FluidML: Fast and Memory Efficient Inference Optimization

Machine learning models deployed on edge devices have enabled numerous exciting new applications, such as humanoid robots, AR glasses, and autonomous vehicles. However, the computing resources available on these edge devices are not catching up with the ever-growing number of parameters in these models. As the models become bigger and more complicated, the novel yet sophisticated structure challenges the inference runtime optimization. ML, a generic runtime memory management and optimization framework that can flexibly transform the model execution blueprint to achieve faster and more memory-efficient inference. ML can consistently reduce the end-to-end inference latency by up to 25.38% for popular language models and reduce peak memory usage by up to 41.47%, compared to state-of-the-art approaches. ML is of 30K line of codes, built for general-purpose usage, and will be released as an open-source inference runtime optimization framework to the community. Fundamentally, there is a lack of a generic, Taking advantage of near-sensor inference, machine learning model-agnostic framework that can provide a holistic plan (ML) models deployed on edge devices enabled many for how the numeric computation should flow throughout low-latency, low-power, and privacy-sensitive applications.