TinyML Talks - Demoing the World’s Fastest Inference Engine for Arm Cortex-M
Explore a live demonstration of the world's fastest inference engine for Arm Cortex-M microcontrollers. Discover how Plumerai's engine outperforms TensorFlow Lite for Microcontrollers with CMSIS-NN kernels, achieving 40% lower latency and 49% less RAM usage without compromising accuracy. Learn about the techniques used to achieve these speedups and memory improvements, including optimized memory planning and INT8 code generation. Examine benchmarks for popular neural network models and gain insights into the MLPerf Tiny models. Understand the machine learning flow and the tasks of an inference engine, with a focus on the TFLM example. Get introduced to Plumerai's public benchmarking service, allowing you to test your own models with their cutting-edge inference engine.