Datasets:
Tasks:
Visual Question Answering
Formats:
parquet
Languages:
English
Size:
10K - 100K
ArXiv:
Tags:
medical
License:
flaviagiammarino
commited on
Commit
•
21341ba
1
Parent(s):
3871cc8
Update README.md
Browse files
README.md
CHANGED
@@ -69,9 +69,9 @@ The question-answer pairs are in English.
|
|
69 |
Each instance consists of an image-question-answer triplet.
|
70 |
```
|
71 |
{
|
72 |
-
'image':
|
73 |
-
'question': '
|
74 |
-
'answer': '
|
75 |
}
|
76 |
```
|
77 |
### Data Fields
|
@@ -80,7 +80,12 @@ Each instance consists of an image-question-answer triplet.
|
|
80 |
- `'answer'`: the expected answer.
|
81 |
|
82 |
### Data Splits
|
83 |
-
The dataset is split into training, validation and test. The split is provided directly by the authors.
|
|
|
|
|
|
|
|
|
|
|
84 |
|
85 |
## Additional Information
|
86 |
|
|
|
69 |
Each instance consists of an image-question-answer triplet.
|
70 |
```
|
71 |
{
|
72 |
+
'image': <PIL.JpegImagePlugin.JpegImageFile image mode=CMYK size=309x272>,
|
73 |
+
'question': 'Where are liver stem cells (oval cells) located?',
|
74 |
+
'answer': 'in the canals of hering'
|
75 |
}
|
76 |
```
|
77 |
### Data Fields
|
|
|
80 |
- `'answer'`: the expected answer.
|
81 |
|
82 |
### Data Splits
|
83 |
+
The dataset is randomly split into training, validation and test. The split is provided directly by the authors.
|
84 |
+
|
85 |
+
| | Training Set | Validation Set | Test Set |
|
86 |
+
|-------------------------|-------------:|---------------:|---------:|
|
87 |
+
| Question-answer pairs |19,654 |6,259 |6,719 |
|
88 |
+
| Images |2,599 |832 |858 |
|
89 |
|
90 |
## Additional Information
|
91 |
|