ClassroomThe classroom
Back to the lesson libraryMECHANISM · 14 MIN
M20.2 CONNECT THE MECHANISM

Learn response behavior from demonstrations

Twenty-three tokens go in, but only eight of them teach anything. See how chat templates and loss masks turn written examples into an assistant that answers and stops.

LESSON OVERVIEW14 min lesson

Lesson overview

Twenty-three tokens go in, but only eight of them teach anything. See how chat templates and loss masks turn written examples into an assistant that answers and stops.

What you’ll explore

  • Trace a supervised conversation example, response-only loss mask, and evaluation split while distinguishing demonstrations from runtime prompts.
GO TO THE SOURCE

Original explanations, connected to the research.

Training language models to follow instructions with human feedback (InstructGPT; Ouyang et al., 2022)LIMA: Less Is More for Alignment (Zhou et al., 2023)Hugging Face Transformers documentation — chat templates
Suggest a correction

A precise note can make an explanation better.

Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.

The file includes this note, the scene title, and lesson metadata. Your saved progress and quiz responses are excluded. Download before leaving or reloading to keep your note.