docs: finalize PAWBench release metadata

Publish arXiv citation, official CLI workflow, release metadata, and explicit code/data license boundaries for PIC-157.

README.md

@@ -6,22 +6,22 @@ tags:
   - benchmark
   - video-generation
   - physical-reasoning
+  - arxiv:2608.27345
 ---
 
 # PAWBench
 
 ### A benchmark for distributional physical realism in generated videos
 
-> [!IMPORTANT]
-> **Private preview.** This repository contains the current materialized
-> PAWBench input package for review and integration. It is not yet a public
-> benchmark release or leaderboard submission endpoint.
-
 PAWBench asks whether a video model reproduces the *range and frequency* of
 outcomes that a physical scene can produce. It evaluates repeated rollouts from
 the same source image and action, rather than treating one plausible-looking
 video as sufficient evidence of physical realism.
 
+[Paper](https://arxiv.org/abs/2608.27345) ·
+[Code](https://github.com/Andrew0613/PAWBench) ·
+[Project website](https://pawbench.github.io/)
+
 ## At a glance
 
 | Component | Contents |
@@ -82,31 +82,27 @@ Illustrative Calibration record:
 
 ## Using the data
 
-Clone or download this dataset locally, generate repeated videos for each
-scene, then pass local video paths to the PAWBench evaluator:
-
-```python
-from pathlib import Path
-
-from pawbench import evaluate
-
-result = evaluate(
-    benchmark_path=Path("/path/to/PAWBench"),
-    videos=[
-        {
-            "sample_id": "my-model::A-01::r000",
-            "scene_id": "A-01",
-            "repeat_index": 0,
-            "video_path": "/results/A-01/r000.mp4",
-        }
-    ],
-    model_or_lane="my-model",
-    vlm={
-        "base_url": "https://your-vlm-provider.example/v1",
-        "model": "your-vlm-model",
-        "api_key_env": "YOUR_VLM_API_KEY",
-    },
-)
+Download this dataset, generate the complete rollout grid described by the
+[PAWBench evaluator](https://github.com/Andrew0613/PAWBench), and run the
+repository's command-line workflow:
+
+```bash
+hf download Andrew613/PAWBench \
+  --repo-type dataset \
+  --local-dir /path/to/PAWBench-data
+
+git clone https://github.com/Andrew0613/PAWBench.git
+cd PAWBench
+pip install -r requirements.txt
+
+python evaluate.py \
+  --benchmark /path/to/PAWBench-data \
+  --videos /path/to/my-model-rollouts \
+  --output runs/my-model/evaluation \
+  --model my-model \
+  --vlm-base-url https://openrouter.ai/api/v1 \
+  --vlm-model google/gemini-3.5-flash \
+  --vlm-api-key-env OPENROUTER_API_KEY
 ```
 
 The evaluator derives the full 50-scene × 50-rollout grid from the scene
@@ -122,12 +118,34 @@ This dataset repository contains benchmark inputs only:
 - generated model videos, provider responses, and experiment result bundles are
   intentionally excluded.
 
-The private preview preserves the currently materialized 50-scene package. Its
-final public roster and evaluator-rubric alignment are being completed before a
-public benchmark release.
+This public repository preserves the materialized 50-scene benchmark package.
+Record the exact Hugging Face revision together with the evaluator commit for
+reproducible evaluations.
 
 ## Citation
 
-The paper and canonical citation will be added with the public release. For
-private preview work, record the exact Hugging Face revision and GitHub commit
-used for an evaluation.
+If you use PAWBench, please cite the paper and record the exact Hugging Face
+revision and GitHub commit used for your evaluation:
+
+```bibtex
+@article{pu2026pawbench,
+  title={PAWBench: How Far Are We from Probabilistically Aligned World Modeling?},
+  author={Yuandong Pu and Le Zhuo and Sayak Paul and Gabriel Jorge Menezes and Avram Đorđević and Shiyang Li and Yifan Zhou and Bin Fu and Wenlong Zhang and Junjun He and Yu Qiao and Yihao Liu and Jinbo Xing and Xi Chen},
+  journal={arXiv preprint arXiv:2608.27345},
+  year={2026},
+  eprint={2608.27345},
+  archivePrefix={arXiv},
+  primaryClass={cs.CV},
+  url={https://arxiv.org/abs/2608.27345}
+}
+```
+
+## License and usage
+
+The Apache-2.0 license in the
+[PAWBench code repository](https://github.com/Andrew0613/PAWBench) applies to
+the evaluator source code, not automatically to this dataset's images, prompt
+material, or scene metadata. No separate license has been selected for this
+dataset repository. Do not infer permission to redistribute or create
+derivatives of its assets without permission from the project owners. Source
+and package provenance is recorded in `scenes.jsonl` and `manifest.json`.

manifest.json

@@ -1,6 +1,7 @@
 {
   "schema_version": "pawbench.benchmark_inputs/v1",
   "dataset_id": "PAWBench",
+  "release_version": "0.1.0",
   "benchmark_revision": 2,
   "scene_count": 50,
   "splits": {
@@ -18,6 +19,13 @@
       "owner_accepted_at": "2026-08-14"
     }
   },
+  "publication": {
+    "title": "PAWBench: How Far Are We from Probabilistically Aligned World Modeling?",
+    "arxiv_id": "2608.27345",
+    "arxiv_url": "https://arxiv.org/abs/2608.27345",
+    "code_url": "https://github.com/Andrew0613/PAWBench",
+    "project_url": "https://pawbench.github.io/"
+  },
   "upstream": {
     "repository": "Andrew613/PAWBench-Results",
     "revision": "5fdea8a1e7a1e6ccf69bf5af9cf7947aefc58190",
@@ -41,5 +49,5 @@
       ]
     }
   ],
-  "claim_boundary": "Benchmark input materialization only; no generation, evaluation, metric, scientific, manuscript, or publication acceptance."
+  "claim_boundary": "Benchmark input materialization only; no generation, evaluation, metric, or scientific-result acceptance."
 }