CH2: Understanding LLM's

Question 1

What is a Large Language Model (LLM)?

Answer 1

A Large Language Model (LLM) is a type of AI trained to understand and generate human language. It reads huge amounts of text and learns how people write and speak.

Example:
It can write essays, answer questions, and chat with you just like a real person.

Question 2

Why is it called ‘Large’ in Large Language Model?

Answer 2

The word ‘Large’ means these models are trained on very big sets of data — sometimes hundreds of gigabytes. They also have billions or even trillions of little pieces inside them called parameters that help them learn.

Example:
Think of it like a super-sized brain trained on almost everything written on the internet.

Question 3

What can LLMs do?

Answer 3

LLMs can write sentences, answer questions, explain things, translate languages, summarize long texts, and even chat like a person.

Example:
It’s like the smartest autocomplete you’ve ever seen — it knows what you want to say next.

Question 4

How does a Large Language Model work?

Answer 4

LLMs work step-by-step:
1. They split up the text into small parts (tokenization)
2. They turn these parts into numbers (embeddings)
3. They focus on important words (attention)
4. They understand the meaning (neural layers)
5. They guess what words should come next (text generation)

Example:
It’s like reading your sentence one word at a time and guessing the next word over and over until the sentence is done.

Question 5

What are the strengths of LLMs?

Answer 5

LLMs are great at reading and writing text that sounds natural. They can understand complicated topics and change how they talk depending on what you need.

Example:
They can write a friendly message or a formal report depending on how you ask.

Question 6

What are the weaknesses of LLMs?

Answer 6

LLMs sometimes make things up — this is called hallucination. They don’t really understand facts like humans do. They also might have bias and don’t know anything that happened after their training ended.

Example:
It might say something confidently even if it’s wrong because it was just guessing based on patterns it learned.