31 terms
Fundamentals
The core ideas behind every AI system: models, data, tokens, and how learning works.
- action spacenounthe complete set of actions an AI system is able or permitted to choose in an environment, such as moving a robot, calling a…
- algorithmnouna set of step-by-step instructions for solving a problem or completing a task
- artificial intelligencenountrendingthe field of building computer systems that perform tasks normally thought to require human intelligence, such as understanding…
- attentionnouna technique that lets a model weigh how relevant every part of the input is to every other part, so it can connect related words…
- backpropagationnounthe method used to train neural networks, which works backward from an error to calculate how much each weight contributed to it…
- computer visionnounthe field of AI that enables computers to interpret images and video, for example recognizing objects, faces, text, or movement
- datasetnounan organized collection of data used to train, test, or evaluate a model
- deep learningnouna type of machine learning that uses neural networks with many layers, allowing a system to learn complex patterns directly from…
- embeddingnouna list of numbers (a vector) that represents the meaning of a piece of text, an image, or other data, so that similar things end…
- generalizationnouna trained model's ability to perform correctly on new data, situations, or tasks that it did not encounter during training rather…
- gradient descentnounan optimization method that improves a model step by step by moving its parameters a little in whichever direction reduces the…
- inferencenountrendingthe act of running a trained model to produce an output, such as answering a prompt or generating an image
- latent spacenounthe hidden, many-dimensional space of numbers in which a model represents concepts, where nearby points have similar meaning
- loss functionnouna formula that measures how wrong a model's output is during training; training tries to make this number as small as possible
- machine learningnouna branch of AI in which computers learn patterns from data instead of following rules written by hand, and use those patterns to…
- modelnounthe trained program that results from machine learning: a set of learned numbers plus the structure that uses them to turn inputs…
- narrow AInounAI that is designed for, and good at, one specific task or a small set of related tasks, without general understanding outside…
- natural language processingnounthe field of AI concerned with getting computers to understand, generate, and work with human language
- neural networknouna computing system made of many simple connected units ("neurons") arranged in layers, which learns by adjusting the strength of…
- overfittingnouna problem in which a model learns its training examples too closely, including their noise and quirks, and so performs poorly on…
- parameternounone of the numbers inside a model that is adjusted during training; the total count of parameters is a rough measure of a model's…
- perplexitynouna measure of how surprised a probabilistic language model is by a sequence of tokens, commonly calculated from its average…
- proprietary datanounnon-public or access-controlled data that an organization or individual can use to train, tune, evaluate, or ground an AI system…
- tokennountrendingthe basic unit of text a language model reads and writes: a whole word, part of a word, a punctuation mark, or a space
- token budgetnountrendingan allowance or limit on the number of tokens an AI system may process or generate during a response, reasoning process, agent…
- tokenizernounthe component that splits text into tokens before it goes into a model, and joins tokens back into text afterward
- trainingnounthe process of feeding data to a model and repeatedly adjusting its parameters so its outputs get better
- training datanounthe collection of examples, observations, labels, actions, or feedback that a model learns from, including text, images, code…
- transformernountrendinga neural network design that processes a whole sequence at once and uses attention to work out which parts relate to each other…
- vectornounan ordered list of numbers; in AI, the form in which models represent words, images, and other data internally
- weightsplural nounthe learned numbers stored in a trained neural network that determine how strongly each part influences the next; the weights…
Advertisement