προ"], "stem": "ἀγ", "inflection": {"pos": "verb", "tense": "present", "voice": "active", "mood": "indicative", "person": "first", "number": "singular"}, "suffix_chain": ["ω"]} +{"morphology_id": "grc-core-morph-016", "surface": "δεῖ", "lemma": "δεῖ", "language": "grc", "root": "δε", "prefix_chain": [], "stem": "δε", "inflection": {"pos": "verb", "tense": "present", "voice": "active", "mood": "indicative", "person": "third", "number": "singular", "subtype": "impersonal"}, "suffix_chain": ["ι"]} +{"morphology_id": "grc-core-morph-017", "surface": "πᾶς", "lemma": "πᾶς", "language": "grc", "root": "παντ", "prefix_chain": [], "stem": "παντ", "inflection": {"pos": "adjective", "case": "nominative", "gender": "masculine", "number": "singular", "declension": "third", "subtype": "universal_quantifier"}, "suffix_chain": ["ς"]} +{"morphology_id": "grc-core-morph-018", "surface": "τις", "lemma": "τις", "language": "grc", "root": "τι", "prefix_chain": [], "stem": "τι", "inflection": {"pos": "pronoun", "case": "nominative", "gender": "masculine", "number": "singular", "subtype": "indefinite", "morphology_note": "enclitic"}, "suffix_chain": ["ς"]} +{"morphology_id": "grc-core-morph-019", "surface": "οὐ", "lemma": "οὐ", "language": "grc", "root": null, "prefix_chain": [], "stem": "οὐ", "inflection": {"pos": "particle", "subtype": "negation"}, "suffix_chain": []} +{"morphology_id": "grc-core-morph-020", "surface": "ὅς", "lemma": "ὅς", "language": "grc", "root": "ὁ", "prefix_chain": [], "stem": "ὁ", "inflection": {"pos": "pronoun", "case": "nominative", "gender": "masculine", "number": "singular", "subtype": "relative"}, "suffix_chain": ["ς"]} diff --git a/language_packs/data/he_core_cognition_v1/lexicon.jsonl b/language_packs/data/he_core_cognition_v1/lexicon.jsonl new file mode 100644 index 00000000..d767774d --- /dev/null +++ b/language_packs/data/he_core_cognition_v1/lexicon.jsonl @@ -0,0 +1,20 @@ +{"entry_id": "he-core-cog-001", "surface": "דבר", "lemma": "דבר", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-001", "semantic_domains": ["logos.utterance.word", "logos.core", "communication.articulation", "meaning.symbol"], "morphology_tags": ["noun", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-002", "surface": "אמת", "lemma": "אמת", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-002", "semantic_domains": ["logos.aletheia.verity", "logos.core", "epistemic.ground", "meaning.verification"], "morphology_tags": ["noun", "feminine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-003", "surface": "אור", "lemma": "אור", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-003", "semantic_domains": ["logos.illumination.photon", "logos.core", "perception.clarity", "meaning.revelation"], "morphology_tags": ["noun", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-004", "surface": "חיים", "lemma": "חיים", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-004", "semantic_domains": ["logos.vitality.animate", "logos.core", "being.animate", "meaning.existence"], "morphology_tags": ["noun", "masculine", "plural", "pluralia_tantum"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-005", "surface": "ראשית", "lemma": "ראשית", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-005", "semantic_domains": ["logos.genesis.origin", "logos.core", "temporal.absolute", "meaning.sequence"], "morphology_tags": ["noun", "feminine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-006", "surface": "בריאה", "lemma": "בריאה", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-006", "semantic_domains": ["logos.ktizo.formation", "logos.core", "formation.origin", "agency.making"], "morphology_tags": ["noun", "feminine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-007", "surface": "דעת", "lemma": "דעת", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-007", "semantic_domains": ["logos.epignosis.knowledge", "logos.core", "epistemic.ground", "memory.semantic"], "morphology_tags": ["noun", "feminine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-008", "surface": "חכמה", "lemma": "חכמה", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-008", "semantic_domains": ["logos.sophia.wisdom", "logos.core", "epistemic.judgment", "reason.order"], "morphology_tags": ["noun", "feminine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-009", "surface": "רוח", "lemma": "רוח", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-009", "semantic_domains": ["logos.pneuma.breath", "logos.core", "being.animating", "meaning.breath"