LLM-learning-path.md

REALLY GOOD STUFF for LLMs

This pathway is designed to guide learners through the essentials of Large Language Models (LLMs), from beginner to advanced levels, with a mix of theoretical concepts and practical coding. It is organized by weeks, with color codes to signify progression:

• 🔵 Week 1: Foundational concepts
• 🟣 Week 2: Intermediate topics
• 🟠 Week 3: Advanced techniques
• 🔴 Must-see (advanced topics)
• ⚫ Optional, specialized content

Starting with a beginner's introduction, you can move into hands-on coding with Langchain, explore structured output with Pydantic, and dive into evaluation methods. The pathway also covers in-context learning, Retrieval-Augmented Generation (RAG), and running LLMs locally, concluding with advanced and optional materials for those who want to go deeper. An alternative pathway includes additional YouTube playlists, books, and courses for further learning.

Start - concept

🔵 Beginner intro to LLMs

⚫ Fun video about the mathematics happening behind the scenes (I promise it's fun)

🟣 RAG Explained

Start - code

🔵 Langchain playlist - fast

OR

🔵 Langchain - master class

Then you can go to each section, as you need, and do something on your PC. Nothing is comparable to getting something running on your own PC (either by API or local LLMs).

Structured Output

🔵 Structured Output w Pydantic

⚫ Optional: JSONformer

evaluation

🟣 Deep dive evaluation

🔵 QA evaluation

⚫ Langsmith

In-context learning (fine-tune)

🟠 Llama3.2 + Ollama

⚫ Llama3

🟠 BERT

RAG

⚫ All about RAG (must see)

🟣 Multimodal RAG + Langchain + GPT4V

⚫ Create your RAG-bot with UI (streamlit)

Running LLMs locally

🔵 LM studio basic ++ Python integration --> recommended for beginners

🟣 Ollama basics ++ Python integration --> Recommended for iterative runs

Advanced - musT

🟠 Using Huggingface Transformer and pre-trained modelslibrary

🟠 Groq API

🟠 Advanced LLM concepts

🟠 Technical Intro to LLMs

🔴 Creating knowledge graphs

Advanced - musT - Agentic

🔴 Agentic Design of LLMs

• https://developer.nvidia.com/blog/introduction-to-llm-agents/

• https://www.superannotate.com/blog/llm-agents

• https://www.promptingguide.ai/research/llm-agents

• https://www.truefoundry.com/blog/llm-agents

• https://towardsdatascience.com/intro-to-llm-agents-with-langchain-when-rag-is-not-enough-7d8c08145834

Advanced - optional

⚫ RAG + Ollama + n8n

⚫ Langflow (UI for creating RAG)

⚫ Local LLMs in minutes

If you finished your first project and want to go deeper:

⚫ Learn the fundamentals of generative AI for real-world applications by deeplearning.io

⚫ Building systems with LLMs

Alternative to this pathway

YT Playlist

⬛ YT playlist covering all aspects, from beginner-level to advanced, including both code and concepts: Generative AI by Washington Prof

Books

⬛ A book covering all aspects of LLMs (really a great book, the second book I tried to read cover-to-cover in my life): Hands-on Large Language Models by J. Alammar

Courses

⬛ Building Production-Ready Apps with Large Language Models

⬛ Learn the fundamentals of generative AI for real-world applications by deeplearning.io

Bootcamps

⬛ Bootcamp for production=level LLM (code are old though)