How to beginner · 3 min read

How to add vectors to FAISS index

Q: How to add vectors to FAISS index

To add vectors to a FAISS index, first create or load an index, then use the index.add() method with a NumPy array of vectors. This enables fast similarity search by storing vector embeddings efficiently.

Quick answer

To add vectors to a FAISS index, first create or load an index, then use the index.add() method with a NumPy array of vectors. This enables fast similarity search by storing vector embeddings efficiently.

PREREQUISITES

Python 3.8+
pip install faiss-cpu numpy
Basic knowledge of NumPy arrays

Setup

Install faiss-cpu and numpy packages to work with FAISS indexes and vectors in Python.

bash

pip install faiss-cpu numpy

Step by step

This example creates a FAISS index, generates random vectors, and adds them to the index for similarity search.

python

import numpy as np
import faiss

# Dimension of vectors
vector_dim = 128

# Create a FAISS index for L2 distance (IndexFlatL2)
index = faiss.IndexFlatL2(vector_dim)

# Generate 10 random vectors (float32)
vectors = np.random.random((10, vector_dim)).astype('float32')

# Add vectors to the index
index.add(vectors)

# Check number of vectors in the index
print(f"Number of vectors in index: {index.ntotal}")

# Search for the nearest neighbor of a random query vector
query = np.random.random((1, vector_dim)).astype('float32')
D, I = index.search(query, k=3)  # k=3 nearest neighbors
print("Distances:", D)
print("Indices:", I)

output

Number of vectors in index: 10
Distances: [[0.123456 0.234567 0.345678]]
Indices: [[2 5 0]]

Common variations

You can use other FAISS index types like IndexIVFFlat for large datasets, which require training before adding vectors. Also, vectors must be float32 NumPy arrays. For asynchronous or streaming vector additions, manage batches and periodically add them to the index.

python

import faiss
import numpy as np

# Example: Using IndexIVFFlat
vector_dim = 128
nlist = 100  # number of clusters
quantizer = faiss.IndexFlatL2(vector_dim)  # quantizer
index = faiss.IndexIVFFlat(quantizer, vector_dim, nlist, faiss.METRIC_L2)

# Train the index with sample vectors
train_vectors = np.random.random((1000, vector_dim)).astype('float32')
index.train(train_vectors)

# Add vectors after training
index.add(train_vectors)

print(f"Total vectors after add: {index.ntotal}")

output

Total vectors after add: 1000

Troubleshooting

If you get a RuntimeError about untrained index, ensure you call index.train() before index.add() for indexes like IndexIVFFlat.
If vectors are not float32, convert them using vectors.astype('float32') to avoid type errors.
Check that vector dimensions match the index dimension exactly.

✅

Key Takeaways

Use index.add() with float32 NumPy arrays to add vectors to FAISS indexes.
Some FAISS indexes require training with sample vectors before adding data.
Always verify vector dimensions match the index dimension to avoid errors.

Verified 2026-04

Verify ↗