Mobile-GUI-Worldmodel-SFT: Mobile GUI Agent Trajectories for World Model Training

Mobile GUI agent data for training and evaluating GUI world models, organized around GUI trajectories. Each step includes a screenshot and page-state annotations such as HTML, plain text, and structured text. The dataset was authored by xwk123 and last updated on 2026-05-11.

Use Cases

• Train GUI world models based on annotated mobile GUI trajectories.
• Evaluate agent performance based on provided evaluation assets.
• Develop multimodal agents using the combination of screenshots and structured page-state annotations.

Strengths

• Data is organized around GUI trajectories, providing a structured format for training.
• Each trajectory step includes multimodal data: a screenshot and multiple annotation types (HTML, plain text, structured text).

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

huggingface

Collection Method

Data annotation scripts and examples are included in the repository.

Freshness

Last updated 2026-05-11 03:24:05.