In Physics
The process of mapping a continuous signal to a finite set of discrete values.
In AI & ML
Quantization reduces the precision of weights and activations by gathering statistics for each layer and using fewer bits for the integer part.
In Music & Film
Dividing the DCT coefficients by bin sizes that increase with frequency to discard high-frequency details the human eye cannot see well.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Classical field theory1Commutation relations1Data representation1Discrete cosine transform1Eigenvalue problems1Energy states1Floating point number1Floating point precision1
Opens up
Concepts that name this one as a prerequisite.
In Physics
6Each in their own terms, from the easiest level.
Alan Guth
The restriction of the energy of a harmonic oscillator to discrete values proportional to its frequency.Barton Zwiebach
The condition that certain physical quantities like energy can only take discrete specific values for normalizable solutions to exist.David Tong
The process of transitioning from a classical field theory to a quantum field theory by imposing commutation relations on the fields and their conjugate momenta.Arindam Kumar Chatterjee
The restriction of a physical property to discrete values rather than a continuous range.Onur Mutlu
Mapping a large set of input values to a smaller set of discrete values to reduce complexity and enable generalization.
In AI & ML
4Each in their own terms, from the easiest level.
Ashish Vaswani
Representing numbers using fewer bits to reduce memory bandwidth and speed up matrix multiplication.Nathan Lambert
Methods such as BitNet that reduce precision to make models more efficient.Onur Mutlu
The process of converting floating-point model weights and activations to lower-precision representations like 8-bit integers.
In Music & Film
1Marc Levoy
Lectures that assume it
11- QED Complete Four-Part Series: Today's Answers to Newton's Queries about Light (1979)
- Applications of Least-Squares Estimation: System Identification & Filtering
- A Hacker's Guide to Language Models
- Coupling Light and Matter
- A Geometer Explores the Universe: The Skyrme Model of the Nucleus
- Solving the Dirac Equation
- Generative Models: Conditional Models
- GSQ-RCO: Better Non-Uniform GGUF Quantization
- Emerging Threats in LLM Privacy and Poisoning
- Zeitlin's Model for Ideal Hydrodynamics on the Sphere
- Infinity Structures, BV Formalism, and Discrete Models of Continuum Vector Calculus
Connections
31The lectures that define it or assume it, and the concepts next to it.
Concepts8
- Classical field theory
- Commutation relations
- Discrete cosine transform
- Adaptive quantized hash-based matching
- Anomaly
- Binary / ternary net
Show 2 moreShow fewer
Lectures23
- Memory Controller Microarchitecture, Row Policies & Request Scheduling
- Arch4Health: 4th Workshop on Architecture for Health
- Autofocus Systems: Active, Contrast, and Phase Detection (CS 178, Lecture 7)
- Course Overview: Applied Linear Algebra & Linear Dynamical Systems
- Quantum Electrodynamics
- Black-Body Radiation and the Early History of the Universe, Part II
Show 17 moreShow fewer
- How I Learned to Stop Worrying and Love the Transformer
- Behavior of the Differential Equation (MIT 8.04 Quantum Physics I)
- Efficient Methods and Hardware for Deep Learning
- Advanced Quantum Mechanics: Lecture 1
- Digital Photography: Light Fields and Computational Cameras
- Aligning Open Language Models
- QED Complete Four-Part Series: Today's Answers to Newton's Queries about Light (1979)
- Applications of Least-Squares Estimation: System Identification & Filtering
- A Hacker's Guide to Language Models
- Coupling Light and Matter
- A Geometer Explores the Universe: The Skyrme Model of the Nucleus
- Solving the Dirac Equation
- Generative Models: Conditional Models
- GSQ-RCO: Better Non-Uniform GGUF Quantization
- Emerging Threats in LLM Privacy and Poisoning
- Zeitlin's Model for Ideal Hydrodynamics on the Sphere
- Infinity Structures, BV Formalism, and Discrete Models of Continuum Vector Calculus