Review for NeurIPS paper: Agnostic Q -learning with Function Approximation in Deterministic Systems: Near-Optimal Bounds on Approximation Error and Sample Complexity