This repository offers a practical, hands-on course for learning about Large Language Models (LLMs), targeting engineers and researchers. It provides a structured curriculum covering LLM techniques, libraries, project implementation, and enterprise solutions, enabling users to build and deploy LLM-powered applications.

How It Works

The course is structured into three main sections: Techniques and Libraries, Projects, and Enterprise Solutions. It emphasizes a learn-by-doing approach, with each lesson featuring accompanying notebooks (often hosted on Google Colab or Kaggle) and detailed articles. The content is grounded in published research papers and covers a wide range of LLM tools and methodologies, including OpenAI API, Hugging Face, LangChain, vector databases, and various fine-tuning techniques like LoRA and QLoRA.

Quick Start & Requirements

• Notebooks are primarily designed to run in Google Colab or Kaggle, with some requiring Colab Pro for higher memory needs.
• Users can also run notebooks in their own environments.
• Key dependencies include Python, OpenAI API access, Hugging Face libraries, and potentially specific databases like ChromaDB.
• Links to specific lessons, notebooks, and articles are provided within the README.

Highlighted Details

• Covers advanced fine-tuning techniques such as PEFT, LoRA, and QLoRA.
• Explores Retrieval Augmented Generation (RAG) with vector databases (ChromaDB, Faiss) and semantic caching.
• Includes practical projects like building chatbots, NL2SQL translators, and data analyst assistants using LangChain agents.
• Details LLM evaluation metrics (BLEU, ROUGE) and monitoring tools (LangSmith, Giskard).

Maintenance & Community

The repository is under permanent development, with new lessons and examples being added. Users are encouraged to contribute by opening issues for problems or starting discussions for suggestions. The author is active on LinkedIn and Twitter.

Licensing & Compatibility

The repository content is provided for educational purposes. Specific library licenses apply to the code within the notebooks.

Limitations & Caveats

Some notebooks require more memory than the free tier of Google Colab offers. The course is an unofficial repository for a book, and does not contain all information from the published book. Some sections are marked as "Work in Progress" (WIP).