Publications

(0001). Model-agnostic policy explanations with large language models. COLM 2025 [top 5%].

(0001). CAPEX: Efficiently Distilling Foundation Model Behavior into Deployable Robot Policies through Experience-Adaptive Reasoning. arXiv preprint, Sep 2026.

(0001). FlowEval: How Much Adaptation Does an Embodied Foundation Model Need at Deployment?. arXiv preprint, Oct 2026.