AI & ML
Finetuning VLAs for Dexterous Manipulation: The Right Checkpoint Is All You Need
arXiv cs.RO0 views
arXiv:2611.23676v1 Announce Type: new Abstract: Vision-Language-Action models have demonstrated strong manipulation capabilities. A limitation of VLA models is that most manipulation datasets focus on parallel-jaw gripper systems, making transfer to multi-finger hands challenging, particularly as teleoperation datasets are harder to collect with multi-fingered robots. In this work, we study how to best adapt VLAs to dexterous hands by pairing supervised fine-tuning (SFT) with diffusion-steering reinforcement learning (DSRL). Using $\pi_{0.5}$ and a LEAP hand, we evaluate five fine-tuning recipes across two simulation tasks using only a handful of demonstrations. Our empirical analysis yields three key insights: (1) SFT success is poorly correlated with post-DSRL success, meaning the best SFT checkpoint is rarely the most effective to steer; (2) freezing the vision-language model (VLM) and tuning only the action expert produces the most steerable policy for both training and out-of-distribution placements; and (3) low-rank adaptation (LoRA) is the least steerable approach. Consequently, we recommend an adaptation recipe that freezes the VLM, fully trains the action expert, and eschews SFT rollout success as a checkpoint selection metric.
Source:https://arxiv.org/abs/2611.23676PreviousRadar-Native Control Barrier Functions: Doppler Range Rate as a Measured Barrier DerivativeNextMeasuring the Conservatism of Latent Safety Filters for World Models
Related posts

Why the Harness Matters: Early Lessons From the Evo Code Scanning Agent Preview
0Snyk Blog
Green software isn't a feature flag — lessons from making conversion pipelines efficient
0dev.to AI
日本AI・LLMデイリーブリーフィング (2026年11月20日)
0dev.to AI
Japan AI & LLM Daily Briefing (Nov 20, 2026)
0dev.to AI
AWS Added GPUs to Lambda. Three Workers Can Mean Three Model Copies.
0dev.to AWS
Matching the Same Product Across Five Stores: How Deal Sites Deduplicate Listings
0dev.to WebDev