- All Courses
- Building LLM Applications
- Talking to a Model
- Prompting fundamentals
Lesson 1.2
Prompting fundamentals
Vague prompts produce vague output.
4mIntermediate15.1k students
Overview
A prompt is a specification
Vague prompts produce vague output. State what the task is, what the output should look like, and what the model must not do. Most disappointing results are underspecified requests rather than model limitations.
When a rule is hard to describe, show it. Two or three examples of an edge case handled correctly communicate more than a paragraph of instruction.
Iterate against a fixed set of inputs. Changing the prompt and the test case at the same time means you learn nothing from the result.
In this lesson you will:
- Specify the task, the format, and the constraints
- Use examples where description is ambiguous
- Iterate against fixed inputs, not fresh ones
Resources
Notes are not saved yet — they clear when you leave this page.
Previous Lesson
How chat completions work
8m
Structured output and JSON schemas
31m