Upload README.md with huggingface_hub

README.md

@@ -13,18 +13,16 @@ size_categories:
   - n<1K
 ---
 
-# OCR Benchmark — Document Corpus
+# OCR Benchmark — Documents
 
-93 real-world document images with multi-format ground truth, used by the
+The 93 document images and ground truth used by the
 [ocr-benchmark](https://github.com/ilsilfverskiold/ocr-benchmark) harness.
-This dataset stores the documents and a reference run — the benchmark code
-lives in the GitHub repo.
+The benchmark code, the reference run results, and the full methodology live
+in the GitHub repo — this dataset is the document corpus only.
 
-## Dataset structure
+## Structure
 
-### Train split (93 rows)
-
-Each row is one document:
+One `train` split, 93 rows, one row per document:
 
 | Column | Type | Description |
 |---|---|---|
@@ -38,20 +36,9 @@ Each row is one document:
 | `markdown_gt` | string | Ground-truth markdown (null if unavailable) |
 | `geometry_gt` | string | Bounding-box ground truth as serialized JSON (null if unavailable) |
 
-### Supplementary files
-
-These are outputs from the reference benchmark run (14 engines × 93 documents)
-and are used by the benchmark harness for report generation and judge caching.
-
-- `results/runs/` — JSONL result records
-- `results/ocr_outputs/` — Raw OCR output per (engine, document, pass)
-- `results/schema_judge_cache/` — Cached LLM judge verdicts (schema/key_value legs)
-- `results/table_judge_cache/` — Cached LLM judge verdicts (tables leg)
-- `results/markdown_judge_cache/` — Cached LLM judge verdicts (markdown leg)
-- `results/judge_checklists/` — Document judge checklists (silver-GT docs)
-- `documents/manifest.json` — Corpus manifest with source dataset provenance
-- `documents/gt_limitations.json` — Known ground-truth limitations
-- `documents/gt_provenance.json` — Per-document GT provenance audit trail
+JSON columns are stored as strings because their internal structure varies per
+document type (a tax form's schema is not a receipt's schema). Parse with
+`json.loads()`.
 
 ## Document types
 
@@ -78,40 +65,50 @@ deterministically (fixed row indices), not cherry-picked by content.
 | [RVL-CDIP](https://huggingface.co/datasets/nielsr/rvl_cdip_10_examples_per_class) (Harley et al.) | 18 | Research use | Silver — Tesseract at build time; judge-only evaluation |
 
 FUNSD, IAM, and RVL-CDIP (30 of 93 docs) carry research-use or non-commercial
-restrictions. If you are a rights holder and want a document removed, open an
-issue — it will be removed promptly.
+restrictions. This dataset redistributes small excerpts solely for reproducible
+research benchmarking, with full attribution. If you are a rights holder and
+want a document removed, open an issue — it will be removed promptly.
 
 ## Ground-truth provenance
 
 - **Human (75 docs)** — source dataset annotations
 - **Model-verified (8 docs)** — model-transcribed, cross-checked against 14-engine consensus
-- **Silver (11 docs)** — machine-generated (Tesseract); evaluated by LLM judge only
-- No GT anywhere is LLM-generated without verification
+- **Silver (11 docs)** — machine-generated (Tesseract); scored by LLM judge only, never by text metrics
 
-Known limitations are recorded in `documents/gt_limitations.json`.
+Per-document provenance, known limitations, and the full audit trail are in the
+[GitHub repo](https://github.com/ilsilfverskiold/ocr-benchmark) under
+`documents/` — and every benchmark run prints them before any score.
 
 ## Usage
 
-This dataset is meant to be used with the benchmark harness:
+Load directly:
+
+```python
+from datasets import load_dataset
+import json
+
+ds = load_dataset("ilsilfverskiold/ocr-benchmark", split="train")
+row = ds[0]
+row["image"]                          # PIL image
+gt = json.loads(row["structured_gt"]) # json_schema + true_json
+```
+
+Or use it through the benchmark harness, which rebuilds the on-disk corpus
+layout and can rerun any engine against these documents:
 
 ```bash
 git clone https://github.com/ilsilfverskiold/ocr-benchmark
 cd ocr-benchmark
 pip install -r requirements.txt
-python scripts/download_data.py
-python run_benchmark.py --report 20260810T130609Z
+python scripts/download_data.py     # fetches this dataset
+python run_benchmark.py --engines tesseract,docling --tiers easy --skip-judge
 ```
 
-Or download supplementary files directly:
-
-```python
-from huggingface_hub import snapshot_download
-snapshot_download("ilsilfverskiold/ocr-benchmark", repo_type="dataset",
-                  allow_patterns="results/**", local_dir=".")
-```
+The reference benchmark results (14 engines × 93 documents × 7 legs) ship with
+the GitHub repo — you only need this dataset if you want to rerun engines.
 
 ## License
 
-Documents are excerpts from third-party datasets — see the source table above
-for per-dataset licensing. The permissive subset (MIT / CC-BY-4.0) covers 57
-of the 93 documents.
+The images are excerpts from third-party datasets and keep their original
+licenses — see the source table above. The permissive subset (MIT / CC-BY-4.0)
+covers 57 of the 93 documents.