Intro to LLMs - What is an LLM? How do they work? 🎓

General
1

Here is a recording of the webinar we ran about LLMs. It was basically a non-technical, in-depth explanation of what a ChatGPT is, how it’s trained, and why it works.

Here’s a link to the 35-minute video What is an LLM?

WEBINAR HIGHLIGHTS
In the webinar we went over:

Definition of LLM: An LLM, or large language model, is basically really smart auto-complete.

Capabilities: How can LLMs do stuff like role-playing, answering questions, doing math, and even reasoning?

Training Process: How are LLMs trained?

  • Pre-Training Phase: Where model absorbs the entire internet (basically).
  • Post-Training Phase: Where the model learns how to act human.
  • Reinforcement Learning: Where the model learns from feedback.

Emergent Behaviors: How can LLMs develop new, unexpected ways of answering questions that weren’t directly taught during training?

Token Usage: What are tokens?

Hallucinations: Why do LLMs sometimes answer incorrectly? How can we prevent that?

3 Likes