embedding

Documentation

Overview

Package embedding defines a small, provider-agnostic interface for generating vector embeddings from text.

The core abstraction is the Embedder interface, which turns a slice of input texts into a slice of embedding vectors while preserving input order.

This package also defines common types used across implementations: - Vector: an embedding vector (typically []float32) - Model: a provider-specific model identifier string

Provider implementations live in subpackages (for example, embedding/openai).

Index

• Variables
• type Embedder
• type Model
• type Vector

Variables

var ErrEmptyInput = errors.New("empty input")

ErrEmptyInput is returned when an embedding call receives no input texts.

Types

type Embedder

type Embedder interface {
	Embed(ctx context.Context, texts []string) ([]Vector, error)
}

Embedder generates vector embeddings for input texts.

Implementations should preserve the ordering of inputs.

type Model

type Model string

Model identifies an embedding model name.

Examples: - OpenAI: "text-embedding-3-small" - Ollama: "nomic-embed-text".

type Vector

type Vector = []float32

Vector is an embedding vector.

Most OpenAI-compatible embedding endpoints return float32 vectors.