Document metadata.json in layout

README.md

@@ -27,7 +27,7 @@ Companion resources: [VBVR-Pro-SFT-Video](https://huggingface.co/datasets/Video-
 | Settings | Video and Image |
 | Resolution | **1024 × 1024** |
 | Video | 16 fps, MPEG-4 Part 2, as generated (no re-encode) |
-| Size | 83.9 MB (video) + 9.8 MB (image) |
+| Size | 86.0 MB (video) + 11.7 MB (image) |
 
 `In-Domain_50` covers task families that also appear in the training splits; `Out-of-Domain_50` covers task families that do **not**, and is the split to report generalization on.
 
@@ -49,8 +49,9 @@ VBVR-Pro-Bench-Video/
 │   │   │   ├── first_frame.png        # conditioning image  (1024 × 1024)
 │   │   │   ├── prompt.txt             # instruction
 │   │   │   ├── ground_truth.mp4       # reference video     (16 fps)
-│   │   │   └── final_frame.png        # last frame of the reference video
-│   │   ├── 00001/ … 00004/            # same four files
+│   │   │   ├── final_frame.png        # last frame of the reference video
+│   │   │   └── metadata.json          # task parameters, ground truth, scoring contract
+│   │   ├── 00001/ … 00004/            # same five files
 │   │   └── …
 │   └── …                              # 50 tasks
 └── Out-of-Domain_50/                  # 50 tasks, same shape
@@ -67,7 +68,8 @@ VBVR-Pro-Bench-Image/
 │   │   │   ├── prompt.txt             # instruction
 │   │   │   ├── frame_1.png            # reference output, step 1
 │   │   │   ├── frame_2.png            # … step 2, when the task has more steps
-│   │   │   └── …                      # up to frame_N.png
+│   │   │   ├── …                      # up to frame_N.png
+│   │   │   └── metadata.json          # task parameters, ground truth, scoring contract
 │   │   ├── 00001/ … 00004/
 │   │   └── …
 │   └── …                              # 50 tasks