Glossary
The vocabulary
Every term gets a plain sentence first and the technical explanation second. If a word in a lesson lost you, it is here.
24 of 24 terms
- AgentsAI agentA system that pursues a goal by planning, choosing tools and taking actions — not just replying.
- EngineeringAPIA defined way for one program to request something from another.
- FoundationsArtificial intelligenceSoftware that performs tasks we would normally say require human intelligence.
- ModelsContext windowThe maximum amount of text, measured in tokens, a model can consider at once.
- DataDatasetThe collection of examples a model is trained on.
- FoundationsDeep learningMachine learning built on neural networks with many layers stacked between input and output.
- DataEmbeddingA list of numbers representing a piece of text, arranged so similar meanings sit close together.
- PromptingFew-shot promptingIncluding a small number of worked examples in the prompt to show the model the pattern you want.
- EngineeringFine-tuningContinuing to train an existing model on a smaller, specific dataset to specialise its behaviour.
- SafetyHallucinationOutput that is fluent and confident but factually wrong or entirely invented.
- EngineeringInferenceRunning a trained model to produce an output — as opposed to training it.
- ModelsLarge language modelA model that predicts the next token of text, trained on very large amounts of written material.
- FoundationsMachine learningAn approach where a system derives its own rules from examples instead of being given rules by a person.
- FoundationsNeural networkA large stack of simple mathematical operations whose adjustable values are tuned during training.
- SafetyPrompt injectionAn attack where instructions hidden in content the model reads are followed as if they came from the user.
- DataRetrieval-augmented generationFetching relevant documents and putting them in the context so the model answers from real sources.
- PromptingSystem promptInstructions placed before the conversation that set a model's role, rules and constraints.
- PromptingTemperatureA setting controlling how much randomness is used when picking each next token.
- ModelsTokenThe unit of text a model actually processes — usually a word, part of a word, or punctuation.
- ModelsTokenisationThe process of cutting text into tokens before a model reads it.
- AgentsTool callingThe mechanism by which a model requests that an external function be run on its behalf.
- EngineeringTrainingThe process of adjusting a model's weights so its outputs match the examples it is shown.
- ModelsTransformerThe neural network architecture behind modern language models.
- DataVector databaseA database built to store embeddings and find the nearest ones to a query quickly.