Spatial Language Models & 3D Scene Understanding

11 repos

Multimodal language models designed to understand and reason about 3D spatial data, particularly point clouds and scene geometry. These repositories focus on extending LLMs with spatial intelligence capabilities, enabling models to process spatial relationships, scene context, and 3D structures alongside traditional language understanding. The cluster centers on SpatialLM variants across different base models (Llama, Qwen) and scales, alongside supporting utilities and datasets for training and evaluating spatial reasoning in language models.

Related clusters