# MultiUI

7.3 million training samples built from webpage UIs and structured with enhanced accessibility trees, for text-rich visual understanding, released with UIX-Qwen2 models.

- Catalog URL: https://awesomecua.github.io/resources/github-neulab-multiui/
- Original source: https://github.com/neulab/multiui
- Kind: dataset
- Review: curated
- Scope: core
- Environments: web
- Topics: screen-understanding, data-generation
- Evidence checked: 2026-10-16T06:36:01Z
- ID: github:neulab/multiui

## Limitations

Inclusion does not establish reproducibility or endorsement.

- Record JSON: https://awesomecua.github.io/resources/github-neulab-multiui/record.json

## Structured profile

- formats: not recorded
- modalities: webpage UI content, text, accessibility-derived structure
- splits: training collection
- supervision: UI understanding, text-rich visual tasks

### Count observations

- ≈ 7,300,000 samples — reported MultiUI training collection (reported; source: https://raw.githubusercontent.com/neulab/multiui/HEAD/README.md)

The README provides a rounded sample count. It does not establish a unique-screenshot inventory or a normalized split breakdown. Synthetic examples and publicly sourced web content have provenance and quality limits; data terms must be checked separately from repository code.

## License declarations

- code: not recorded
- data: not recorded
- document: not recorded
- weights: not recorded

## Primary evidence

- https://raw.githubusercontent.com/neulab/multiui/HEAD/README.md (retrieved 2026-10-16T06:36:01Z; SHA-256 fe0c03fe73529a29788d559103c4aed48dca6ed9ba3bfd5f151f902fa8858b38)
- https://raw.githubusercontent.com/neulab/multiui/HEAD/README.md (retrieved 2026-10-16T06:36:01Z; SHA-256 fe0c03fe73529a29788d559103c4aed48dca6ed9ba3bfd5f151f902fa8858b38)