Massachusetts Institute of Technology 6.5940 EfficientML.ai
Professor Song Han · Massachusetts Institute of Technology · 8 lectures with notes
Students in this class: ask your lecturer for the class code, and these lectures will already be in your library when you sign up.
EfficientML.ai Lecture 7 - Neural Architecture Search (Part I) (MIT 6.5940 Fall 2026)
NAS automates the search for accurate, efficient models by designing suitable search spaces and evaluating candidate architectures against resource constraints.
EfficientML.ai Lecture 6 - Quantization (Part II) (MIT 6.5940 Fall 2026)
Choosing quantization granularity, calibration, and precision per layer determines whether low-bit models retain accuracy and run efficiently.
EfficientML.ai Lecture 5 - Quantization (Part I) (MIT 6.5940 Fall 2026)
How K-means and affine quantization shrink neural networks and enable integer-only inference.
EfficientML.ai Lecture 4 - Pruning and Sparsity (Part II) (MIT 6.5940 Fall 2026)
Effective pruning requires layer-aware sparsity, gradual fine-tuning, and execution paths designed to exploit zeros.
EfficientML.ai Lecture 3 - Pruning and Sparsity (Part I) (MIT 6.5940 Fall 2026)
Pruning can sharply reduce model size, but useful speedups depend on matching sparsity patterns to hardware.
EfficientML.ai Lecture 2 - Basics of Neural Networks (MIT 6.5940 Fall 2026)
A practical guide to neural-network building blocks and measuring their compute, memory, latency, and energy costs.
EfficientML.ai Lecture 1 - Introduction (MIT 6.5940 Fall 2026)
Song Han’s course shows how compression and systems techniques can make large AI models practical on laptops, phones, and microcontrollers.
EfficientML.ai Lecture 8 - Neural Architecture Search (Part II) (MIT 6.5940 Fall 2026)
Hardware-aware NAS finds efficient, platform-specific models through weight sharing, direct latency feedback, and joint hardware–network search.