Ruby

DATABASE_URL = "postgres://postgres:postgres@localhost:5432/postgres"
embedding = [1, 2, 3]
query = "My text input"

gem install pg

require 'pg'

conn = PG.connect(DATABASE_URL)

# Enable the extension
conn.exec('CREATE EXTENSION IF NOT EXISTS lantern')

# Create a table
conn.exec('CREATE TABLE IF NOT EXISTS items (id bigserial PRIMARY KEY, embedding REAL[3])')

# Insert a vector
conn.exec_params("INSERT INTO books (book_embedding) VALUES ($1)", [embedding])

# Find nearest rows to a vector
conn.exec_params("SELECT * FROM books ORDER BY book_embedding <-> $1 LIMIT 5", [embedding])

# Find nearest rows to a vector generated from text
conn.exec_params("SELECT * FROM books ORDER BY book_embedding <-> text_embedding('BAAI/bge-small-en', $1) LIMIT 5", [query])

conn.close

gem install sequel

require 'sequel'

# Connect to the database
DB = Sequel.connect(DATABASE_URL)

# Create a table
DB.create_table?(:items) do
  primary_key :id
  column :embedding, 'REAL[3]'
end

# Insert a vector
DB[:books].insert(book_embedding: embedding)

# Find nearest rows to a vector
DB[:books].order(Sequel.lit('book_embedding <-> ?', embedding)).limit(5)

# Find nearest rows to a vector generated from text
DB[:books].order(Sequel.lit("book_embedding <-> text_embedding(?, ?)", 'BAAI/bge-small-en', query)).limit(5)

gem install lanterndb

require 'active_record'
require 'lantern'
ActiveRecord::Base.establish_connection(DATABASE_URL)
ActiveRecord::Base.connection.enable_extension("lantern")
conn = ActiveRecord::Base.connection

ActiveRecord::Migration.create_table :movies do |t|
  t.column :movie_embedding, :real, array: true
end
conn.execute("INSERT INTO movies (movie_embedding) VALUES ('{0,1,0}'), ('{3,2,4}')")

embedding1 = Lantern.text_embedding('BAAI/bge-base-en', 'Your text here')

Lantern.set_api_token(openai_token: 'your_openai_token')
embedding2 = Lantern.openai_embedding('text-embedding-3-small', 'Hello')

Lantern.set_api_token(cohere_token: 'your_cohere_token')
embedding3 = Lantern.cohere_embedding('embed-english-v3.0', 'Hello')

class Document < ApplicationRecord
  has_neighbors :embedding
end

# Find 5 nearest neighbors using L2 distance
Document.nearest_neighbors(:embedding, [0.1, 0.2, 0.3], distance: 'l2').limit(5)

# Given a document, find 5 nearest neighbors using cosine distance
document = Document.first
document.nearest_neighbors(:embedding, distance: 'cosine').limit(5)

class Book < ApplicationRecord
  has_neighbors :embedding
end

# Find 5 nearest neighbors using open-source model
Book.nearest_neighbors(:embedding, 'The quick brown fox', model: 'BAAI/bge-small-en', distance: 'l2').limit(5)

# Find 5 nearest neighbors using OpenAI
Lantern.set_api_token(openai_token: 'your_openai_token')
Book.nearest_neighbors(:embedding, 'The quick brown fox', model: 'openai/text-embedding-3-small', distance: 'cosine').limit(5)

class CreateVectorIndex < ActiveRecord::Migration[7.0]
  def up
    add_index :books, :embedding, using: :lantern_hnsw, opclass: :dist_l2sq_ops, name: 'book_embedding_index'
  end

  def down
    remove_index :books, name: 'book_embedding_index'
  end
end

class CreateHnswIndex < ActiveRecord::Migration[7.0]
  def up
    execute <<-SQL
      CREATE INDEX movie_embedding_hnsw_idx
      ON movies
      USING lantern_hnsw (movie_embedding dist_l2sq_ops)
      WITH (
        ef = 15,
        m = 16,
        ef_construction = 64
      )
    SQL
  end

  def down
    remove_index :movies, name: 'movie_embedding_hnsw_idx'
  end
end

gem 'lanterndb'

bundle install

rails generate lantern:install
rails db:migrate