Towards a green communication discipline, more energy-efficient wireless networks are needed. In this paper, an energy-efficient interference management technique is proposed for heterogeneous networks (HetNets) with downlink nonorthogonal multiple access (NOMA) operating in millimeter (mm) waves. The proposed technique utilizes data-driven-based power and resource blocks (RBs) allocation scheme to maximize the energy efficiency of the entire network while managing inter-user, co-tier, and cross-tier interference. The studied interference management problem is mathematically formulated as maximizing the energy efficiency of NOMA-HetNets under constraints; quality of service (QoS), successive interference cancelation condition (SIC), maximum power allocation, and the maximum capacity of each RB, which is a non-convex and combinatorial NP-hard problem that needs exhaustive search for an optimum solution. The formulated problem is relaxed into a minimization of a linear <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">l</i> <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> -norm problem. Then, a lowcomplex model-free data-driven based approach is proposed to find a near-optimum allocation for power and RBs based on a dynamic linearization representation of a time-varying pseudogradient parameter estimation procedure. Numerical analyses show that the proposed scheme achieves higher network energy efficiency with lower computational complexity compared to the benchmark schemes.