Correct CalVer generator smoke contract

README.md

@@ -26,9 +26,9 @@ OmniDocBench is used as one document-parsing evaluation protocol. The methods
 and infrastructure in this repository are designed around the broader research
 problem of table structure recognition.
 
-## Research snapshot 2026.08.12.1
+## Research snapshot 2026.08.12.2
 
-The `2026.08.12.1` snapshot establishes the shared representation, evaluation,
+The `2026.08.12.2` snapshot establishes the shared representation, evaluation,
 and synthetic-corpus specification for two independent modeling tracks. It
 includes Canonical Table normalization, order-invariant topology targets,
 candidate-integrity checks, table-only model interfaces, a frozen 256-record
@@ -143,7 +143,7 @@ compression, blur, background, and scanning artifacts.
 Dataset roles are assigned by document, template, content, renderer, and seed
 families before rendering. Exact and near-duplicate audits operate on images,
 text, normalized structure, geometry, and provenance. See the
-[synthetic data specification](docs/corpus/SYNTHETIC_DATA_SPECIFICATION_2026.08.12.1.md)
+[synthetic data specification](docs/corpus/SYNTHETIC_DATA_SPECIFICATION_2026.08.12.2.md)
 and [data governance guide](docs/corpus/DATA_GOVERNANCE_2026.08.12.1.md).
 
 ## Installation

configs/generation_parameters_2026.08.12.2.json

@@ -0,0 +1,77 @@
+{
+  "configuration_release": "2026.08.12.2",
+  "dataset_release": "data-smoke-2026.08.12.2",
+  "schema_release": "synthetic-table-record-2026.08.12.2",
+  "seed_namespace": "borderless-table-structuring-data-smoke-2026.08.12.2",
+  "seed_start": 202608120200000,
+  "requested_records": 256,
+  "allow_failed_seed_replacement": false,
+  "gate_costs": {
+    "false_edit": 4.0,
+    "missed_edit": 1.0,
+    "uncertain_action": "KEEP"
+  },
+  "categories": {
+    "exact_keep": {"count": 64, "gate_label": "KEEP"},
+    "hard_keep": {"count": 64, "gate_label": "KEEP"},
+    "single_minimal_edit": {"count": 64, "gate_label": "EDIT"},
+    "complex_correction": {"count": 64, "gate_label": "EDIT"}
+  },
+  "roles_per_category": {
+    "train": 44,
+    "development": 12,
+    "holdout": 8
+  },
+  "counterfactual_pairs": {
+    "minimum_pairs": 64,
+    "same_role_required": true,
+    "shared_fields": [
+      "content_family_id",
+      "template_family_id",
+      "font_family_id",
+      "renderer_family_id",
+      "base_degradation"
+    ]
+  },
+  "structure": {
+    "rows": {"minimum": 2, "maximum": 28},
+    "columns": {"minimum": 2, "maximum": 14},
+    "header_depth": {"minimum": 0, "maximum": 4},
+    "span_fraction": {"minimum": 0.0, "maximum": 0.45},
+    "empty_cell_fraction": {"minimum": 0.0, "maximum": 0.35},
+    "maximum_cells": 240
+  },
+  "rendering": {
+    "canvas_width_px": [960, 1280, 1600, 2048],
+    "dpi": [96, 150, 200, 300],
+    "font_size_px": {"minimum": 12, "maximum": 32},
+    "cell_padding_px": {"minimum": 4, "maximum": 18},
+    "rotation_degrees": {"minimum": -4.0, "maximum": 4.0},
+    "border_alpha": {"minimum": 0.0, "maximum": 1.0},
+    "jpeg_quality": {"minimum": 55, "maximum": 100},
+    "gaussian_blur_radius": {"minimum": 0.0, "maximum": 1.8},
+    "noise_sigma": {"minimum": 0.0, "maximum": 12.0}
+  },
+  "corruptions": {
+    "single_minimal_edit_exactly_one": true,
+    "complex_correction_minimum_operations": 2,
+    "complex_correction_maximum_operations": 4,
+    "accepted_prior_must_remain_canonical_legal": true,
+    "invalid_grid_fixtures_are_test_only": true,
+    "allowed": [
+      "extra_split",
+      "missing_split",
+      "incorrect_merge",
+      "missing_merge",
+      "span_extent_error",
+      "token_ownership_error",
+      "geometry_inconsistency",
+      "row_or_column_assignment_error"
+    ]
+  },
+  "determinism": {
+    "semantic_replay_runs": 2,
+    "normalized_pixel_replay_runs": 2,
+    "sha256_algorithm": "sha256"
+  }
+}

