Techniques and patterns for writing prompts that get reliable results from language and image models.
55 terms
Socratic prompting guides a language model or a learner through a structured sequence of probing questions instead of asking for an immediate final answer.
Step-back prompting asks a language model to state the general principle behind a question first, then answer the specific question using that principle.
Tree of thoughts (ToT) is a prompting technique in which a language model explores, scores and prunes several reasoning branches before choosing an answer.
Visual prompting steers an AI model with points, boxes, masks, marks or example images instead of, or alongside, a written text instruction.
Zero-shot chain-of-thought is a prompting technique that adds a reasoning trigger to a question so a language model reasons step by step without examples.