Nearby lessons

20 of 30

Encoding & Decoding

Concept in One Line

Encoding means converting text into numbers, and decoding means converting numbers back into readable text.

Why This Concept Is Important

  • Computers understand only numbers.
  • GPT works internally using numbers.
  • Helps students understand how AI really processes language.

Key Points to Remember

  • GPT cannot process text directly.
  • Text is first converted into tokens.
  • Tokens are converted into numbers (encoding).
  • The model processes numbers internally.
  • Output numbers are converted back to text (decoding).

Encoding Explained

What is Encoding?

Converting tokens into numbers.

Each token gets a numeric ID.

Numbers carry learned meaning.

Example:

"Hello AI"

-> Tokens -> ["Hello", " AI"]

-> Numbers -> [13225, 481]

Numbers are for machine use only.

Decoding Explained

What is Decoding?

Converting numbers back to tokens.

Tokens are converted to readable text.

Final answer is shown to the user.

Example:

[13225, 481]

-> Tokens -> ["Hello", " AI"]

-> Text -> "Hello AI"

Simple Flow

Text -> Tokens -> Numbers (Encoding) -> Model Processing -> Numbers -> Tokens -> Text (Decoding)

Interview / Exam Points

Q1: What is encoding in GPT?

Encoding is the process of converting text tokens into numbers so that the model can process them.

Q2: What is decoding in GPT?

Decoding is the process of converting numbers back into readable text.

Common Confusions to Avoid

  • Encoding is not encryption.
  • Numbers do not mean human words.
  • Encoding is for machine understanding.
  • Decoding is for human reading.

One Line to Remember

GPT thinks in numbers, humans read text.

Quick Self-Check

For Students

Encoding converts text to numbers - Yes / No?

Decoding converts numbers to text - Yes / No?

(Correct answers: Yes, Yes)

🧠 Test Your Knowledge

3 Questions

Progress: 0 / 3
Keep Going!Simple Tokenizer Logic