Zero-Shot Learning
ゼロショット学習(ゼロショットがくしゅう)
IntermediateModels & Architecture
An AI model's ability to perform a task it was never explicitly trained on, using only its general knowledge.
Why It Matters
Zero-shot capability means you don't need custom training data for every new task — the model generalizes.
Example in Practice
Asking GPT-4 to classify restaurant reviews as positive/negative without providing any examples.