Choose cosine similarity or dot product for your vectors

Ranking the same two vectors by dot product gave a different top match than ranking them by cosine similarity, and I had no way to know which one to trust.

BeginnerRead 4 minDoing: under an hour6 steps

▶ Start

What you'll have at the end

A short note recording which metric your project uses and why, based on how the model was trained.

Before you start

You need

A working way to turn real text into embedding vectors, an API call or a loaded model, that already returns real numbers for at least two different pieces of text.

Not covered

Configuring which metric a specific vector database or ANN index enforces at query time; this only covers the metric's own math and reading it off your model's documentation.

Leans on

6 steps

Where to go next

Verify your first embedding works

go there first if you don't yet have a real, working embedding call to measure the length of

Speed up nearest neighbor search with a FAISS index

go there once the metric is settled and the open question becomes configuring the index that enforces it at query time

Checked 15 Aug 2026

Part of the Embeddings cookbook