Weight
A weight is a numerical quantity assigned to an element to represent its contribution, influence, importance, strength, or resource amount in a calculation or system.
Data Without Labels
A weight is a learned numerical parameter that scales the contribution of an input or intermediate activation in a neural network.
- How it is used
- Uses weights to control each connection’s contribution to a neuron and to parameterize the transformations learned by network layers.
- How deep it goes
- Defines weighted inputs and a threshold perceptron, then explains iterative weight updates through loss gradients and backpropagation.
- Examples
- Illustrates weights with a two-input threshold perceptron and a face-identification network trained on images and names.
- What you can do after
- Evaluate/understand how changing network weights changes predictions and how training updates them to reduce loss.
Fabulous Adventures in Data Structures and Algorithms
An integer weight is a whole-number quantity representing the relative contribution of an outcome to a weighted distribution.
- How it is used
- Represents relative sampling probabilities in finite discrete distributions and carries those ratios through projections, joint distributions, and posterior inference.
- How deep it goes
- Defines normalization by total weight, implements uniform and projected weights, and works through LCM scaling for integer joint weights.
- Examples
- Uses a uniform integer distribution, a color-projected die, and cold-and-sneezing prior, joint, and posterior weights.
- What you can do after
- Build it from scratch: implement weighted discrete distributions and preserve integer weights when projecting outcomes or constructing joint distributions.
Read how Fabulous Adventures in Data Structures and Algorithms explains weight →
Introduction to Generative AI, Second Edition
A model weight is a learned numerical parameter that determines the strength of a computational connection in a machine-learning model.
- How it is used
- Defines the connection strengths that neural networks adjust during training.
- How deep it goes
- Explains weights as connection strengths, their iterative adjustment, and their role in parameter counts, but defers the underlying mathematics.
- Examples
- Uses the leak of Meta’s Llama 1 weights and rapid reproduction of a model based on it as a security example.
- What you can do after
- Understand how training sets connection strengths and why access to model weights can enable model reproduction.
Read how Introduction to Generative AI, Second Edition explains weight →
Kubernetes in Action, Second Edition
A weight is a routing value that determines a backend's proportional share of requests relative to the weights of other backends.
- How it is used
- Routes a proportional share of HTTP requests to each backend, including a small share to a canary Service.
- How deep it goes
- Explains proportional shares with a 9:1 worked example and a request-count check, without detailing routing internals.
- Examples
- Uses stable and new Services with weights of 9 and 1, and suggests checking request counts with curl.
- What you can do after
- Configure an HTTPRoute to send a chosen proportional share of requests to multiple backend Services.
Read how Kubernetes in Action, Second Edition explains weight →
Related concepts
- fine-tuningused by · Fine-tuning is adapting a pretrained machine-learning model by training it further on targeted data or objectives for a task, domain, or behavior.
- deep learningpart of · Deep learning is a machine-learning approach that uses neural networks with multiple computational layers to learn representations from data.
- neural networkpart of · A neural network is a computational model of interconnected, parameterized units that learns transformations from inputs to outputs.
- parameterequivalent to · A model parameter is a numerical value, such as a weight or bias, that is learned or set to determine how a model computes outputs.
- linear regressionused by · Linear regression is a statistical modeling method that estimates a continuous response as a linear combination of predictor variables and fitted coefficients.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.