Policy Regret for Embedding Model Routing: Contextual Bandits with Low-Rank Experts

In three linesTheoretical paper on dynamic routing of queries to multiple embedding models. Formalizes the problem as an adversarial contextual linear bandit with low-rank experts. Proposes Hypentropy Policy Gradient (HPG) algorithm achieving Õ(s√MT) linearized policy regret without curse of dimensionality.

Summary generated by Claude — human-verified