Context window
Level 2
The maximum amount of text a model can consider at one time.
Term 4 of 8 in Language models
In plain language
Everything the model uses to answer must fit inside its context window: your question, the conversation so far, and any documents attached. Once the window is full, the oldest material is pushed out.
Think of it like this
A desk of a fixed size. New pages can always come on, but something has to come off.
Why it matters
This is why long chats start to forget things, and why a huge document sometimes has to be summarised before the model can use it.
Next terms in Language models
- Large language modelA model trained on enormous amounts of text to predict what words come next.
- TokenA small chunk of text that a model reads and writes, usually part of a word.
- PromptThe instruction and information you give a model.
- TemperatureA setting that controls how predictable or varied a model's output is.