VisGym/visgym_data
Image Text To TextEN
VisGym/visgym_data is a image text to text-focused dataset in EN distributed in Parquet format.
About VisGym/visgym_data
VisGym Dataset
Project Page | Paper | GitHub
VisGym consists of 17 diverse, long-horizon environments designed to systematically evaluate, diagnose, and train Vision-Language Models (VLMs) on visually interactive tasks. In these environments, a...
Details
- Task
- Image Text To Text
- Language
- EN
- Format
- Parquet
- Rows / instances
- N/A
- Creator
- VisGym
- Year
- 2026