], "morphology_tags": ["noun", "feminine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-010", "surface": "אדם", "lemma": "אדם", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-010", "semantic_domains": ["logos.anthropos.person", "agency.subject", "identity.personal", "relation.dialogic"], "morphology_tags": ["noun", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-011", "surface": "חושך", "lemma": "חושך", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-011", "semantic_domains": ["logos.skotos.darkness", "logos.privation", "perception.obscurity", "meaning.concealment"], "morphology_tags": ["noun", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-012", "surface": "שחר", "lemma": "שחר", "language": "he", "pos": "NOUN", "morphology_id": "he-core-morph-012", "semantic_domains": ["logos.orthros.dawn", "temporal.boundary", "perception.emergence", "meaning.revelation"], "morphology_tags": ["noun", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-013", "surface": "גילה", "lemma": "גלה", "language": "he", "pos": "VERB", "morphology_id": "he-core-morph-013", "semantic_domains": ["action.reveal", "agency.disclosure", "meaning.manifestation", "logos.illumination"], "morphology_tags": ["verb", "qal", "perfect", "third_person", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-014", "surface": "מצא", "lemma": "מצא", "language": "he", "pos": "VERB", "morphology_id": "he-core-morph-014", "semantic_domains": ["action.find", "agency.discovery", "meaning.recognition", "logos.encounter"], "morphology_tags": ["verb", "qal", "perfect", "third_person", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-015", "surface": "קדם", "lemma": "קדם", "language": "he", "pos": "VERB", "morphology_id": "he-core-morph-015", "semantic_domains": ["action.precede", "temporal.order", "relation.sequence", "logos.priority"], "morphology_tags": ["verb", "qal", "perfect", "third_person", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-016", "surface": "צריך", "lemma": "צרך", "language": "he", "pos": "ADJ", "morphology_id": "he-core-morph-016", "semantic_domains": ["action.require", "agency.necessity", "meaning.obligation", "logos.bondedness"], "morphology_tags": ["adjective", "predicate", "masculine", "singular"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-017", "surface": "כל", "lemma": "כל", "language": "he", "pos": "DET", "morphology_id": "he-core-morph-017", "semantic_domains": ["logos.totality", "quantification.universal", "meaning.wholeness"], "morphology_tags": ["determiner", "quantifier", "universal"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-018", "surface": "יש", "lemma": "יש", "language": "he", "pos": "PART", "morphology_id": "he-core-morph-018", "semantic_domains": ["logos.existence", "quantification.existential", "meaning.particularity"], "morphology_tags": ["particle", "existential"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-019", "surface": "לא", "lemma": "לא", "language": "he", "pos": "PART", "morphology_id": "he-core-morph-019", "semantic_domains": ["logic.negation", "syntactic.modifier", "meaning.denial"], "morphology_tags": ["particle", "negation"], "provenance_ids": ["seed:core_cognition_v1"]} +{"entry_id": "he-core-cog-020", "surface": "אשר", "lemma": "אשר", "language": "he", "pos": "PART", "morphology_id": "he-core-morph-020", "semantic_domains": ["syntax.relative_marker", "logic.predication", "meaning.relation"], "morphology_tags": ["particle", "relativizer"], "provenance_ids": ["seed:core_cognition_v1"]} diff --git a/language_packs/data/he_core_cognition_v1/manifest.json b/language_packs/data/he_core_cognition_v1/manifest.json new file mode 100644 index 00000000..ab428cc3 --- /dev/null +++ b/language_packs/data/he_core_cognition_v1/manifest.json @@ -0,0 +1,13 @@ +{ + "pack_id": "he_core_cognition_v1", + "language": "he", + "role": "depth_root", + "script": "Hebrew", + "normalization_policy": "unitize_versor", + "source_manifest": "he_core_cognition_v1.lexicon.jsonl", + "determinism_class": "D0", + "checksum": "dee1e8c6ad9a76d58665333070f5e6d9acb89e13e79701ba89a8ae0aceb39616", + "version": "1.0.0", + "gate_engaged": true, + "oov_policy": "fail_closed" +} diff --git a/language_packs/data/he_core_cognition_v1/morphology.jsonl b/language_packs/data/he_core_cognition_v1/morphology.jsonl new file mode 100644 index 00000000..0ccdf51f --- /dev/null +++ b/language_packs/data/he_core_cognition_v1/morphology.jsonl @@ -0,0 +1,20 @@ +{"morphology_id": "he-core-morph-001", "surface": "דבר", "lemma": "דבר", "language": "he", "root": "ד-ב-ר", "prefix_chain": [], "stem": "דבר", "inflection": {"pos": "noun", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-002", "surface": "אמת", "lemma": "אמת", "language": "he", "root": "א-מ-נ", "prefix_chain": [], "stem": "אמת", "inflection": {"pos": "noun", "gender": "feminine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-003", "surface": "אור", "lemma": "אור", "language": "he", "root": "א-ו-ר", "prefix_chain": [], "stem": "אור", "inflection": {"pos": "noun", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-004", "surface": "חיים", "lemma": "חיים", "language": "he", "root": "ח-י-ה", "prefix_chain": [], "stem": "חי", "inflection": {"pos": "noun", "gender": "masculine", "number": "plural", "subtype": "pluralia_tantum"}, "suffix_chain": ["ים"]} +{"morphology_id": "he-core-morph-005", "surface": "ראשית", "lemma": "ראשית", "language": "he", "root": "ר-א-ש", "prefix_chain": [], "stem": "ראש", "inflection": {"pos": "noun", "gender": "feminine", "number": "singular"}, "suffix_chain": ["ית"]} +{"morphology_id": "he-core-morph-006", "surface": "בריאה", "lemma": "בריאה", "language": "he", "root": "ב-ר-א", "prefix_chain": [], "stem": "בריא", "inflection": {"pos": "noun", "gender": "feminine", "number": "singular"}, "suffix_chain": ["ה"]} +{"morphology_id": "he-core-morph-007", "surface": "דעת", "lemma": "דעת", "language": "he", "root": "י-ד-ע", "prefix_chain": [], "stem": "דע", "inflection": {"pos": "noun", "gender": "feminine", "number": "singular"}, "suffix_chain": ["ת"]} +{"morphology_id": "he-core-morph-008", "surface": "חכמה", "lemma": "חכמה", "language": "he", "root": "ח-כ-מ", "prefix_chain": [], "stem": "חכמ", "inflection": {"pos": "noun", "gender": "feminine", "number": "singular"}, "suffix_chain": ["ה"]} +{"morphology_id": "he-core-morph-009", "surface": "רוח", "lemma": "רוח", "language": "he", "root": "ר-ו-ח", "prefix_chain": [], "stem": "רוח", "inflection": {"pos": "noun", "gender": "feminine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-010", "surface": "אדם", "lemma": "אדם", "language": "he", "root": "א-ד-מ", "prefix_chain": [], "stem": "אדם", "inflection": {"pos": "noun", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-011", "surface": "חושך", "lemma": "חושך", "language": "he", "root": "ח-ש-כ", "prefix_chain": [], "stem": "חשכ", "inflection": {"pos": "noun", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-012", "surface": "שחר", "lemma": "שחר", "language": "he", "root": "ש-ח-ר", "prefix_chain": [], "stem": "שחר", "inflection": {"pos": "noun", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-013", "surface": "גילה", "lemma": "גלה", "language": "he", "root": "ג-ל-ה", "prefix_chain": [], "stem": "גלה", "inflection": {"pos": "verb", "binyan": "piel", "aspect": "perfect", "person": "third", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-014", "surface": "מצא", "lemma": "מצא", "language": "he", "root": "מ-צ-א", "prefix_chain": [], "stem": "מצא", "inflection": {"pos": "verb", "binyan": "qal", "aspect": "perfect", "person": "third", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-015", "surface": "קדם", "lemma": "קדם", "language": "he", "root": "ק-ד-מ", "prefix_chain": [], "stem": "קדם", "inflection": {"pos": "verb", "binyan": "qal", "aspect": "perfect", "person": "third", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-016", "surface": "צריך", "lemma": "צרך", "language": "he", "root": "צ-ר-כ", "prefix_chain": [], "stem": "צריכ", "inflection": {"pos": "adjective", "subtype": "predicate_participle", "gender": "masculine", "number": "singular"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-017", "surface": "כל", "lemma": "כל", "language": "he", "root": null, "prefix_chain": [], "stem": "כל", "inflection": {"pos": "determiner", "subtype": "universal_quantifier"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-018", "surface": "יש", "lemma": "יש", "language": "he", "root": null, "prefix_chain": [], "stem": "יש", "inflection": {"pos": "particle", "subtype": "existential"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-019", "surface": "לא", "lemma": "לא", "language": "he", "root": null, "prefix_chain": [], "stem": "לא", "inflection": {"pos": "particle", "subtype": "negation"}, "suffix_chain": []} +{"morphology_id": "he-core-morph-020", "surface": "אשר", "lemma": "אשר", "language": "he", "root": null, "prefix_chain": [], "stem": "אשר", "inflection": {"pos": "particle", "subtype": "relativizer"}, "suffix_chain": []}