A toolkit to optimize ML models for deployment for Keras and TensorFlow, including quantization and pruning.
A Systems View of Efficient Machine Learning, from Silicon to Agents. Fourteen parts, from roofline analysis and vector units up through kernels, compilers, quantisation, compression, vision, on-device language models, robotics, profiling, serving and agents.