Learn Before
Concept

Problems with Natural Language Prompts

While natural language prompts are a common and direct method for interacting with LLMs, they present two main issues. First, they can be overly complex and long, which creates a significant computational load and makes them inefficient, especially for repetitive tasks. Second, these prompts are represented as discrete token sequences (hard prompts), but LLMs internally convert them into low-dimensional vectors, which raises questions about whether more direct and efficient prompt representations are possible.

0

1

Updated 2026-05-02

Contributors are:

Who are from:

Tags

Ch.3 Prompting - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Related