dataset/README.md

@@ -6,13 +6,13 @@ dataset references used by reproducible experiments.
 ## Active dataset entry template
 
 ```yaml
-dataset_name: data-2026.08.12.1
+dataset_name: data-2026.08.12.2
 service: private-hugging-face-or-approved-object-store
 repository: DearKarl/borderless-table-structuring-data
 revision: null
 root_manifest_sha256: null
-schema_release: synthetic-table-record-2026.08.12.1
-research_snapshot: 2026.08.12.1
+schema_release: synthetic-table-record-2026.08.12.2
+research_snapshot: 2026.08.12.2
 roles_available: []
 license_manifest_sha256: null
 overlap_audit_sha256: null

docs/corpus/ACCEPTANCE_CRITERIA_2026.08.12.2.md

@@ -0,0 +1,65 @@
+# Acceptance Criteria 2026.08.12.2
+
+Status: `FROZEN_FOR_GENERATOR_SMOKE`
+
+## Preregistered smoke gate
+
+`data-smoke-2026.08.12.2` passes only when all criteria below pass. No failed
+seed may be silently replaced, no category quota may be rebalanced after
+generation, and no model performance metric may be used to select records.
+
+| Criterion | Required result | Failure action |
+|---|---:|---|
+| Requested records | 256 | Stop |
+| Generated records | 256 | Stop |
+| Accepted records | 256 | Stop |
+| Exact KEEP / Hard KEEP / Single / Complex | 64 / 64 / 64 / 64 | Stop |
+| Counterfactual pairs | at least 64 complete same-role pairs | Stop |
+| Schema validation | 256/256 | Stop |
+| Canonical legality of Gold and prior | 256/256 | Stop |
+| Complete grid coverage of Gold and prior | 256/256 | Stop |
+| Gold recompilation | 256/256 semantic identity | Stop |
+| Unique OCR-token ownership | 100% of owned tokens | Stop |
+| Finite complete geometry | 256/256 | Stop |
+| KEEP semantic identity | 128/128 | Stop |
+| EDIT replay to Gold | 128/128 | Stop |
+| Explicit KEEP empty diff | 128/128 | Stop |
+| LoRA target completeness | 256/256 | Stop |
+| Semantic replay | identical across 2 runs | Stop |
+| Normalized-pixel replay | identical across 2 runs | Stop |
+| Source and license inventory | complete and approved | Stop |
+| Cross-role exact overlap | 0 | Stop |
+| Cross-role unresolved near overlap | 0 | Stop |
+| Terminal-input usage | false for every record | Stop |
+| Incomplete-grid rejection fixture | rejected as `GRID_INCOMPLETE` | Stop |
+| Complete count accounting | requested/generated/accepted/quarantined/failed | Stop |
+
+## Validation reason codes
+
+Every failure uses a stable reason code, including:
+
+- `SCHEMA_INVALID`;
+- `CANONICAL_ILLEGAL`;
+- `GRID_INCOMPLETE`;
+- `GOLD_RECOMPILE_MISMATCH`;
+- `TOKEN_OWNERSHIP_INVALID`;
+- `GEOMETRY_INVALID`;
+- `KEEP_NOT_IDENTITY`;
+- `EDIT_NOT_REPLAYABLE`;
+- `NONDETERMINISTIC_SEMANTICS`;
+- `NONDETERMINISTIC_PIXELS`;
+- `SOURCE_OR_LICENSE_INCOMPLETE`;
+- `CROSS_ROLE_EXACT_OVERLAP`;
+- `CROSS_ROLE_NEAR_OVERLAP`;
+- `TERMINAL_DERIVATION_RISK`.
+
+Quarantine manifests preserve every failed record identifier, frozen seed,
+reason code, and diagnostic hash. The intentionally incomplete-grid fixture is
+test evidence and is not part of the 256-record corpus manifest.
+
+## Post-pass boundary
+
+A sealed pass authorizes a new non-overwriting shared-corpus build with frozen
+distributions. It does not authorize Explicit or LoRA training. Corpus scale is
+chosen from measured generation cost, coverage, and quarantine rate, not from
+current official benchmark scores.

docs/corpus/COVERAGE_MATRIX_2026.08.12.2.csv

@@ -0,0 +1,30 @@
+category,gate_label,phenomenon,requested_count,train_count,development_count,holdout_count,counterfactual_pair_required,primary_risk
+exact_keep,KEEP,simple_regular,12,8,2,2,false,false_edit
+exact_keep,KEEP,hierarchical_header,12,8,2,2,false,false_edit
+exact_keep,KEEP,row_or_column_span,12,8,2,2,false,false_edit
+exact_keep,KEEP,empty_or_sparse_cells,8,6,2,0,false,false_edit
+exact_keep,KEEP,dense_or_multiline_text,8,6,2,0,false,text_loss
+exact_keep,KEEP,mixed_script_or_formula,6,4,2,0,false,text_loss
