Papers2

#Test-time compute scaling

Recurrent-Depth VLA: Implicit Test-Time Compute Scaling of Vision-Language-Action Models via Latent Iterative Reasoning

Yalcin Tur, Jalal Naghiyev et al.Feb 8arXiv

Robots often use the same amount of thinking for easy and hard moves, which wastes time on easy steps and isn’t enough for tricky ones.

#Recurrent depth#Latent iterative reasoning#Vision-Language-Action

PaCoRe: Learning to Scale Test-Time Compute with Parallel Coordinated Reasoning

Intermediate

Jingcheng Hu, Yinmin Zhang et al.Jan 9arXiv

PaCoRe is a way for AI to think in many parallel paths and then coordinate them, so it can use a lot more brainpower at test time without running out of context window space.

#Parallel Coordinated Reasoning#Test-time compute scaling#Message passing