• Widely adopted

  48,116 stars

• Repeat trending

  9 trending appearances

Dive-into-llms provides an extensive, open-source educational repository focused on the comprehensive understanding of Large Language Models (LLMs). It demystifies the complex architecture and underlying mechanisms of state-of-the-art models, offering deep technical insights into pre-training, fine-tuning techniques (like LoRA and QLoRA), and efficient inference optimization. The project combines theoretical explanations with practical, hands-on code examples to guide users through the complete lifecycle of LLM development. By bridging the gap between academic research and practical engineering, it enables developers and researchers to build, deploy, and optimize their own models.

This project is tailored for AI researchers, machine learning engineers, and software developers who want to move beyond using LLM APIs.

• Architectural Deep Dives: Detailed explanations and breakdowns of Transformer architectures and modern LLM variants.
• Fine-Tuning Frameworks: Step-by-step guides and code implementations for advanced fine-tuning methods including LoRA, QLoRA, and P-Tuning.
• Inference Optimization: Techniques and tools for optimizing model inference, reducing memory footprint, and increasing throughput in production.
• Practical Hands-On Code: Provides executable notebooks and scripts to directly apply theoretical knowledge to real-world datasets and models.
• Evaluation Methodologies: Details techniques for accurately benchmarking and evaluating model performance across specific domains.

Getting started: Clone the repository and explore the Jupyter notebooks provided in the root directory.

1. dive-into-llmsv1Jun 12, 2025

• microsoft/generative-ai-for-beginners

  21 Lessons, Get Started Building with Generative AI

  116.9K stars · Jupyter Notebook

• rasbt/LLMs-from-scratch

  Implement a ChatGPT-like LLM in PyTorch from scratch, step by step

  100.8K stars · Jupyter Notebook

• microsoft/ai-agents-for-beginners

  18 Lessons to Get Started Building AI Agents

  71.5K stars · Jupyter Notebook

• microsoft/AI-For-Beginners

  12 Weeks, 24 Lessons, AI for All!

  63.1K stars · Jupyter Notebook

• anthropics/claude-cookbooks

  A collection of notebooks/recipes showcasing some fun and effective ways of using Claude.

  51.1K stars · Jupyter Notebook

• DataTalksClub/data-engineering-zoomcamp

  Data Engineering Zoomcamp is a free 9-week course on building production-ready data pipelines. The next cohort starts in January 2026. Join the course here 👇🏼

  44.4K stars · Jupyter Notebook