This study presents an automated high-accuracy optimization approach for designing high-performance radio frequency high power amplifiers (HPAs). The amplifier is designed by applying a top-down pruning optimization approach that automatically converts the given HPA with lumped elements (LEs) to the HPA with distributed elements (DEs). Firstly, the lumped element HPA is designed based on a bottom-up optimization presented in Kouhalvandi et al. (2019 11th international conference on electrical and electronics engineering (ELECO), pp 510-513, 2019, 10.23919/ELECO47770.2019.8990407), then the LE amplifier is decomposed into basic unit cells that consist of one capacitor (C) and one inductor (L). For each LC unit, a suitable transmission line cell network is selected from predefined models by considering the maximum a posterior (MAP) metric. The component values of the resulting HPA design with DEs are optimized using Bayesian Optimization to achieve the desired design specifications. The overall proposed automated optimization accelerates the design process and outperforms the amplifier's specifications that is constructed with DEs, automatically. The optimization starts with LE amplifier designs for keeping high linear gain performance and is converted to the HPA with transmission lines for having ready to fabricate circuit design. The proposed approach is validated by designing three HPAs with GaN HEMT from 1.8 to 2.2 GHz operational band frequency with drain efficiency more than 50% and with minimum linear power gain of 14.5 dB in all band frequency.