Dataset · ScaleAI
Intuitive Visual Reasoning Benchmark
ScaleAI/HSS
A collection of images and short videos with questions from Scale AI to evaluate models' ability to understand hidden context.
What's inside
Contains English-language images, short video clips, and tabular data, with a dataset size of under 1,000 examples. Each task includes a media file, a human-written question, a reference answer, and a set of evaluation criteria. Created by the Scale AI team to test how well models capture implicit temporal, spatial, social, and abstract relationships.
What you can do with it
- Evaluating multimodal LLMs on visual context understanding tasks (VQA)
- Benchmarking video models for logical reasoning on short clips
- Comparing neural network reasoning quality against human level (93.1%)
- Tasks
- Visual question answering, Video text → text
- Languages
- English
- Size
- up to 1K rows
- Downloads this month
- 3.8K
- Likes
- 19
How to load it
from datasets import load_dataset
ds = load_dataset("ScaleAI/HSS")