+exact_keep,KEEP,clean_high_resolution_control,6,4,0,2,false,false_edit
+hard_keep,KEEP,multilevel_irregular_header,12,8,2,2,true,false_edit
+hard_keep,KEEP,mixed_two_dimensional_spans,12,8,2,2,true,false_edit
+hard_keep,KEEP,weak_missing_or_partial_borders,12,8,2,2,true,false_edit
+hard_keep,KEEP,empty_near_empty_or_sparse,8,6,2,0,true,false_edit
+hard_keep,KEEP,dense_small_long_or_multiline_text,8,6,2,0,true,text_loss
+hard_keep,KEEP,rotation_blur_compression_or_noise,8,4,2,2,true,false_edit
+hard_keep,KEEP,color_background_or_alignment_trap,4,4,0,0,true,false_edit
+single_minimal_edit,EDIT,extra_split,12,8,2,2,true,cell_inflation
+single_minimal_edit,EDIT,missing_split,10,6,2,2,true,under_segmentation
+single_minimal_edit,EDIT,incorrect_merge,10,6,2,2,true,content_reassignment
+single_minimal_edit,EDIT,missing_merge,10,6,2,2,true,over_segmentation
+single_minimal_edit,EDIT,span_extent_error,8,6,2,0,true,topology_error
+single_minimal_edit,EDIT,token_ownership_error,6,4,2,0,true,text_loss
+single_minimal_edit,EDIT,geometry_inconsistency,4,4,0,0,true,location_error
+single_minimal_edit,EDIT,row_or_column_assignment_error,4,4,0,0,true,topology_error
+complex_correction,EDIT,multilevel_header_with_spans,12,8,2,2,false,topology_error
+complex_correction,EDIT,joint_split_and_merge,12,8,2,2,false,cell_inflation
+complex_correction,EDIT,weak_border_with_topology_error,10,6,2,2,false,missed_edit
+complex_correction,EDIT,empty_cells_with_spans,8,6,2,0,false,ownership_error
+complex_correction,EDIT,dense_multiline_with_topology_error,8,6,2,0,false,text_loss
+complex_correction,EDIT,mixed_script_formula_with_topology_error,6,4,2,0,false,text_loss
+complex_correction,EDIT,rotation_noise_with_topology_error,8,6,0,2,false,geometry_error

docs/corpus/SYNTHETIC_DATA_SPECIFICATION_2026.08.12.2.md

@@ -0,0 +1,68 @@
+# Synthetic Data Specification 2026.08.12.2
+
+Status: `FROZEN_FOR_GENERATOR_SMOKE`
+
+Dataset identifier: `data-smoke-2026.08.12.2`
+
+Schema identifier: `synthetic-table-record-2026.08.12.2`
+
+Supersedes: `2026.08.12.1`
+
+## Purpose of this revision
+
+This revision resolves one contradiction discovered during implementation
+review. The earlier coverage matrix requested four accepted records whose prior
+contained incomplete grid coverage, while the acceptance gate correctly
+required complete and legal prior geometry for all 256 accepted records.
+
+Incomplete-grid examples remain useful validator rejection fixtures, but they
+must be `QUARANTINE` records outside the accepted 256-record smoke. The four
+accepted records are now legal row-or-column assignment errors that preserve
+complete grid coverage while requiring a deterministic topology correction.
+
+## Inherited research contract
+
+All unchanged requirements of
+[`SYNTHETIC_DATA_SPECIFICATION_2026.08.12.1.md`](SYNTHETIC_DATA_SPECIFICATION_2026.08.12.1.md)
+remain binding, including:
+
+- `KEEP`, `EDIT`, and `QUARANTINE` gate semantics;
+- direct Canonical Table and order-invariant supervision;
+- Explicit topology-only targets with prior text frozen;
+- complete LoRA Canonical Table targets;
+- a four-to-one false-edit versus missed-edit cost;
+- uncertainty resolving to KEEP;
+- 64 Exact KEEP, 64 Hard KEEP, 64 Single Minimal Edit, and 64 Complex
+  Correction records;
+- 176 train, 48 development, and 32 holdout records;
+- at least 64 same-role counterfactual pairs;
+- no failed-seed replacement;
+- source, license, provenance, deterministic replay, and overlap gates;
+- no model training or performance claim.
+
+## Active frozen artifacts
+
+- Coverage matrix:
+  [`COVERAGE_MATRIX_2026.08.12.2.csv`](COVERAGE_MATRIX_2026.08.12.2.csv)
+- Generation parameters:
+  [`generation_parameters_2026.08.12.2.json`](../../configs/generation_parameters_2026.08.12.2.json)
+- Record schema:
+  [`synthetic_table_record_2026.08.12.2.json`](../../schemas/synthetic_table_record_2026.08.12.2.json)
+- Acceptance criteria:
