These paths are sequenced — each step builds on the last. The goal is not to skim everything; it’s to understand one thing well enough that the next thing makes sense without re-reading the prerequisites.
Each path has a recommended reading order with a brief bridge explaining why that order matters. Follow the links to the concept or source pages for the actual material.
Paths
- llm-training — How LLMs Are Trained: From Scratch to RLHF
- efficient-inference — Making LLMs Fast: The Inference Efficiency Stack
- vision-language — From Pixels to Understanding: Vision-Language Models