Verified evaluation set: 719 questions, metadata and normalised ground-truth geometry

README.md

@@ -0,0 +1,143 @@
+---
+license: mit
+language:
+  - en
+  - zh
+task_categories:
+  - visual-question-answering
+  - document-question-answering
+tags:
+  - evidence-attribution
+  - visual-document-understanding
+  - attribution-hallucination
+  - document-grounding
+size_categories:
+  - n<1K
+configs:
+  - config_name: default
+    data_files: verified_eval_set.jsonl
+---
+
+# Verified evaluation set for evidence attribution in visual documents
+
+The 719-question evaluation set used in [**"Evidence Attribution in Visual Document
+Understanding without Coordinates or Region Labels"**](https://arxiv.org/abs/2607.24651)
+(Liu, Zhang, Xiao, 2026).
+
+Code: [github.com/Ryenhails/quote-and-retrieve](https://github.com/Ryenhails/quote-and-retrieve) ·
+Model: [Ryenhails/quote-and-retrieve-8b-grpo](https://huggingface.co/Ryenhails/quote-and-retrieve-8b-grpo)
+
+## What this is
+
+CiteVQA links to source PDFs that are no longer all reachable, and some of the reachable ones
+differ from the version that was annotated. Evaluating evidence attribution on those questions
+measures broken links rather than models. This release is the subset whose annotations we could
+verify against the PDF actually available, together with the normalised ground-truth geometry
+needed to score against it.
+
+Starting from the 987 single-document questions in the CiteVQA validation release, the filter
+removes 117 questions whose source PDF does not resolve or is not a valid PDF, 4 whose annotated
+evidence pages lie outside the downloaded file, 131 whose annotated evidence text does not match
+the extracted page text, and 16 that cannot be byte-verified, retaining **719 questions over 440
+documents (72.9%)**. The filter inspects only ground-truth annotations and document content, never
+model outputs, so it cannot favour any system. Documents dropped for text mismatch have text
+layers at least as rich as the retained ones, so the filter does not select for easy cases.
+
+| | |
+|---|---|
+| questions | 719 (386 English, 333 Chinese) |
+| documents | 440 PDFs, median 34 pages, longest 182 |
+| necessary evidence elements | 1,034 |
+| all annotated elements | 1,763 |
+| question types | Complex Synthesis 394, Multimodal Parsing 136, Factual Retrieval 122, Quantitative Reasoning 67 |
+
+## What this contains, and what it does not
+
+This release contains **identifiers, our verification metadata, and our normalised ground-truth
+boxes**. It does **not** redistribute CiteVQA's question text, answers, or PDFs; download those
+from the original release and join locally with the included script.
+
+| field | meaning |
+|---|---|
+| `index` | question identifier, the join key to the CiteVQA release |
+| `pdf_stem` | source PDF filename without extension |
+| `language` | `en` or `zh` |
+| `qtype` | question type, used for the per-type breakdowns in the paper |
+| `n_pdf_pages` | document length in pages |
+| `alignment` | our verification outcome: `verified` (366) when annotated evidence text was found in the extracted page text, `unverifiable` (353) when the page has no usable text layer to check against, for example a scanned page. Both are retained; questions whose text actively contradicted the annotation were removed. |
+| `gt_pages` | 1-based pages holding necessary evidence |
+| `gt_necessary` | necessary evidence elements as `[page, x1, y1, x2, y2]`; the recall denominator |
+| `gt_all` | necessary plus optional supporting elements; the precision target set |
+
+### Coordinate normalisation
+
+The released CiteVQA evidence boxes are ordered `[y1, x1, y2, x2]` on a 0-1000 scale, which does
+not match the `x1y1x2y2` description in the benchmark's own prompt. We established the true order
+by extracting page text inside candidate boxes and checking it against the annotated evidence
+content. The boxes here are already converted to `x1 y1 x2 y2` in rendered-page pixel
+coordinates, with 1-based page indices. Using the raw released order will silently produce near
+zero recall.
+
+## Usage
+
+```bash
+pip install huggingface_hub
+huggingface-cli download Ryenhails/quote-and-retrieve-eval --repo-type dataset --local-dir eval_set
+```
+
+Load the metadata directly:
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("Ryenhails/quote-and-retrieve-eval", split="train")
+print(ds[0]["index"], ds[0]["gt_necessary"])
+```
+
+To get records the pipeline can run on, join with your own CiteVQA download:
+
+```bash
+python rehydrate.py \
+    --release verified_eval_set.jsonl \
+    --citevqa /path/to/CiteVQA/data/validation/CiteVQA.json \
+    --pdf-dir /path/to/CiteVQA/data/pdf \
+    --out citevqa_singledoc_pub.jsonl
+```
+
+The output is byte-identical in schema to what `src/build_citevqa_pub.py` produces in the code
+repository, so every downstream script runs unchanged. See
+[docs/using_the_release.md](https://github.com/Ryenhails/quote-and-retrieve/blob/main/docs/using_the_release.md)
+for the end-to-end path from this file to the paper's tables.
+
+## Scoring
+
+A citation matches an annotated element when both lie on the same page and their IoU is at least
+0.5. Recall is computed per question over `gt_necessary` and macro-averaged over questions;
+precision is scored against `gt_all`, so citing optional supporting evidence is not an error. The
+exact semantics, including the two protocol choices that are ours rather than the benchmark's, are
+in
+[docs/scoring_protocol.md](https://github.com/Ryenhails/quote-and-retrieve/blob/main/docs/scoring_protocol.md)
+and are asserted by the repository's test suite.
+
+## Licensing
+
+The metadata and derived geometry in this release are MIT licensed. The underlying questions,
+answers, annotations, and documents belong to CiteVQA and remain under its terms; obtain them
+from the original release.
+
+## Citation
+
+```bibtex
+@article{liu2026evidence,
+  title         = {Evidence Attribution in Visual Document Understanding without Coordinates or Region Labels},
+  author        = {Liu, Zhuchenyang and Zhang, Yao and Xiao, Yu},
+  journal       = {arXiv preprint arXiv:2607.24651},
+  year          = {2026},
+  eprint        = {2607.24651},
+  archivePrefix = {arXiv},
+  primaryClass  = {cs.CV},
+  url           = {https://arxiv.org/abs/2607.24651}
+}
+```
+
+Please also cite CiteVQA, whose annotations this release builds on.

