| { |
| "completed": 76, |
| "planned": 76, |
| "totals": { |
| "cells": 8667, |
| "valid_cells": 8600, |
| "exactly_repaired_cells": 48, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 59, |
| "flagged_valid_cells": 1, |
| "faulty_cells": 67, |
| "incorrect_replacements": 1, |
| "hard_negative_cases": 18, |
| "hard_negative_cases_harmed": 0 |
| }, |
| "categories": { |
| "text.encoding": { |
| "gold_cells": 9, |
| "detected_cells": 9, |
| "exactly_repaired_cells": 7 |
| }, |
| "text.invisible": { |
| "gold_cells": 2, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 2 |
| }, |
| "text.spelling": { |
| "gold_cells": 9, |
| "detected_cells": 9, |
| "exactly_repaired_cells": 8 |
| }, |
| "format.number": { |
| "gold_cells": 2, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 2 |
| }, |
| "format.date": { |
| "gold_cells": 2, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 2 |
| }, |
| "format.boolean": { |
| "gold_cells": 2, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 2 |
| }, |
| "format.case": { |
| "gold_cells": 2, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 2 |
| }, |
| "format.whitespace": { |
| "gold_cells": 3, |
| "detected_cells": 3, |
| "exactly_repaired_cells": 3 |
| }, |
| "format.unit": { |
| "gold_cells": 3, |
| "detected_cells": 3, |
| "exactly_repaired_cells": 3 |
| }, |
| "schema.type": { |
| "gold_cells": 3, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 2 |
| }, |
| "schema.identifier": { |
| "gold_cells": 5, |
| "detected_cells": 5, |
| "exactly_repaired_cells": 5 |
| }, |
| "schema.category": { |
| "gold_cells": 11, |
| "detected_cells": 11, |
| "exactly_repaired_cells": 11 |
| }, |
| "consistency.unit": { |
| "gold_cells": 2, |
| "detected_cells": 0, |
| "exactly_repaired_cells": 0 |
| }, |
| "consistency.dependency": { |
| "gold_cells": 3, |
| "detected_cells": 0, |
| "exactly_repaired_cells": 0 |
| }, |
| "consistency.temporal": { |
| "gold_cells": 2, |
| "detected_cells": 0, |
| "exactly_repaired_cells": 0 |
| }, |
| "missing.required": { |
| "gold_cells": 2, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 0 |
| }, |
| "missing.conditional": { |
| "gold_cells": 2, |
| "detected_cells": 2, |
| "exactly_repaired_cells": 0 |
| }, |
| "structure.shift": { |
| "gold_cells": 4, |
| "detected_cells": 4, |
| "exactly_repaired_cells": 0 |
| } |
| }, |
| "languages": { |
| "ja": { |
| "cells": 774, |
| "valid_cells": 766, |
| "exactly_repaired_cells": 4, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 6, |
| "flagged_valid_cells": 0, |
| "faulty_cells": 8 |
| }, |
| "de": { |
| "cells": 1372, |
| "valid_cells": 1363, |
| "exactly_repaired_cells": 7, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 9, |
| "flagged_valid_cells": 0, |
| "faulty_cells": 9 |
| }, |
| "fr": { |
| "cells": 1670, |
| "valid_cells": 1656, |
| "exactly_repaired_cells": 12, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 13, |
| "flagged_valid_cells": 0, |
| "faulty_cells": 14 |
| }, |
| "nl": { |
| "cells": 1124, |
| "valid_cells": 1116, |
| "exactly_repaired_cells": 6, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 7, |
| "flagged_valid_cells": 1, |
| "faulty_cells": 8 |
| }, |
| "pl": { |
| "cells": 1655, |
| "valid_cells": 1644, |
| "exactly_repaired_cells": 9, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 9, |
| "flagged_valid_cells": 0, |
| "faulty_cells": 11 |
| }, |
| "en": { |
| "cells": 1342, |
| "valid_cells": 1332, |
| "exactly_repaired_cells": 6, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 9, |
| "flagged_valid_cells": 0, |
| "faulty_cells": 10 |
| }, |
| "el": { |
| "cells": 730, |
| "valid_cells": 723, |
| "exactly_repaired_cells": 4, |
| "valid_cells_harmed": 0, |
| "detected_faulty_cells": 6, |
| "flagged_valid_cells": 0, |
| "faulty_cells": 7 |
| } |
| }, |
| "decisions": { |
| "applied": 49, |
| "relation contradiction: ambiguous faulty member; unchanged": 6, |
| "abstained: output violates a declared rule": 8, |
| "preserved": 1, |
| "abstained: perplexity exceeds validation threshold": 2 |
| }, |
| "model_revision": "913f3efbf4f0557b1ad5ec1693730d8b7535ac58", |
| "dataset_revision": "602b8a454ca448713d8a5658650d4feaad8d579c", |
| "scope": "Representative held-out synthetic test panel, not the entire test split. All context cells scanned. Gold used only after inference. Thresholds frozen from validation. Relation groups are not counted as localized cell predictions. Compound category counts may overlap.", |
| "historical_same_55_tables": { |
| "v1": { |
| "exactly_repaired_cells": 23, |
| "faulty_cells": 45, |
| "valid_cells_harmed": 55, |
| "valid_cells": 6193 |
| }, |
| "v2": { |
| "exactly_repaired_cells": 30, |
| "faulty_cells": 45, |
| "valid_cells_harmed": 0, |
| "valid_cells": 6193 |
| }, |
| "interpretation": "Whole pipeline comparison, not isolated backbone improvement; frozen old result artifacts reused without rerunning training." |
| } |
| } |
|
|
