🧠 LLM Learning Hub

A comprehensive, structured learning repository for Large Language Models — from fundamentals to production deployment.

---

📚 Repository Structure

• llm-learning-hub/
  • 01-foundations/
    • 01-mathematics-for-ml/ - Linear algebra, calculus, probability
    • 02-deep-learning-basics/ - Neural networks, backprop, optimization
    • 03-transformers/ - Attention mechanism, architecture
  • 02-llm-science/
    • 01-tokenization/ - BPE, WordPiece, SentencePiece
    • 02-pretraining/ - Data pipelines, scaling laws
    • 03-fine-tuning/ - SFT, LoRA, QLoRA
    • 04-alignment/ - RLHF, DPO, GRPO
    • 05-evaluation/ - Benchmarks, LLM-as-a-Judge
  • 03-llm-engineering/
    • 01-rag-systems/ - Retrieval-Augmented Generation
    • 02-agents/ - AI agents, tool use
    • 03-prompt-engineering/ - Prompt patterns, CoT
    • 04-deployment/ - Quantization, inference
  • 04-projects/
    • 01-chatbot/ - Production chatbot
    • 02-code-assistant/ - Coding assistant
    • 03-research-assistant/ - Paper analysis tool
  • 05-resources/
    • papers.md - Research papers
    • courses.md - Courses & books
  • scripts/
    • setup.sh
    • download-datasets.sh
  • requirements.txt
  • LICENSE

---

🔰 Beginner Path (No ML Background)

Week	Topic	Notebook / Path
1-2	Math Basics	01-mathematics-for-ml/
3-4	Deep Learning	02-deep-learning-basics/
5-6	Transformers	transformer-from-scratch.ipynb
7-8	Tokenization	01-bpe-from-scratch.ipynb
9-10	Fine-tuning	01-lora-finetuning.ipynb
11-12	RAG Systems	01-basic-rag.ipynb
13-14	Agents	01-react-agent.ipynb
15-20	Projects	04-projects/

---

🚀 Quick Start (5 Minutes)

Step 1: Clone & Setup

# Clone the repository
git clone https://github.com/jobiryasir/llm-learning-hub.git
cd llm-learning-hub

# Set up environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate
pip install -r requirements.txt

Step 2: Verify Installation

# Check everything works
python -c "import torch; print('PyTorch:', torch.__version__)"
python -c "import transformers; print('Transformers:', transformers.__version__)"

# Start Jupyter
jupyter lab

Step 3: Start Learning! 🎯

👉 Open your first notebook:

# In Jupyter, navigate to:
01-foundations/03-transformers/02-transformer-from-scratch.ipynb

---

🗺️ What to Do After Setup

🔰 Beginner Path (No ML Background)

Week	Topic	Notebook
1-2	Math Basics	Review 01-foundations/01-mathematics-for-ml/
3-4	Deep Learning	Run 01-foundations/02-deep-learning-basics/notebooks/
5-6	Transformers	Complete 01-foundations/03-transformers/02-transformer-from-scratch.ipynb
7-8	Tokenization	Implement 02-llm-science/01-tokenization/01-bpe-from-scratch.ipynb
9-10	Fine-tuning	Run 02-llm-science/03-fine-tuning/01-lora-finetuning.ipynb
11-12	RAG Systems	Build 03-llm-engineering/01-rag-systems/01-basic-rag.ipynb
13-14	Agents	Create 03-llm-engineering/02-agents/01-react-agent.ipynb
15-20	Projects	Complete any project in 04-projects/

🚀 Intermediate Path (Knows ML Basics)

Skip 01-foundations and start directly from:

# Week 1-2: Transformers from scratch
01-foundations/03-transformers/02-transformer-from-scratch.ipynb

# Week 3-4: Fine-tuning with LoRA
02-llm-science/03-fine-tuning/01-lora-finetuning.ipynb

# Week 5-6: RAG Systems
03-llm-engineering/01-rag-systems/01-basic-rag.ipynb

# Week 7+: Projects
04-projects/01-chatbot/

🎯 Quick Start (Today!)

Want to see something cool right now?

# 1. Start Jupyter
jupyter lab

# 2. Open this notebook and run all cells:
# 01-foundations/03-transformers/02-transformer-from-scratch.ipynb

# 3. You'll build a complete GPT model from scratch!

---

🛠️ Tech Stack

Category	Tools
Frameworks	PyTorch, Hugging Face Transformers, TRL
Training	DeepSpeed, Unsloth, Axolotl
Inference	vLLM, llama.cpp, Ollama
RAG	LangChain, LlamaIndex, ChromaDB
Agents	CrewAI, AutoGen, smolagents
Evaluation	LM Evaluation Harness, OpenCompass
Tracking	Weights & Biases, MLflow

---

📖 How to Use This Repository

1. Follow the roadmap in 05-resources/roadmap.md
2. Complete notebooks in each section — they contain hands-on implementations
3. Build projects in 04-projects/ to apply what you've learned
4. Refer to resources in 05-resources/ for papers, courses, and tools

---

🎯 Milestones

Milestone	Check
⬜ Build transformer from scratch	Week 4
⬜ Fine-tune a 7B model	Week 8
⬜ Train with DPO/GRPO	Week 9
⬜ Build working RAG system	Week 11
⬜ Deploy a model to production	Week 16
⬜ Complete 3 end-to-end projects	Week 20

---

---

🤝 Contributing

This is a personal learning repository, but suggestions and improvements are welcome! See CONTRIBUTING.md for guidelines.

---

📄 License

This project is licensed under the MIT License — see the LICENSE file.

---

Note: This repository is completely independent and not affiliated with any previous projects. Built from scratch for focused LLM learning.