零样本学习

Asking the model to perform an unseen task from the description alone, without examples.

Zero-shot relies on generalization learned during pretraining, great for quick validation. The harder the task, the more likely mistakes—switching to few-shot usually improves stability.