LLMs, tokens, Git and GitHub

This whole document is from Class 1 of the AI Agents module, the part taught by Sir Faheem Akhtar.

AI models and LLMs

• LLM means Large Language Model. LLMs are trained on billions of data, and you can use them for your own work.
• ChatGPT and Claude are the names of apps. Opus, Sonnet, Astra and so on are the LLMs.¹

Context window

Every LLM has a specific context: a capacity for how much data it takes in. If a model's window is 1M tokens, that LLM can process 1M tokens of data.¹

Tokens

A token is the model's scale: how many words' weight it can carry. The easy way to think of it is that usually 1 token = 0.75 words, so 4 words are 5 tokens.¹

Git

• Git is a version control system. It keeps your work shipped.
• Say you did some work yesterday, and five days later you want to see what a file looked like five days ago. You can see that through Git.
• Git puts checkpoints in place. It takes snapshots of the code's history.
• Every time you commit while coding, a checkpoint is made, and you write a comment there, so that the changes you made can be read later.¹

GitHub

GitHub is file storage, or code storage. Just as a computer or a drive has folders, GitHub has repositories.¹

Footnotes

1. iSkills AI Agents Class 1 notes, introductory class, taught by Sir Tanveer Nandla and Sir Faheem Akhtar. ↩ ↩² ↩³ ↩⁴ ↩⁵