Pre-rendered 3D multi-room environments support the Theory of Space benchmark for evaluating spatial reasoning in Vision Language Models. The dataset is designed to test whether foundation models can construct spatial beliefs through active exploration. It was created by MLL-Lab and last updated on February 11, 2026.
Use Cases
- Benchmarking spatial reasoning in VLMs based on pre-rendered 3D environments
- Testing active exploration and belief construction in models based on multi-room scenes
- Evaluating foundational model performance on the Theory of Space benchmark
Strengths
- Dataset is specifically designed for the Theory of Space benchmark, providing a targeted evaluation tool.
- Contains pre-rendered 3D multi-room environments, which suggests a controlled and consistent test setting.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- MLL-Lab
- Collection Method
- Pre-rendered 3D environments, likely generated for benchmark purposes.
- Freshness
- Last updated 2026-02-11 03:27:39; freshness should be verified.