A high-confidence subset of VisualWebInstruct curated by TIGER-Lab, last updated October 24, 2025. It contains verified multimodal question–answer pairs where correctness, reasoning quality, and image–text alignment have been explicitly validated. The dataset is designed for Reinforcement Learning and Reward Model training pipelines.
Use Cases
- Training reinforcement learning agents based on verified multimodal question–answer pairs.
- Training reward models based on validated reasoning quality and image–text alignment.
- Fine-tuning vision-language models on high-confidence instruction-following data.
Strengths
- Contains verified question–answer pairs where correctness has been explicitly validated.
- Data quality is curated for high-confidence use in Reinforcement Learning training.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- TIGER-Lab
- Collection Method
- Curated subset of VisualWebInstruct.
- Freshness
- Last updated 2025-10-24 15:52:27.