Skip to main content

Exercise 1 - Your First Embedding

Goal: Call the embedding API, inspect the raw output, and understand what the numbers represent.

Assignment​

Open 01_first_embedding.py and run it:

python exercises/01_first_embedding.py

The script embeds one sentence and prints the vector's shape, range, and magnitude. Read the output carefully.

Now make these changes:

  1. Try changing one word in the sentence. Re-run. Do the values shift noticeably?
  2. Try embedding a completely unrelated sentence (e.g. "The price of eggs went up this quarter."). Compare the first 8 values side by side. Are they different?
  3. Try embedding an empty string "". What happens?

Thinking questions​

  • The vector has 768 dimensions. Each number is meaningless on its own. Where is the meaning?
  • The magnitude (L2 norm) is probably not 1.0. What would you need to do to fix that, and why does it matter?
  • If you embed the same sentence twice, do you get identical vectors? Why?

Next: Exercise 2 - Similarity →