Prompt engineering

Last updated: Saturday, 29 June 2024

Crafting prompts or instructions given to generative language models and other AI systems to elicit desired outputs.

An iterative process, that requires a certain level of computational thinking?

Think about in-context learning, chain-of-thought prompting, one-shot vs. multi-shot prompts, etc.

• [?] What are some best practices for balancing specificity and flexibility in prompts, to avoid overfitting while eliciting relevant responses?
• [&] See also: dialogue systems, retrieval-augmented generation (rag)?

Backlinks

: Oulipics