Understanding LLMs
Lesson 4 of 6
0%Lesson 4
40 min
Training: Pre-training, Fine-tuning, and RLHF
Listen to the full lesson
AI NarrationQuick Summary
Pre-training builds raw language ability. Fine-tuning teaches a model to follow instructions in a specific format. RLHF aligns it to human preferences. The same base weights become wildly different products depending on which of these stages you customize.
What you will learn
- ·Understand the difference between pre-training, fine-tuning, and RLHF
- ·Explain what instruction-tuning and alignment mean
- ·Know why a raw pre-trained model behaves differently from a chat model