The words you will encounter.
- AI: software performing tasks associated with human intelligence
- MACHINE LEARNING: systems learning patterns from data
- DEEP LEARNING: machine learning using layered neural networks
- NEURAL NETWORK: layers of connected units, loosely inspired by neurons
- MODEL: the trained system that turns input into output
- TRAINING: adjusting a model using example data
- PARAMETER: one adjustable value inside a model
- LLM: large language model, trained to predict and generate text
- TRANSFORMER: the architecture behind most modern language models
- TOKEN: a fragment of text, the unit models process
- CONTEXT WINDOW: how much text a model can consider at once
- PROMPT: the input you give a model
- HALLUCINATION: confident, fluent, false output
- FINE-TUNING: further training a model on specific data
- RAG: retrieval augmented generation, answering from retrieved documents
- EMBEDDING: text represented as numbers capturing meaning
- TEMPERATURE: a setting controlling randomness in output
- INFERENCE: running a trained model to produce output
- AGENT: a system taking actions over multiple steps
- MULTIMODAL: handling text, images and other data together
- ALIGNMENT: making a model behave as intended
