Update README.md

README.md

@@ -1,156 +1,114 @@
 ---
 license: cc-by-4.0
+pretty_name: MedPhyGraph Support-Graph Data
+viewer: false
 tags:
   - digital-twins
   - scene-graphs
-  - healthcare
+  - support-relations
+  - counterfactual-evidence
   - structured-scene-states
-  - reproducibility
   - procedural
+  - robotics
+  - reproducibility
 ---
 
-# MedPhyGraph — Procedural Support-Graph Data
-
-**MedPhyGraph: Counterfactual Support-Graph Maintenance for Dynamic Built-Environment Digital Twins**
-
-| Resource | Link |
-|----------|------|
-| **Project Website** | https://medphygraph.github.io/ |
-| **Model (CF-SupportNet)** | https://huggingface.co/MedPhyGraph/CF-SupportNet |
-| **Paper** | \<PAPER_URL\> |
-| **Code** | \<GITHUB_URL\> |
-
-## Dataset description
-
-This repository contains structured procedural training and evaluation data used in
-the MedPhyGraph paper. It supports the paper's method contribution and is not
-presented as a standalone benchmark contribution.
-
-All redistributed artifacts are **MedPhyGraph-owned procedural structured JSON**.
-No NVIDIA Isaac Sim / Isaac for Healthcare assets, USD binaries, meshes, textures,
-or Isaac-derived structured scene files are included.
-
-This repository contains the **Procedural subset (136 of 217 cases)** of the
-expanded transfer suite. The paper additionally reports **81 Isaac for Healthcare
-cases**, which are **not redistributed here**.
-
-### Paper evaluation vs. this public release
-
-| Suite | Paper (reported) | This repository |
-|-------|------------------|-----------------|
-| Expanded support-transfer cases | **217** total | **136** procedural only |
-| Procedural transfers | 136 | **136** |
-| Isaac for Healthcare transfers | 81 | **0** (excluded) |
-
-The public expanded-transfer subset is **`expanded_transfer_procedural_v1/`**.
-It is **not** the complete 217-case paper suite.
-
-## Components
+# MedPhyGraph Support-Graph Data
 
-### 1. `training_data/`
+Structured procedural training and evaluation data for **MedPhyGraph**, from *"MedPhyGraph: Counterfactual Support-Graph Maintenance for Dynamic Built-Environment Digital Twins"* (ECCV 2026, TwinWorld Workshop). This repository provides the candidate-edge training split and procedural evaluation data used to train and evaluate [`MedPhyGraph/CF-SupportNet`](https://huggingface.co/MedPhyGraph/CF-SupportNet), the learned scorer inside MedPhyGraph. It supports the paper's method contribution and is not presented as a standalone benchmark contribution.
 
-Paper-frozen CF-SupportNet training/evaluation feature dataset (`dataset_hard`).
+| Dataset | Model | Project |
+|---|---|---|
+| You are here | [MedPhyGraph/CF-SupportNet](https://huggingface.co/MedPhyGraph/CF-SupportNet) | [medphygraph.github.io](https://medphygraph.github.io/) |
 
-| File | Description |
-|------|-------------|
-| `dataset.json` | 533 analytic counterfactual edge samples |
-| `split.json` | Scene-level train/val/test split (324 / 101 / 108 samples) |
+## At a glance
 
-**Provenance:** Generated from procedurally authored `hard_000`–`hard_019` scenes via
-`hard_scenes.py` and `generate_counterfactual_dataset.py`. No Isaac-derived scenes.
+| | |
+|---|---|
+| Candidate-edge samples | 533 (train 324 / val 101 / test 108) |
+| Procedural scene states | 54 |
+| Procedural transfer cases | 136 of 217 in the paper's full expanded suite |
+| Isaac for Healthcare cases | 0 — not redistributed (81 in the paper) |
+| License | CC BY 4.0 |
 
-### 2. `procedural_scenes/`
+## Download
 
-Structured **Procedural** evaluation states (TwinWorld Phase-2 dynamic corpus).
+```python
+from huggingface_hub import snapshot_download
 
-| Property | Value |
-|----------|-------|
-| `scene.json` directories | 54 |
-| Geometry | Procedural AABB proxies (`proxy://aabb/...`) |
-
-### 3. `expanded_transfer_procedural_v1/`
-
-Public procedural subset of the paper's expanded transfer evaluation suite.
-
-| Path | Description |
-|------|-------------|
-| `predeclared_manifest.json` | Eligibility manifest (procedural cases only) |
-| `targets.json` | Ground-truth targets (**136** procedural cases) |
-| `scenes/procedural/` | Authored structured scene states for each case |
+snapshot_download(
+    repo_id="MedPhyGraph/support-graph-data",
+    repo_type="dataset",
+    local_dir="./support-graph-data",
+)
+```
 
-## NVIDIA Isaac for Healthcare: reproduction scope
+This downloads the complete public release while preserving the directory structure below. Individual files can also be browsed under **Files and versions**.
 
