Gathering the raw digital ingredients
Washing and prepping the data
Slow-cooking in the neural oven
Taste-testing and seasoning
YOUR GOAL
Master of The digital cookbook for hungry algorithms
007skaty Artificial intelligence systems use loss functions to measure and correct mistakes
007skaty Stirring Weights and Biases
We stir the weights and nudge the bias high, To feed a clever recipe to our AI.
007skaty Originally designed for video games, GPUs in massive server farms perform millions of math calculations simultaneously to power modern AI.
007skaty Since 2014, the Adam algorithm uses mathematical loss scores to adjust AI dials
007skaty Sort data ingredients into baking recipes to discover how we feed and train AI.
GPU heat: Training a single giant AI model can emit as much carbon as five cars over their lifetimes.
The process of making small, precise adjustments to something in order to achieve the best possible performance. Engineers in Geneva use this process over forty hours to adapt neural networks for domestic assistants.
Which hardware component, originally built for 3D video games, is the primary engine used to train neural networks?
007skaty The landmark paper Attention Is All You Need explains how self-attention links distant words to feed our digital AI families.
Meta built a massive computing cluster in 2024 featuring how many high-performance graphics processing units?
Which highly popular optimization algorithm, introduced in 2014, is used to adjust AI weights based on the loss function?
What mathematical tool measures the difference between an AI's prediction and the actual correct target answer?
In machine learning, what is the term for one complete pass of the entire training dataset through the neural network?
What is the term for the training error where a model memorizes its training data too well but fails on new data?
How many parameters (weights and biases) did the base GPT-3 model possess when it was released in 2020?