ValueFormer: A Causal Transformer Value Function with Stage-Aware Labels for Semi-Autonomous Vision-Language-Action Policies
ValueFormer is presented, a compact policy-agnostic causal transformer over a frozen DINOv3 backbone that emits two per-frame signals in one forward pass: a smooth Monte Carlo value, V_mc, for advantage estimation and a sharp binary value for online mistake detection, targets that pull in opposite directions by design.
Inkyu Sa, Konstantin Stulov, Rajat Bhageria
· 1 citation