Implementation:Neuml Txtai Annoy ANN

Knowledge Sources	Neuml_Txtai
Domains	Vector_Search, ANN
Last Updated	2026-02-10 01:00 GMT

Overview

Concrete ANN backend for approximate nearest neighbor search using the Annoy (Approximate Nearest Neighbors Oh Yeah) library, provided by txtai.

Description

Annoy is an ANN implementation that builds approximate nearest neighbor indexes using Spotify's Annoy library. It creates a forest of random projection trees for fast similarity search using dot product distance (equivalent to cosine similarity on normalized vectors). The index is built by adding items one at a time and then constructing the tree forest. Annoy is a read-only index after building -- it does not support append or delete operations natively.

Usage

Use the Annoy backend when you need a simple, memory-mapped read-only ANN index with fast search. Select this backend by setting the ANN backend configuration to "annoy". Requires the annoy Python package, installed via the txtai "ann" extra. Key tuning parameters include ntrees (number of trees in the forest) and searchk (search parameter controlling accuracy/speed tradeoff).

Code Reference

Source Location

• Repository: Neuml_Txtai
• File: src/python/txtai/ann/dense/annoy.py
• Lines: 1-73

Signature

class Annoy(ANN):
    """Builds an ANN index using the Annoy library."""

    def __init__(self, config)
    def load(self, path)
    def index(self, embeddings)
    def search(self, queries, limit)
    def count(self)
    def save(self, path)

Import

from txtai.ann import ANNFactory

I/O Contract

Inputs

Name	Type	Required	Description
config	dict	Yes	ANN configuration dictionary containing backend settings
config["backend"]	str	Yes	Must be set to "annoy" to select this backend
config["dimensions"]	int	Yes	Dimensionality of the embedding vectors
ntrees	int	No	Number of trees in the forest (default: 10)
searchk	int	No	Search parameter controlling accuracy vs speed (default: -1, which uses ntrees * limit)

Outputs

Name	Type	Description
search() returns	list	List of lists of (id, score) tuples, one list per query
count() returns	int	Number of items in the Annoy index
save() side-effect	file	Persists the Annoy index to a binary file at the specified path

Usage Examples

from txtai import Embeddings

# Create embeddings with Annoy backend
embeddings = Embeddings({
    "path": "sentence-transformers/all-MiniLM-L6-v2",
    "backend": "annoy",
    "annoy": {
        "ntrees": 10,
        "searchk": 1000
    }
})

# Index data
embeddings.index([
    "US tops 5 million confirmed virus cases",
    "Canada's last intact ice shelf has broken up",
    "Beijing urges strong action on climate change",
    "New York battles severe winter storm"
])

# Search
results = embeddings.search("climate change effects", 2)
print(results)

Related Pages

• Environment:Neuml_Txtai_Python_Core_Dependencies