Visual Instruction Bottleneck Tuning
Paper • 2505.13946 • Published • 10
Datasets:
Dataset Viewer
The dataset could not be loaded because the splits use different data file formats, which is not supported. Read more about the splits configuration. Click for more details.
Couldn't infer the same data file format for all splits. Got {NamedSplit('validation'): ('imagefolder', {}), NamedSplit('test'): ('json', {})}
Error code: FileFormatMismatchBetweenSplitsError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Vittle - Visually Perturbed POPE Benchmark
This dataset provides visually perturbed variants of the POPE (Polling-based Object Probing Evaluation) benchmark, built on COCO val2014 images. It is released as part of the Vittle (Visual Instruction Bottleneck Tuning) project (NeurIPS 2025).
Overview
- Questions: 9,000 yes/no object hallucination probing questions (3,000 each for adversarial / popular / random splits)
- Images: 500 unique COCO val2014 images, each with 9 visual perturbation variants (severity level 3)
- Total image files: 4,500 (500 images x 9 perturbations)
Visual Perturbations
All perturbations are at severity level 3, generated following MM-Robustness:
| Perturbation | Folder |
|---|---|
| Gaussian Noise | images/COCO_IP_gaussian_noise_3/ |
| Shot Noise | images/COCO_IP_shot_noise_3/ |
| Speckle Noise | images/COCO_IP_speckle_noise_3/ |
| Fog | images/COCO_IP_fog_3/ |
| Contrast | images/COCO_IP_contrast_3/ |
| Brightness | images/COCO_IP_brightness_3/ |
| Defocus Blur | images/COCO_IP_defocus_blur_3/ |
| Zoom Blur | images/COCO_IP_zoom_blur_3/ |
| Frost | images/COCO_IP_frost_3/ |
File Structure
.
├── README.md
├── llava_pope_test.jsonl # 9,000 questions
├── annotations/
│ ├── coco_pope_adversarial.json # 3,000 adversarial split labels
│ ├── coco_pope_popular.json # 3,000 popular split labels
│ └── coco_pope_random.json # 3,000 random split labels
└── images/
├── COCO_IP_gaussian_noise_3/ # 500 images
├── COCO_IP_shot_noise_3/
├── COCO_IP_speckle_noise_3/
├── COCO_IP_fog_3/
├── COCO_IP_contrast_3/
├── COCO_IP_brightness_3/
├── COCO_IP_defocus_blur_3/
├── COCO_IP_zoom_blur_3/
└── COCO_IP_frost_3/
Question Format (JSONL)
{"question_id": 0, "image": "COCO_val2014_000000007991.jpg", "text": "Is there a snowboard in the image?\nAnswer the question using a single word or phrase.", "category": "adversarial"}
Citation
@inproceedings{
oh2025visual,
title={Visual Instruction Bottleneck Tuning},
author={Changdae Oh and Jiatong Li and Shawn Im and Sharon Li},
booktitle={The Thirty-ninth Annual Conference on Neural Information Processing Systems},
year={2025},
url={https://openreview.net/forum?id=yzHiEmLSk8}
}
License
MIT
- Downloads last month
- 1,710