-This release distinguishes between the fully public Procedural evaluation and
-the Isaac for Healthcare evaluation, which depends on separately licensed
-NVIDIA software and assets.
+The Dataset Viewer is intentionally disabled. The release preserves the paper-frozen raw JSON artifacts as-is, and their schemas differ by directory — reshaping them into one inferred table would mean no longer matching the exact files the paper's numbers were computed from. This is a deliberate choice, not a broken repository.
 
-### 1. Procedural experiments — fully reproducible
+## Contents
 
-The procedural training data, frozen train/validation/test split, structured
-procedural scenes, CF-SupportNet checkpoints, and the Procedural subset of the
-expanded transfer evaluation are publicly released.
+| Released as | Contains | Local MedPhyGraph path |
+|---|---|---|
+| `training_data/` | `dataset.json`, `split.json` — the frozen 324/101/108 candidate-edge split | `outputs/dyphygraph_health/dataset_hard/` |
+| `procedural_scenes/` | 54 procedural structured scene states, plus indexes/transitions | `outputs/dyphygraph_health/scenes_twinworld_phase2/` |
+| `expanded_transfer_procedural_v1/` | 136 procedural transfer cases (predeclared manifest, targets, before/after scene pairs) | `outputs/reviewer_response/expanded_transfer_procedural_v1/` |
 
-The public expanded-transfer subset contains 136 Procedural cases. These
-artifacts support end-to-end reproduction of the released Procedural
-experiments.
+`MANIFEST.json` indexes every released file; `SHA256SUMS.txt` lets you verify byte-for-byte integrity against this exact release.
 
-### 2. Isaac for Healthcare — NVIDIA software and assets are required
+## Scope
 
-The MedPhyGraph paper also reports experiments in NVIDIA Isaac for Healthcare
-environments.
+**Included** — MedPhyGraph-owned, procedurally generated:
+- Candidate-edge training/validation/test split
+- Procedural structured scene states
+- The procedural subset (136 cases) of the paper's expanded transfer suite
 
-NVIDIA Isaac for Healthcare assets, USD/USDA/USDC files, meshes, textures, and
-Isaac-derived structured scene states are not redistributed in this release.
+**Not included:**
+- NVIDIA Isaac for Healthcare assets, USD/USDA/USDC files, meshes, textures
+- Isaac-derived structured scene states
+- The paper's 81 Isaac cases from the expanded transfer suite
 
-Users interested in running MedPhyGraph with Isaac for Healthcare must obtain
-the required NVIDIA Isaac Sim / Isaac for Healthcare software and assets
-directly from NVIDIA under the applicable NVIDIA license terms.
+**136 vs. 217:** this repository contains the Procedural subset only. The paper's expanded transfer suite reports 217 cases total (136 Procedural + 81 Isaac for Healthcare). The 81 Isaac cases are not redistributed here, for the reasons below.
 
-The public MedPhyGraph codebase contains relevant components for working with
-locally installed Isaac environments, but the complete paper-frozen Isaac scene
-generation workflow is not currently packaged as a standalone end-to-end public pipeline.
+## NVIDIA Isaac for Healthcare
 
-### 3. Scope of the paper-frozen Isaac evaluation
+NVIDIA Isaac for Healthcare assets are not redistributed in this release. To work with the Isaac side of MedPhyGraph: obtain NVIDIA's software and assets directly from NVIDIA under NVIDIA's own license terms, then use MedPhyGraph's tooling locally to construct structured scene states from those assets.
 
-The paper's expanded transfer evaluation contains 217 cases in total:
+| Resource | Link |
+|---|---|
+| NVIDIA Isaac for Healthcare | [developer.nvidia.com/isaac/healthcare](https://developer.nvidia.com/isaac/healthcare) |
+| Getting started | [docs.nvidia.com — Getting Started](https://docs.nvidia.com/learning/physical-ai/getting-started-with-isaac-for-healthcare/latest/index.html) |
+| Setup guide | [docs.nvidia.com — Setup](https://docs.nvidia.com/learning/physical-ai/getting-started-with-isaac-for-healthcare/latest/training-healthcare-robots-from-scratch/01-setup/01-setup.html) |
+| Loading custom assets | [docs.nvidia.com — Asset loading](https://docs.nvidia.com/learning/physical-ai/getting-started-with-isaac-for-healthcare/latest/training-healthcare-robots-from-scratch/02-data-loading/02-loading-custom-assets.html) |
+| Isaac for Healthcare on GitHub | [github.com/isaac-for-healthcare](https://github.com/isaac-for-healthcare) |
+| i4h-workflows | [github.com/isaac-for-healthcare/i4h-workflows](https://github.com/isaac-for-healthcare/i4h-workflows) |
 
