In this work, we describe a single-loop method, with no back-tracking or line searches, which does not require knowledge of $D$ yet asymptotically achieves the optimal rate of convergence for the complexity class of convex Lipschitz functions. Code: https://github.com/facebookresearch/dadaptation