Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -1,27 +1,26 @@
|
|
| 1 |
---
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
dtype: binary
|
| 15 |
-
- name: original_row_index
|
| 16 |
-
dtype: int64
|
| 17 |
-
- name: original_filename
|
| 18 |
-
dtype: string
|
| 19 |
-
- name: original_caption
|
| 20 |
-
dtype: string
|
| 21 |
-
splits:
|
| 22 |
-
- name: train
|
| 23 |
-
num_bytes: 38109167352
|
| 24 |
-
num_examples: 254138
|
| 25 |
-
download_size: 36808549629
|
| 26 |
-
dataset_size: 38109167352
|
| 27 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- question-answering
|
| 5 |
+
- image-to-text
|
| 6 |
+
language:
|
| 7 |
+
- en
|
| 8 |
+
tags:
|
| 9 |
+
- vlm
|
| 10 |
+
- visual-question-answering
|
| 11 |
+
- dataset-polishing
|
| 12 |
+
size_categories:
|
| 13 |
+
- 100K<n<1M
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
---
|
| 15 |
+
|
| 16 |
+
# Polished VLM Dataset
|
| 17 |
+
|
| 18 |
+
This dataset contains polished and augmented Visual Question Answering (VQA) pairs designed for training and evaluation of Vision-Language Models (VLMs).
|
| 19 |
+
|
| 20 |
+
## Dataset Structure
|
| 21 |
+
|
| 22 |
+
- **Columns:** `question`, `answer`, `image`, `original_row_index`, `original_filename`, `original_caption`
|
| 23 |
+
- **Total Rows:** 254,138
|
| 24 |
+
|
| 25 |
+
## Summary
|
| 26 |
+
This dataset is uploaded directly from cleaned, parquet-sharded preprocessing outputs.
|