-- 136 Procedural cases
-- 81 Isaac for Healthcare cases
+## Reproducibility
 
-This repository redistributes the 136 Procedural cases only. The 81 Isaac cases
-and their corresponding structured scene states are not included.
+This release distinguishes three levels. Read this before assuming what is or isn't reproducible from the public artifacts alone.
 
-Accordingly, the exact Isaac results reported in the paper should not be
-expected to be reproducible from the currently released public artifacts alone.
-Reconstructing that portion requires the appropriate NVIDIA software and assets
-as well as additional MedPhyGraph generation tooling.
+**1. Procedural experiments — fully reproducible.**
+Everything needed is in this repository: training data, procedural scenes, and the procedural transfer suite.
 
-This limitation does not affect reproduction of the publicly released
-Procedural experiments.
+**2. Isaac for Healthcare environments — reproducible, with software and assets obtained separately.**
+This is a standard, permanent boundary reflecting asset licensing, not a gap in what MedPhyGraph provides. Use the NVIDIA links above to obtain Isaac Sim / Isaac for Healthcare, then apply MedPhyGraph's tooling locally.
 
-## Expected layout after download
+**3. The paper's exact frozen Isaac results — not yet available as a packaged public pipeline.**
+The structured states behind the paper's 81 Isaac cases are not redistributed, and the full generation workflow used to produce them is not yet packaged as a one-command public pipeline. Some relevant MedPhyGraph modules are already public; the complete workflow is still being prepared. Until then, regenerating Isaac scenes with the public tooling can support validating MedPhyGraph's qualitative findings on Isaac data, but should not be expected to reproduce the paper's exact reported numbers.
 
-Mirror paths relative to the MedPhyGraph code repository:
+## Relationship to CF-SupportNet
 
-```
-outputs/dyphygraph_health/dataset_hard/              # from training_data/
-outputs/dyphygraph_health/scenes_twinworld_phase2/   # from procedural_scenes/
-outputs/reviewer_response/expanded_transfer_procedural_v1/
-```
+[`MedPhyGraph/CF-SupportNet`](https://huggingface.co/MedPhyGraph/CF-SupportNet) is the learned candidate-edge scorer used within MedPhyGraph, released as five paper-frozen checkpoints. MedPhyGraph itself additionally applies deterministic State Consistency and Union-Based Transition-Aware Consistency on top of CF-SupportNet's scores — these are non-learned, code-only operations, not part of the checkpoint. This dataset provides the data used to train and evaluate CF-SupportNet; it does not itself constitute or require the full MedPhyGraph framework to use.
 
 ## Integrity
 
-- `SHA256SUMS.txt` — per-file SHA256 digests
-- `MANIFEST.json` — public metadata index (relative paths only)
+- **`MANIFEST.json`** — index of every artifact in this release
+- **`SHA256SUMS.txt`** — per-file hashes for verifying an exact, unmodified copy of the frozen release
 
-## Not included
+## License
 
-| Artifact | Reason |
-|----------|--------|
-| `scenes_twinworld_i4h_dynamic/` | Isaac-derived structured states — not redistributed |
-| Isaac expanded-transfer cases (81) | Isaac-derived — excluded from public subset |
-| NVIDIA / i4h USD assets | Third-party license |
-| Canonical rebuild `outputs/medphygraph/` | Different research track |
+The MedPhyGraph-owned content in this repository is released under **CC BY 4.0**. NVIDIA Isaac for Healthcare assets are not included in this repository and are therefore not being relicensed by MedPhyGraph; they remain governed by NVIDIA's own terms.
 
 ## Citation
 
@@ -161,13 +119,4 @@ outputs/reviewer_response/expanded_transfer_procedural_v1/
   booktitle={ECCV 2026 TwinWorld Workshop},
   year={2026}
 }
-```
-
-## License
-
-This dataset is released under the **Creative Commons Attribution 4.0 International
-License (CC BY 4.0)**. See the `LICENSE` file in the MedPhyGraph code repository
-or https://creativecommons.org/licenses/by/4.0/.
-
-Third-party NVIDIA Isaac for Healthcare assets are **not** included and are **not**
-covered by this license.
+```