rehydrate.py

@@ -0,0 +1,88 @@
+#!/usr/bin/env python3
+"""Join this release with your own CiteVQA download to get runnable records.
+
+This dataset ships identifiers, our verification metadata, and our normalised
+ground-truth boxes. It deliberately does not redistribute the benchmark's
+question text, answers, or PDFs. Download those from the original CiteVQA
+release, then run this script to produce the record format the pipeline reads.
+
+    python rehydrate.py \
+        --release verified_eval_set.jsonl \
+        --citevqa /path/to/CiteVQA/data/validation/CiteVQA.json \
+        --pdf-dir /path/to/CiteVQA/data/pdf \
+        --out citevqa_singledoc_pub.jsonl
+
+The output is exactly what `src/build_citevqa_pub.py` in the code repository
+produces, so every downstream script runs unchanged:
+
+    https://github.com/Ryenhails/quote-and-retrieve
+"""
+import argparse
+import json
+import os
+import sys
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument("--release", default="verified_eval_set.jsonl")
+    ap.add_argument("--citevqa", required=True,
+                    help="path to the CiteVQA validation JSON from the original release")
+    ap.add_argument("--pdf-dir", required=True,
+                    help="directory holding the source PDFs")
+    ap.add_argument("--out", default="citevqa_singledoc_pub.jsonl")
+    args = ap.parse_args()
+
+    release = [json.loads(line) for line in open(args.release)]
+    print(f"release records: {len(release)}")
+
+    raw = json.load(open(args.citevqa))
+    if isinstance(raw, dict):
+        raw = raw.get("data") or next(v for v in raw.values() if isinstance(v, list))
+    by_index = {}
+    for item in raw:
+        idx = item.get("index") or item.get("id") or item.get("question_id")
+        if idx is not None:
+            by_index[str(idx)] = item
+    print(f"benchmark records loaded: {len(by_index)}")
+
+    out, missing_q, missing_pdf = [], 0, 0
+    for r in release:
+        src = by_index.get(r["index"])
+        if src is None:
+            missing_q += 1
+            continue
+        pdf = os.path.join(args.pdf_dir, r["pdf_stem"] + ".pdf")
+        if not os.path.exists(pdf):
+            missing_pdf += 1
+            continue
+        out.append({
+            "index": r["index"],
+            "question": src.get("question") or src.get("Question"),
+            "standard_answer": src.get("answer") or src.get("standard_answer"),
+            "language": r["language"],
+            "qtype": r["qtype"],
+            "n_pdf_pages": r["n_pdf_pages"],
+            "alignment": r["alignment"],
+            "pdf_path": pdf,
+            "gt_pages": r["gt_pages"],
+            "gt_necessary": r["gt_necessary"],
+            "gt_all": r["gt_all"],
+        })
+
+    with open(args.out, "w") as f:
+        for d in out:
+            f.write(json.dumps(d, ensure_ascii=False) + "\n")
+
+    print(f"wrote {len(out)} records -> {args.out}")
+    if missing_q:
+        print(f"  {missing_q} identifiers not found in the benchmark file")
+    if missing_pdf:
+        print(f"  {missing_pdf} source PDFs not found in {args.pdf_dir}")
+    if len(out) != len(release):
+        print("  note: the paper's numbers assume all 719 records are present")
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()