The ability of a model to perform a task or classify an object it was never explicitly trained on, relying on generalised knowledge and semantic relationships.
Zero-shot learning is the ability of an AI model to correctly perform a task on classes or instructions it has never seen during training, by leveraging generalised knowledge about related concepts or following natural language instructions directly.
In the context of LLMs, zero-shot prompting means asking a model to complete a task without providing any examples in the prompt — the model generalises from its pre-training knowledge alone.
🔍 Click image to zoom
Zero-shot vs few-shot prompting
Share
Frequently Asked Questions
What is Zero-shot Learning?
The ability of a model to perform a task or classify an object it was never explicitly trained on, relying on generalised knowledge and semantic relationships. Zero-shot learning is the ability of an AI model to correctly perform a task on classes or instructions it has never seen during training, by leveraging generalised knowledge about related concepts or following natural language instructions directly.
How is Zero-shot Learning used in practice?
In the context of LLMs, zero-shot prompting means asking a model to complete a task without providing any examples in the prompt — the model generalises from its pre-training knowledge alone.
Why is Zero-shot Learning important in AI?
Zero-shot Learning is a foundational concept in Learning Paradigm. The ability of a model to perform a task or classify an object it was never explicitly trained on, relying on generalised knowledge and semantic relationships.