Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore efficient deep learning computing techniques through this comprehensive MIT course that addresses the computational challenges of large generative models like language models and diffusion models. Master essential optimization methods including model compression, pruning, quantization, and neural architecture search to deploy powerful AI applications on resource-constrained devices. Delve into distributed training strategies, data and model parallelism, gradient compression, and on-device fine-tuning techniques that make large-scale models more accessible. Learn application-specific acceleration methods for large language models, diffusion models, video recognition, and point cloud processing, while gaining exposure to emerging quantum machine learning concepts. Engage with hands-on deployment exercises featuring large language models such as LLaMA 2 on laptop computers. Progress through 23 detailed lectures covering neural network fundamentals, advanced pruning and sparsity techniques, multi-part quantization methods, neural architecture search strategies, knowledge distillation, TinyML for microcontrollers, transformer architectures, vision transformers, GANs, distributed training methodologies, transfer learning, prompt engineering, quantum computing basics, and noise-robust quantum machine learning approaches.
