Artificial Intelligence
Course Overview
Conquer Artificial Intelligence (AI) frontier and realize the power of generative AI and complex large language models. Build a strong foundation with AI to propel you into the new AI-powered landscape.
The AI/LLM track is designed to introduce participants to the world of Large Language Models (LLMs) and their applications. LLMs are revolutionizing how we interact with data, offering capabilities like natural language understanding, text generation, and more. This track will cover theoretical concepts, practical exercises, and real-world use cases to equip you with the knowledge and skills to leverage LLMs effectively.
This course requires no prior coding experience but python code knowledge can enable you to create more advanced projects.
Instructions
Following the kick-off interview all participants receive an invitation to Datacamp to access the recommended course links below. In any case, the track courses and labs do not require any specific software installation on the participants laptop, everything can be completed on-line.
For this track, you will need access to various online learning platforms and tools such as ChatGPT, Gemini, Copilot, and HuggingChat. Please ensure you have accounts set up on at least 2 of these platforms.
Materials required include:
- A computer with internet access.
- Access to DataCamp, YouTube, and other specified resources.
- Accounts on at least 2 of the following:
Materials
Week 1
This week focuses on the fundamentals of Large Language Models (LLMs). You’ll learn about what LLMs are, their development, and the underlying technologies that make them powerful. Topics include Deep Neural Networks, Attention mechanisms, the importance of scaling training data, and the concepts of zero-shot and few-shot learning. We’ll compare LLMs with traditional Machine Learning models, explore embeddings and tokens, and discuss fine-tuning and prompt basics.
Resources
| Topic | Duration | Platform |
|---|---|---|
| Large Language Models (LLMs) Concepts | 2h | Datacamp |
| (Optional) Karpathy Introduction to LLMs | 1h | YouTube |
| (Optional) Andrew NG Introduction to LLMs (week 1) | 2h | Coursera (week 1 only) |
Practical Exercises
Try and document the following tasks and compare the answers with at least 2 services (Gemini, ChatGPT, Copilot, HuggingChat). Are they similar? Different?
- Ask about a recent event (e.g. the result of a football game from last week)
- Ask a Mathematical question
- Simple
- Complex
- Take a movie/game/music review and ask the LLM to classify the sentiment (positive/negative)
- Summarize a news article you read
- Complex instructions (What could we do here?)
- Generate some simple python code (e.g. code to train a python scikit model)
- Draft a work email asking a coworker for a project update
- Find 1 example where the answer given by the LLM is false
- Find 1 example where 1 LLM is correct but not the other one(s)
What were some surprise findings?
Week 2
This week builds on your experiences from Week 1 by focusing on the art and science of crafting effective prompts. We’ll review what worked and what didn’t, and delve into best practices for prompt engineering. You’ll see demonstrations on how to refine prompts to get better results.
Resources
| Topic | Duration | Platform |
|---|---|---|
| Understanding Prompt Engineering | 2h | Datacamp |
| A Beginner’s Guide to Prompt Engineering with ChatGPT | 1h | Datacamp Webinar |
| (Optional) Prompt Engineering Guide | - | Website |
| (Optional) OpenAI Prompt Engineering Guide | - | Website |
Practical Exercises
Complete the following:
- E-commerce description Task: Google Presentation
- (Optional) Test your prompt engineering skills against Gandalf in this Prompt Engineering game: Gandalf Game
Week 3
This week explores the real-world applications and integration of LLMs in various industries. We’ll discuss use-cases from companies like UBS and Unit8, and compare open-source vs closed-source LLMs. You’ll learn about QnA architectures, vector embeddings, and how LLMs can interact with external systems, such as calling APIs.
Resources
| Topic | Duration | Platform |
|---|---|---|
| Generative AI for Everyone (week 2 + 3) | 3h | Coursera |
| (Optional) Implementing AI Solutions in Business | 2h | Datacamp |
Practical Exercises
You’ll create a specialized Q&A system based on gathered documents and improve it to cite its sources correctly:
- Pick a topic of your choice
- Gather documents related to the topic
- Create a specialized QnA tool to answer questions on the documents
- Allow the LLM to cite the document source for the information provided (e.g. “[1] Document 1”)
Week 4
The week 4 of the Data Track is in general shorter and left free for the final presentation and any catch up that may be required.