Case Study — Educational Tool
Perceptron LLM
An interactive playground for watching a language model choose its next word, one probability at a time.
What I Did
You type a prompt and it generates one token at a time, surfacing the candidate next words and their probabilities at each step. Temperature, top-p, and max tokens are exposed as live controls, so the sampling process that usually happens out of sight becomes something you can watch and adjust.
