Chapter One · complete
Good job!
Here’s what you learned.
1 · Prompt and tokens
A prompt is split into tokens: whole words or smaller pieces of text.
2 · Context
The context keeps the prompt’s tokens and each new output, so the next answer can use what came before.
3 · Embedding
An embedding turns each token into numbers the network can use.
4 · Neurons
A neuron multiplies inputs by weights, adds them and a bias, then produces an output.
5 · Weights and bias
Weights set each input’s influence. Bias adds a steady nudge.
6 · Training
Training adjusts weights and biases to reduce loss, a measure of error. An optimizer automates those changes.