PhysMoE paper proposes single-token coordinate representation for spatial grounding in VLMs
A new arXiv paper introduces PhysMoE, a modality-decoupled vision-language model approach using single-token coordinate representation to speed up spatial grounding for embodied AI tasks.
Read the full story at arXiv cs.CV ↗
Timeline · 1 report
- 2026-11-11 05:00 · arXiv cs.CV
PhysMoE: Fast Spatial Grounding with Single-Token Coordinate Representation in Modality-Decoupled VLMs