MolmoWeb-SyntheticGround: Webpage Screenshots with Synthetic Grounding Tasks

AllenAI created a dataset of webpage screenshots paired with synthetic grounding tasks. Each example asks a model to identify a target element on the page, with ground-truth bounding boxes and natural-language thoughts for a subset. The dataset was last updated on March 24, 2026.

Use Cases

• Train object detection models based on ground-truth bounding boxes for webpage elements.
• Evaluate visual reasoning models based on tasks requiring identification of target elements on screenshots.
• Benchmark multimodal models based on tasks combining image and natural-language thought data.

Strengths

• Dataset includes ground-truth bounding boxes for target elements.
• A subset of examples includes natural-language thoughts, likely for GPT-based models.
• Dataset is hosted by AllenAI, a known research organization.

Limitations

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

Provenance

Source

allenai

Collection Method

Likely contains synthetically generated grounding tasks paired with webpage screenshots.

Time Range

null

Freshness

Last updated 2026-03-24 02:34:29; freshness should be verified.

Geography

null