Fetching the paper…

Optimal Contextual Bandits with Knapsacks under Realizability via Regression Oracles · Around