GR00T N1.5 (AI model): news timeline & CVEs

The Low-Rank Structure of VLA Reinforcement Learning

RL post-training of flow-based VLA models produces low-rank updates concentrated in action-expert Timestep Modules.

Researchers find reinforcement learning on flow-based vision-language-action models, including pi0.5 and GR00T N1.5/N1.6, produces substantially lower-rank updates concentrated in the action expert's Timestep Modules. Tests span LIBERO, ManiSkill, MetaWorld, and CALVIN, and module-replacement experiments show these modules account for a disproportionate share of RL gains. RL specializes the modules to discrete denoising timesteps; shift-vector update directions predict task success with ROC-AUC up to 99.6% and reflect cross-task transfer. Steering along those directions further improves policies without additional RL.

Hugging Face daily papersSep 27, 20:00 UTC · 11d agoAI research 2 sources

GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies

GeoAAC adaptively sizes action chunking horizons in flow-based VLA policies using denoising trajectory geometry, raising real-world manipulation success from 53.3% to 74.4%.

GeoAAC exploits geometric variation across action prefixes in flow-matching denoising trajectories as a process-level signal of prediction reliability. It constructs a horizon-wise geometric profile and adaptively determines the action horizon from a single generation without additional training. Experiments with GR00T N1.5 and π0.5 on LIBERO, LIBERO-Pro, and RoboCasa365 show gains up to 8.7 percentage points over fixed-horizon and adaptive baselines, with real-world success rate rising from 53.3% to 74.4%.

arXiv cs.AI / cs.LG / cs.CLSep 17, 17:48 UTC · 21d agoAI research1

GR00T N1.5

Timeline2 articles

Appears with

The Low-Rank Structure of VLA Reinforcement Learning

GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policies