Richard Richard, inspired by a childhood programming project, has transformed into a powerful neural network tool. The original mischievous “virus” Richard has evolved into a benevolent CLI application specializing in classification using dense, convolutional, and max pooling layer types with GPU acceleration through Vulkan compute shaders. With support for hand-written digits and cat vs. dog classification, Richard is a versatile and user-friendly platform for machine learning experiments. Users can obtain detailed profiling insights by leveraging Google Perftools for CPU profiling, offering a comprehensive look at performance metrics and how time is allocated within the system.
https://github.com/robjinman/richard