Breaking the 'Lock-In' Dilemma: Preserving Steerability in Low-Data VLA Post-Training
New research reveals that vision-language-action models exhibit a 'lock-in' phenomenon after few-shot fine-tuning, causi…
🏷️ VLA Models
3 articles about 'VLA Models'
Comprehensive Survey on VLA Robot Data Infrastructure Released
A new arXiv survey systematically reviews datasets, benchmarks, and data engines for Vision-Language-Action models in ro…
Libra-VLA: Dual-System Architecture Solves Hierarchical Challenges in Robot Manipulation
A research team proposes the Libra-VLA framework, which addresses the problem of single-generation paradigms in vision-l…
1