OLIVIA is proposed, a lightweight online adaptation layer that replaces the frozen action-selection head with a per-action contextual linear bandit, selects actions via LinUCB exploration on the LLM's hidden state and updates only its per-action sufficient statistics from step-level feedback via a rank-one Sherman–Morrison rule.
This work forms personalization of a frozen agent as online learning of a per-user execution policy from scalar feedback observed only for the executed action, and proposes FABLE (Factorized Adaptive Bandit Layer for Execution), a lightweight policy layer outside a potentially black-box host agent.
Dian Jin, Zhi Zhang, Huichao Li et al.· 0 citations
This work extends contrastive reinforcement learning (CRL), a prototypical self-supervised method, to operate over action chunks, and finds that this results in large, pervasive gains across established offline and online benchmarks: +31.7% and +93.1% across 18 and 11 environments respectively.
Michal Korniak, Kamil Dybek, Benjamin Eysenbach et al.· 3 citations· ⚡1
SPACE induces two-level programmatic skills from successful trajectories, where subskill boundaries serve as direct chunk-boundary supervision from trajectory-induced programmatic skills, and distilled into a primitive-chunk policy via hybrid on-/off-policy optimization.
Yan-Ting Yang, Can Jin, Jin-Man Zhao et al.· 2 citations
TEMPO is proposed, a semantic-action decoupled, two-timescale RL post-training framework for VLA models that consistently outperforms both pretrained state-of-the-art VLA models and the RL post-training baseline, while reaching and maintaining higher evaluation rewards on two real-world tasks.
A lightweight LoRA parameterization that conditions adaptation at each action-chunk prediction step using two weakly supervised descriptors: fine-control tendency and event/boundary intensity is proposed, establishing within-trajectory conditioning as an effective lightweight PEFT axis for continuous-action VLA policie...
Yu-Fei Guo, Yi-Nan Wu, Hao-Ran Duan et al.· 0 citations
This work proposes TAPO: Transition-Aware Policy Optimization for LLM Agents, a unified training framework that alternates between policy optimization and transition supervision, and demonstrates that TAPO consistently improves task performance over pure policy optimization baselines.
