Prompt Engineering Techniques - AWS BedRock

What is Prompt Engineering?

In simple terms add more context to the user request is Prompt Engineering. Adding more context will basic aid the LLMs to generate appropriate responses.

Let me illustrate by an example

Here I am using AWS Bedrock Claude 3 Haiku as an example.

• First I am adding a context that I am an experienced pilot and I wanted the LLM to explain me about the use of Flaps in Aircraft and you could see the response from the LLM that is more suited for an experienced pilot.

• Next I am adding a context to the LLM that I am a school student. The LLM have changed the response now that is more suited for the School student as below

Zero Shot Prompting

Here we provide the prompt without any specific examples and the response generated by LLMs are vry diverse.

Few Shot Prompting

Here we provide multiple examples for a request and example response such that the LLMs understand well about and produce expected results

Chain of Thought Prompting(COT)

COT prompting is basically used to breakdown a complex problem and try to produce the results step by step, basically it adds some thinking process to the LLMs.