In-Context Learning & Prompting

You now understand how LLMs learn from examples without changing their weights: The secret of In-Context Learning, the power of System Prompts, and why format often matters more than content.

In-Context Learning System Prompts Attention Heatmaps Lost-in-the-Middle Format vs. Content N-Shot Scaling Few-Shot Learning

Continue with Chapter 6

Training & Inference

Dive into the world of LLM training: RLHF, DPO, sampling strategies like Temperature and Top-K/Top-P, as well as optimizations like Quantization and Speculative Decoding.

Progress: Chapter 5 of 8