Same Prompt, Twice
Why asking an AI the same question twice can give two different answers. Worked solutions on a toy language model: the NOT, OR and AND rules, how many calls it takes before a rare output shows up, and the temperature dial.
Worked solutions and Manim animations from my section of Stanford's Probability for AI. Every week has a notes PDF to download and a video to watch.
Probability for AI teaches the math language that AI speaks. I'm a section leader in the course, working with top tech minds, researchers, and working professionals from around the world.
After each class I write detailed worked solutions and build a Manim animation that shows the week's ideas moving. Everything lands here, one week at a time.
Why asking an AI the same question twice can give two different answers. Worked solutions on a toy language model: the NOT, OR and AND rules, how many calls it takes before a rare output shows up, and the temperature dial.
How do you measure something people will not say out loud? The list experiment, built on Bernoulli and binomial random variables, and what an accuracy number really means on a finite test set.
Worked solutions and a Manim animation for this week's section.
New weeks are added as the course moves along.
For teaching, research collaboration, or PhD-related conversations, reach out directly.
sohaibhassan199@gmail.com