Prompt Engineering Workshop

A comprehensive interactive workshop for learning prompt engineering techniques with large language models (LLMs). This repo contains materials from our Prompt Engineering workshop @ AI Convention 2025 (IHK Schwaben) including practical examples, hands-on exercises, and real-world applications of advanced prompting strategies.

We also provide the complete presentation materials here:

📄 Download Workshop Presentation PDF

This workshop covers essential prompt engineering techniques from basic prompting to advanced frameworks like ReAct and Tree of Thoughts. An interactive Streamlit application provides hands-on experience with real LLMs models, while comprehensive Jupyter notebooks offer detailed explanations and examples.

The following details the functionality provided by the workshop:

• 8 Core Techniques: From basic prompting to advanced reasoning frameworks

  • Basic Prompting - Simple text completions and natural language interactions
  • Instruction-based Prompting - Explicit task directions and structured guidance
  • Zero/One/Few-Shot Learning - Learning from examples with varying amounts of context
  • Chain-of-Thought Reasoning - Step-by-step logical problem solving
  • Self-Consistency - Multiple solution paths with consensus voting
  • Tree of Thoughts - Systematic exploration of solution spaces
  • ReAct Framework - Combining reasoning with external tool interaction
  • Real-world Applications - Practical implementations and use cases

• Interactive Learning Environment: Web-based interface for hands-on experimentation

• Comprehensive Examples: Over 50 practical examples across different domains

• Model Flexibility: Support for OpenAI GPT and Anthropic Claude models

• Educational Materials: Detailed notebooks with theory and implementation

The workshop is compatible with Python 3.8+ and runs on Linux, MacOS X and Windows. It is distributed under the Apache 2.0 license.

📖 Contents

• Installation
• Quick Start
• Workshop Techniques
• Configuration
• Contribute
• License

⚙️ Installation

There are two ways to set up the Prompt Engineering Workshop:

• Clone from GitHub (recommended):

git clone https://github.com/alexanderlammers/prompt_engineering_guide.git
cd prompt_engineering_guide
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

• Direct download:

# Download and extract the workshop materials
# Install dependencies
pip install streamlit openai anthropic python-dotenv pandas numpy

API Keys Setup

Create a .env file in the project root:

OPENAI_API_KEY=your_openai_api_key_here
ANTHROPIC_API_KEY=your_anthropic_api_key_here

🚀 Quick Start

Interactive Web Application

Launch the interactive Streamlit workshop:

streamlit run prompt_engineering_workshop_app.py

This opens a web interface where you can:

• Experiment with different prompting techniques
• Adjust model parameters in real-time
• See immediate results from LLM models
• Follow guided tutorials for each technique

Jupyter Notebooks

For detailed learning, explore the notebooks:

jupyter notebook notebooks/

Start with 0_Environmental_Setup.ipynb to configure your environment.

Basic Example

Here's a simple example of chain-of-thought reasoning:

from modules.prompt_engineering_workshop import PromptEngineering

# Initialize the prompt engineering framework
pe = PromptEngineering()

# Chain-of-thought prompting example
problem = """
A library has 1,200 books on the first floor, 40% more books on the second floor, 
and 25% fewer books on the third floor than the second floor. How many books total?
"""

prompt = f"Problem: {problem}\n\nLet's solve this step-by-step:"
messages = [{"role": "user", "content": prompt}]

response = pe.get_completion(messages, temperature=0.1)
print(response)

Advanced Framework Example

Using the ReAct framework with tools:

from modules.react_framework import ReActAgent, CalculatorTool, WikipediaTool

# Initialize tools and agent
tools = [CalculatorTool(), WikipediaTool()]
agent = ReActAgent(pe, tools)

# Solve a complex problem requiring multiple tools
question = """
What's the area of a circle with radius 15 meters, and how does this 
compare to the area of Tokyo Disneyland?
"""

result = agent.solve_with_react(question)
print(result)

🎯 Workshop Techniques

1. Basic Prompting

Simple, natural language inputs that leverage the model's pretrained knowledge.

prompt = "Explain quantum computing in simple terms"

2. Instruction-based Prompting

Explicit task directions with format specifications.

prompt = """
Task: Classify the sentiment of the following text as positive, negative, or neutral.
Text: "This workshop is incredibly helpful for learning AI!"
Format: Output only the classification.
"""

3. Few-Shot Learning

Learning from examples with varying amounts of context.

prompt = """
Examples:
Text: "Amazing product!" → Positive
Text: "Terrible service." → Negative
Text: "It's okay." → Neutral

Text: "This workshop exceeded my expectations!" → 
"""

4. Chain-of-Thought Reasoning

Step-by-step problem solving with explicit reasoning.

prompt = """
Problem: A company's revenue increased by 15% in Q1, then decreased by 8% in Q2. 
If they started with $1M, what's their Q2 revenue?

Let's solve this step-by-step:
"""

5. Self-Consistency

Multiple solution paths with consensus voting.

# Generate multiple solutions and find consensus
solutions = []
for i in range(5):
    response = pe.get_completion(messages, temperature=0.7)
    solutions.append(response)

consensus = find_consensus(solutions)

6. Tree of Thoughts

Systematic exploration of multiple solution branches.

from modules.tree_of_thoughts import TreeOfThoughts

tot = TreeOfThoughts(pe, max_depth=3, breadth=3)
result = tot.solve_with_tree_of_thoughts(complex_problem)

7. ReAct Framework

Reasoning combined with external tool interaction.

# Combines thinking, acting, and observing in iterative cycles
# Thought → Action → Observation → Thought → ...

⚙️ Configuration

Model Parameters

Control the LLM model behavior through various parameters:

# Temperature: Controls randomness (0.0 = deterministic, 1.0 = creative)
temperature = 0.7

# Top P: Nucleus sampling (0.5 = conservative, 1.0 = full vocabulary)
top_p = 0.9

# Max Tokens: Response length limit
max_tokens = 500

# Model Selection
model = "gpt-4"  # or "gpt-3.5-turbo", "claude-3-sonnet", etc.

🤝 Contribute

We welcome contributions to improve the workshop:

• Add Examples: Contribute new prompting examples
• Improve Documentation: Enhance explanations and tutorials
• Fix Issues: Report and fix bugs
• New Techniques: Add emerging prompting strategies

See the contribution guidelines for more details.

🏗 Maintainers

• Alexander Lammers, DATANOMIQ GmbH - GitHub

© Copyright

Copyright 2025 DATANOMIQ GmbH. All rights reserved.

See LICENSE for details.

---

Built for AI Convention 2025 (IHK Schwaben) - Empowering developers with advanced prompt engineering techniques.