+  [`ACCEPTANCE_CRITERIA_2026.08.12.2.md`](ACCEPTANCE_CRITERIA_2026.08.12.2.md)
+- License policy:
+  [`LICENSE_AND_SOURCE_POLICY_2026.08.12.1.md`](LICENSE_AND_SOURCE_POLICY_2026.08.12.1.md)
+- Split policy:
+  [`SPLIT_AND_ISOLATION_POLICY_2026.08.12.1.md`](SPLIT_AND_ISOLATION_POLICY_2026.08.12.1.md)
+
+## Negative rejection fixtures
+
+The generator test suite must additionally create at least one intentionally
+invalid incomplete-grid fixture. It must be rejected with
+`GRID_INCOMPLETE`, remain outside the 256 accepted records, and appear in test
+evidence rather than the released corpus manifest.
+
+## Escalation boundary
+
+A sealed pass of `data-smoke-2026.08.12.2` authorizes a new non-overwriting
+shared-corpus build. It does not authorize Explicit or LoRA model training.

docs/experiment-records/EVIDENCE_CARD_SYNTHETIC_DATA_SPECIFICATION_2026.08.12.2.md

@@ -0,0 +1,55 @@
+# Evidence Card: Synthetic Data Specification 2026.08.12.2
+
+Status: `PASS_SEALED_FOR_BOUNDED_GENERATOR_IMPLEMENTATION`
+
+## Objective
+
+Resolve the implementation-review contradiction between an accepted
+incomplete-grid prior and the mandatory complete-grid acceptance rule, while
+preserving the frozen 256-record composition and all source, license, split,
+determinism, and non-training boundaries.
+
+## Change
+
+- Replaced four accepted `localized_grid_coverage_error` records with four
+  accepted `row_or_column_assignment_error` records.
+- Required both Gold and prior to be legal Canonical Tables with complete grid
+  coverage for all 256 records.
+- Moved incomplete-grid corruption to an explicit negative rejection fixture
+  outside the accepted corpus manifest.
+- Advanced the specification, configuration, schema, dataset smoke, and public
+  research snapshot to `2026.08.12.2`.
+
+## Unchanged boundary
+
+- Requested/generated/accepted records: `256/256/256`.
+- Category counts: `64/64/64/64`.
+- Role counts: `176/48/32` train/development/holdout.
+- Current OmniDocBench or Customer50 contents used: `false`.
+- Model training started: `false`.
+- Current official benchmark evaluation run: `false`.
+- Historical tags and sealed artifacts modified: `false`.
+
+## Validation
+
+- All JSON and JSON Schema documents parsed successfully.
+- Coverage and role totals closed exactly.
+- All local Markdown links resolved.
+- The data-free regression suite passed `18/18` tests under Python 3.12.
+- Repository diff passed `git diff --check`.
+
+## Frozen artifact hashes
+
+| Artifact | SHA256 |
+|---|---|
+| `configs/generation_parameters_2026.08.12.2.json` | `4d5730db1e814256bd0c3287e3a8f36bc6f968038da01fcb0ef7300ebd29b625` |
+| `docs/corpus/SYNTHETIC_DATA_SPECIFICATION_2026.08.12.2.md` | `0c66f64d6e9be7fbb8e554e57500b176b7486a93445ad6f7d259b01668818cb7` |
+| `docs/corpus/COVERAGE_MATRIX_2026.08.12.2.csv` | `dd35f920d5614e609213dc9bce22bf0dd0549d2c8b78a64b7788aa22fd9ed8d2` |
+| `docs/corpus/ACCEPTANCE_CRITERIA_2026.08.12.2.md` | `c232d2a638005ef13ecb72c1fbe697c7b31e37be64413981398e3a56ecc7c147` |
+| `schemas/synthetic_table_record_2026.08.12.2.json` | `ccab367184358f75cf653235034eb33de0688618b2e9923b0c58871fae059f0e` |
+
+## Result
+
+The `2026.08.12.2` specification revision passes and is the only active input
+to the bounded generator smoke. The immutable `2026.08.12.1` tag remains
+preserved as historical research evidence.

docs/experiment-records/EVIDENCE_CARD_SYNTHETIC_DATA_SPECIFICATION_2026.08.12.2.sha256

@@ -0,0 +1 @@
+f095803788923f32bf633e9cc86fadaf7564d2fbdf6a68692fce607b257ff377  EVIDENCE_CARD_SYNTHETIC_DATA_SPECIFICATION_2026.08.12.2.md

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "borderless-table-structuring-lab"
-version = "2026.8.12.1"
+version = "2026.8.12.2"
 description = "Research on canonical representations, explicit topology modeling, and generative adaptation for borderless table structure recognition."
 readme = "README.md"
 requires-python = ">=3.10"