WebRISE is a benchmark for evaluating web artifacts generated by multimodal large language models. The dataset contains task specifications, multimodal inputs, ground-truth HTML artifacts, and Interaction Contract Graphs for browser-based evaluation. It was created by IIGroup and last updated on June 3, 2026.
Use Cases
- Benchmarking MLLM-generated HTML quality based on provided ground-truth artifacts
- Evaluating model compliance with task requirements using annotated task specifications
- Analyzing multimodal input-to-output consistency using the multimodal inputs
- Testing browser-based interaction logic via the Interaction Contract Graphs
Strengths
- Includes multiple structured components: task specifications, multimodal inputs, ground-truth HTML artifacts, and Interaction Contract Graphs
- Provides a benchmark specifically designed for browser-based evaluation of web artifacts
- Created by a known organization (IIGroup) and updated in 2026
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
- IIGroup
- Freshness
- Last updated 2026-06-03 03:03:46