Kestutis Saltonas developed a feature for the openvinotoolkit/openvino repository that generalizes CWAI3 pattern matching to improve inference performance. By expanding the handling of weight scales as constants, Kestutis enabled more flexible recognition across various data types and operations, incorporating optional slicing and conversion steps. This C++ implementation focused on graph optimization and pattern matching, resulting in an observed reduction of approximately 240 milliseconds per inference across representative workloads. The work demonstrated depth in performance optimization, addressing both generalization and efficiency, while also identifying a regression in certain operations that is being tracked for further resolution within the project.