Text to Numbers

Machines Only Read Numbers

A model does not see the word cat. It sees numbers.

Computers cannot directly understand human language. They only process numbers.

So every prompt you write has to be translated before a neural network can touch it.

That translation step is where LLM work actually begins.

What a Vector Actually Is

A vector is an array of numbers.

More precisely, it is a single-dimensional array: one ordered list, no rows and columns.

Notice that the whole word collapses into one ordered row of numbers, with no second dimension of rows and columns.

AI systems convert words and sentences into these arrays so neural networks can process them.

The vector is the representation. The text is just what you typed.

Think of it as the data type your prompt becomes.

From Prompt to Numbers

The conversion is not decorative. It is a required stage in the pipeline.

Your text goes in as language and comes out as arrays of numbers ready for a neural network.

Everything downstream depends on this handoff being done first.

Vectors Flow Through Every Layer

Vectors are not a one-time entry ticket.

They are the fundamental data structure that flows through every layer of an LLM.

Each layer receives numbers, transforms them, and passes numbers onward.

Nothing turns back into text in the middle.

Why Numbers Enable Understanding

Once text is numeric, math becomes possible.

Vectors enable the mathematical operations that ultimately produce intelligent language understanding.

Understanding here is the output of arithmetic performed on arrays, layer after layer.

That is a useful reframe when you debug a prompt: you are shifting the numbers that enter the network.

Common Misreadings

A vector is not a stored copy of your sentence.

It is a numeric representation the network can process, nothing more.

It is also not two-dimensional. A vector is a single-dimensional array of numbers.

Keeping these straight matters before you look at how those numbers get produced.

What to Carry Forward

Three ideas do the work here.

Computers process only numbers, so text must be converted. Vectors are the single-dimensional arrays that carry that text.

Those vectors move through every layer of an LLM, enabling the math behind language understanding.

Next comes the question of how words become the specific numbers inside a vector.

Does a neural network directly process the word "cat" as human-readable text?

• Yes
• No

Answer: No

No. Computers process numbers, so the word must be converted into a numeric representation before the neural network can use it.

A text prompt becomes an _____ representation before a _____ can process it.

Answer: A text prompt becomes an numeric representation before a neural network can process it.

Text must become numeric before processing begins. The resulting numbers can be handled by a neural network.

Which statement best describes how vectors move through an LLM?

• Each layer receives and passes numeric vectors.
• Each layer converts the data back into text.
• Only the input layer uses numeric vectors.
• The model stores the original sentence in every layer.

Answer: Each layer receives and passes numeric vectors.

Vectors are not used only at the input. Each layer receives numeric data, transforms it, and passes numeric data onward.