New Method Restores Language Capabilities in Vision-Language-Action Models via Parameter Merging
A new arXiv paper proposes parameter merging to restore language capabilities in vision-language-action models that are lost when VLMs are fine-tuned on robot demonstration data for action tasks.
Read the full story at arXiv cs.AI ↗
Timeline · 1 report
- 2026-11-17 05:00 · arXiv cs.AI
Bring Back Language to Actions: Restoring VLM capabilities in VLA Models via Parameter Merging