Hugging Face's logo

Datasets:
Video-Reason
/
VBVR-Pro-Bench

Modalities:
Image
Text
Formats:
parquet
Languages:
English
Size:
< 1K
ArXiv:
Tags:
video-generation
image-to-video
visual-reasoning
benchmark
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
junxiangjoe commited on
Commit
31bf80c
·
verified ·
1 Parent(s): ea06c24

Update dataset card: badges, scorer link

Browse files
Files changed (1) hide show
  1. README.md +20 -2
README.md CHANGED
@@ -13,9 +13,27 @@ size_categories:
13
 
14
  # VBVR-Pro-Bench
15
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
  The frozen **evaluation split** of VBVR-Pro. 100 parameterized reasoning tasks, 5 held-out instances each, in both a video (I2V) and an interleaved-image setting.
17
 
18
- Companion resources: [VBVR-Pro-SFT-Video](https://huggingface.co/datasets/Video-Reason/VBVR-Pro-SFT-Video) · [VBVR-Pro-SFT-Image](https://huggingface.co/datasets/Video-Reason/VBVR-Pro-SFT-Image) · [VBVR-Pro-RL](https://huggingface.co/datasets/Video-Reason/VBVR-Pro-RL)
19
 
20
  ## At a glance
21
 
@@ -85,7 +103,7 @@ huggingface-cli download Video-Reason/VBVR-Pro-Bench --repo-type dataset --local
85
  cd VBVR-Pro-Bench && tar xzf VBVR-Pro-Bench-Video.tar.gz
86
  ```
87
 
88
- Generate one video (or image sequence) per instance from `first_frame.png` + `prompt.txt`, then score with [VBVR-EvalKit](https://github.com/Video-Reason/VBVR-EvalKit).
89
 
90
  > This benchmark is **1024 × 1024**, as generated. The training splits
91
  > ([SFT-Video](https://huggingface.co/datasets/Video-Reason/VBVR-Pro-SFT-Video),
 
13
 
14
  # VBVR-Pro-Bench
15
 
16
+ <p align="center">
17
+ <a href="https://huggingface.co/datasets/Video-Reason/VBVR-Pro-SFT-Video" target="_blank">
18
+ <img alt="VBVR-Pro-SFT-Video" src="https://img.shields.io/badge/%F0%9F%A4%97%20_VBVR_Pro_SFT-Video-ffc107?color=ffc107&logoColor=white" height="20" />
19
+ </a>
20
+ <a href="https://huggingface.co/datasets/Video-Reason/VBVR-Pro-SFT-Image" target="_blank">
21
+ <img alt="VBVR-Pro-SFT-Image" src="https://img.shields.io/badge/%F0%9F%A4%97%20_VBVR_Pro_SFT-Image-ffc107?color=ffc107&logoColor=white" height="20" />
22
+ </a>
23
+ <a href="https://huggingface.co/datasets/Video-Reason/VBVR-Pro-RL" target="_blank">
24
+ <img alt="VBVR-Pro-RL" src="https://img.shields.io/badge/%F0%9F%A4%97%20_VBVR_Pro-RL-ffc107?color=ffc107&logoColor=white" height="20" />
25
+ </a>
26
+ <a href="https://github.com/Video-Reason/VBVR-Pro-Bench" target="_blank">
27
+ <img alt="Scorer" src="https://img.shields.io/badge/Scorer-VBVR_Pro_Bench-100000?style=flat-square&logo=github&logoColor=white" height="20" />
28
+ </a>
29
+ <a href="LICENSE">
30
+ <img alt="License" src="https://img.shields.io/badge/License-Apache_2.0-blue.svg" height="20" />
31
+ </a>
32
+ </p>
33
+
34
+
35
  The frozen **evaluation split** of VBVR-Pro. 100 parameterized reasoning tasks, 5 held-out instances each, in both a video (I2V) and an interleaved-image setting.
36
 
 
37
 
38
  ## At a glance
39
 
 
103
  cd VBVR-Pro-Bench && tar xzf VBVR-Pro-Bench-Video.tar.gz
104
  ```
105
 
106
+ Generate one video (or image sequence) per instance from `first_frame.png` + `prompt.txt`, then score with [VBVR-Pro-Bench](https://github.com/Video-Reason/VBVR-Pro-Bench).
107
 
108
  > This benchmark is **1024 × 1024**, as generated. The training splits
109
  > ([SFT-Video](https://huggingface.co/datasets/Video-Reason/VBVR-Pro-SFT-Video),