Prompt Engineering

Prompt Engineering#

Part 1#

LLMs#

  • Large Language Models are supervised models

  • The objective is to predict the next word in a sentence

Generating text#

In order to generate new text, the model needs a starting point

  • Provide some initial text for the LLM to complete

  • We call this a prompt

How prompts work#

First, a little game 😄#

https://gandalf.lakera.ai/baseline

Prompt structure#

A typical LLM prompt structure consists of the following messages:

  • System message:

    • Unique, applies to all instructions on the model

    • Is defined only once based on our requirements

  • User message:

    • Has the question that needs to be answered

    • Expecting a new message each time we run inference

  • (Optional) Assistant message:

    • Includes the expected response from the LLM

    • Usually used for training/fine-tuning models only

Prompt Elements#

A prompt contains any of the following elements:

  • Instruction - a specific task or instruction you want the model to perform

  • Context - external information or additional context that can steer the model to better responses

  • Input Data - the input or question that we are interested to find a response for

  • Output Indicator - the type or format of the output.

Example:

Classify the text into neutral, negative, or positive
Text: I think the food was okay.
Sentiment:

Prompt Templates#

Zero-shot Prompting#

Prompt:

Classify the text into neutral, negative or positive. 
Text: I think the vacation is okay.
Sentiment:

Output:

Neutral

Few-shot Prompting#

Prompt:

Classify the text into neutral, negative or positive. 
Follow these examples:

Example 1: 
Text: I think the movie is very bad
Sentiment: Negative

Example 2:
Text: Today was a good day!
Sentiment: Positive

Example 3:
Text: Today we learnt about Machine Learning
Sentiment: Neutral

Now classify the following text:

Text: I think the vacation is okay.
Sentiment:

Output:

Neutral

Chain-of-Thought (CoT) Prompting#

References#

Contents