Lilly Tech Systems
Learn LLM Models
Comprehensive guide to understanding Large Language Models — how they work, their architecture, training, fine-tuning, and practical usage across open and closed ecosystems.
Your Learning Path
Follow these lessons in order, or jump to any topic that interests you.
1. Introduction
What are LLMs? History from GPT-1 to modern models, capabilities, limitations, and the Transformer revolution.
2. How LLMs Work
Transformer architecture, tokenization, self-attention, pre-training objectives, and emergence.
3. Training LLMs
Pre-training process, data collection, compute requirements, RLHF, Constitutional AI, and DPO.
4. Fine-tuning
Full fine-tuning, LoRA, QLoRA, instruction tuning, and when to fine-tune vs prompt vs RAG.
5. Open vs Closed Models
Compare GPT-4, Claude, Gemini with LLaMA, Mistral, Qwen. Licensing, pros, cons, and use cases.
6. Running Local LLMs
Hardware requirements, quantization, Ollama, LM Studio, llama.cpp, and performance benchmarks.
7. LLM APIs
API providers, pricing comparison, streaming, function calling, and best practices.
8. Best Practices
Choosing the right LLM, cost optimization, safety, evaluation benchmarks, and building LLM apps.
What You'll Learn
By the end of this course, you'll be able to:
Understand LLM Architecture
Know how Transformers, attention, and tokenization work under the hood of modern LLMs.
Fine-tune Models
Apply LoRA, QLoRA, and instruction tuning to adapt LLMs for your specific use case.
Choose the Right Model
Evaluate open vs closed models, compare APIs, and select the best option for your needs.
Run LLMs Locally
Set up and run quantized models on your own hardware with Ollama, LM Studio, and more.