Add task category and link to paper
#2
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,4 +1,6 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
dataset_info:
|
| 3 |
features:
|
| 4 |
- name: id
|
|
@@ -36,6 +38,8 @@ configs:
|
|
| 36 |
|
| 37 |
# VDiff-Bench
|
| 38 |
|
|
|
|
|
|
|
| 39 |
VDiff-Bench is a multiple-choice benchmark for fine-grained visual difference identification. Each example presents two similar images and four candidate descriptions, exactly one of which states a real difference between the images.
|
| 40 |
|
| 41 |
## Dataset structure
|
|
@@ -78,4 +82,4 @@ example = dataset[0]
|
|
| 78 |
|
| 79 |
## Citation
|
| 80 |
|
| 81 |
-
Citation information will be added with the accompanying paper release.
|
|
|
|
| 1 |
---
|
| 2 |
+
task_categories:
|
| 3 |
+
- image-to-text
|
| 4 |
dataset_info:
|
| 5 |
features:
|
| 6 |
- name: id
|
|
|
|
| 38 |
|
| 39 |
# VDiff-Bench
|
| 40 |
|
| 41 |
+
[Paper](https://huggingface.co/papers/2609.06245) | [Project page](https://huggingface.co/spaces/elaine1wan/vdiff-bench)
|
| 42 |
+
|
| 43 |
VDiff-Bench is a multiple-choice benchmark for fine-grained visual difference identification. Each example presents two similar images and four candidate descriptions, exactly one of which states a real difference between the images.
|
| 44 |
|
| 45 |
## Dataset structure
|
|
|
|
| 82 |
|
| 83 |
## Citation
|
| 84 |
|
| 85 |
+
Citation information will be added with the accompanying paper release.
|