35:30Why Language Models Surprise You
36K views · 3mo ago · 35:30
In this video, I explain the softmax function and the concept of temperature in the context of language models and neural networks. I start by revisiting the Markov chain coding challenge, then break down how softmax converts raw neural network outputs (logits) into probabilities, and how temperature can be used to adjust those probabilities making outputs more random (higher temperature) or more deterministic (lower temperature).
Topics: Hello! · Markov chains as language models · From logits to probabilities · The softmax formula · Implementing softmax in p5.js · Weighted selection (sampling)
▶️ Mentioned videos: Previous · Next · Video on youtube.com · Video on youtube.com
📰 Mentioned news: Softmax function on Wikipedia · Euler's number on Wikipedia






















