Zero-Shot
A model performing a task without any examples of that task in the prompt.
In plain language
Zero-shot relies on task patterns and instruction-following learned during pretraining. It usually works for common tasks such as summarization, classification and rewriting, but is unstable for strict formats, heavy domain terminology or ambiguous judgment criteria. The key is not withholding information but specifying the goal, output format and criteria. Complex tasks should move to few-shot or fine-tuning.
Example
“Summarize this meeting into three action items” is zero-shot; adding two examples of what counts as an action item makes it few-shot.
Recommended tools
Related terms
Few-Shot
Providing a small number of input-output examples in the prompt so the model can imitate the pattern.
In-Context Learning
A model learning a task temporarily from instructions and examples in the prompt without updating its weights.
Prompt Engineering
Designing instructions, examples and constraints so a model reliably produces the desired output.
Agent Memory
A mechanism for an agent to store and retrieve user preferences, past tasks and key facts across sessions.