…NAVER AI Lab Authors: , , , , , Abstract Retrieval-augmented vision-language-action policies eliminate per-task fine-tuning costs by using pre-trained models with indexed demonstrations, enabling efficient cross-embodiment generalization and task…
…ELBO-based surrogates with Group Relative Policy Optimization, achieving faster and more efficient alignment of generative models with human preferences compared to existing approaches. AI-generated summary Aligning denoising generative models with…
…Bo Yin , Qi Li , Abstract FATE is an on-policy framework that uses failure trajectories to improve agent safety and performance through self-evolution and Pareto-aware optimization. AI-generated summary Tool…
…improvement of Vision-Language-Action policies through fleet-scale offline-to-online reinforcement learning with distributed robot experience and human interventions. AI-generated summary Generalist robot policies increasingly benefit from large-scale…
…To study this setting, we construct two human-verified benchmarks, VRQABench for controllable spatial lookahead and OpenWorldQA for open-domain physical prediction, and propose Privileged-Future On-Policy Self-Distillation ( PF-OPSD…
…AI-generated summary We investigate the temporal concatenation of sub-policies in Markov Decision Processes (MDP) with time-varying reward functions . We introduce General Dijkstra Search (GDS), and prove that globally optimal…
…AURA-Mem matches the ungated base policy (0.233) and slightly exceeds an always-write KV arm (0.217), while using 7.0 times fewer writes and constant memory. We also instantiate…
…policy side, we present LabVLA , trained with a two-stage recipe : FAST action token pretraining first makes the Qwen3-VL-4B-Instruct backbone action aware before any continuous control is learned, and…
…they sample k responses, build a target distribution via a temperature-controlled softmax of rewards, and then project the current policy onto that target using a chosen divergence, which gives closed-form…
…following fidelity through real-world data retraining and incorporates modules for state decoding, world scoring, and accelerated inference to enable scalable policy learning. AI-generated summary We introduce GE-Sim 2.0…