Class: LLM::Gemini

Inherits:
Provider show all
Includes:
Format
Defined in:
lib/llm/providers/gemini.rb,
lib/llm/providers/gemini/audio.rb,
lib/llm/providers/gemini/files.rb,
lib/llm/providers/gemini/format.rb,
lib/llm/providers/gemini/images.rb,
lib/llm/providers/gemini/models.rb,
lib/llm/providers/gemini/error_handler.rb,
lib/llm/providers/gemini/response_parser.rb

Overview

The Gemini class implements a provider for [Gemini](ai.google.dev/).

The Gemini provider can accept multiple inputs (text, images, audio, and video). The inputs can be provided inline via the prompt for files under 20MB or via the Gemini Files API for files that are over 20MB

Examples:

#!/usr/bin/env ruby
require "llm"

llm = LLM.gemini(ENV["KEY"])
bot = LLM::Chat.new(llm).lazy
bot.chat LLM::File("/images/capybara.png")
bot.chat "Describe the image"
bot.messages.select(&:assistant?).each { print "[#{_1.role}]", _1.content, "\n" }
#!/usr/bin/env ruby
require "llm"

llm = LLM.gemini(ENV["KEY"])
bot = LLM::Chat.new(llm).lazy
bot.chat ["Describe the image", LLM::File("/images/capybara.png")]
bot.messages.select(&:assistant?).each { print "[#{_1.role}]", _1.content, "\n" }

Defined Under Namespace

Modules: Format, ResponseParser Classes: Audio, ErrorHandler, Files, Images, Models

Constant Summary collapse

HOST = 
"generativelanguage.googleapis.com"

Instance Method Summary collapse

Methods included from Format

#format

Methods inherited from Provider

#chat, #chat!, #inspect, #respond, #respond!, #responses, #schema

Constructor Details

#initialize(secret) ⇒ Gemini

Returns a new instance of Gemini.

Parameters:

  • secret (String) —

    The secret key for authentication



44
45
46
 
# File 'lib/llm/providers/gemini.rb', line 44

def initialize(secret, **)
  super(secret, host: HOST, **)
end

Instance Method Details

#assistant_role ⇒ String

Returns the role of the assistant in the conversation. Usually “assistant” or “model”

Returns:

  • (String) —

    Returns the role of the assistant in the conversation. Usually “assistant” or “model”



119
120
121
 
# File 'lib/llm/providers/gemini.rb', line 119

def assistant_role
  "model"
end

#audio ⇒ Object

Provides an interface to Gemini’s audio API

See Also:



91
92
93
 
# File 'lib/llm/providers/gemini.rb', line 91

def audio
  LLM::Gemini::Audio.new(self)
end

#complete(prompt, role = :user, model: default_model, schema: nil, **params) ⇒ LLM::Response::Completion

Provides an interface to the chat completions API

Examples:

llm = LLM.openai(ENV["KEY"])
messages = [
  {role: "system", content: "Your task is to answer all of my questions"},
  {role: "system", content: "Your answers should be short and concise"},
]
res = llm.complete("Hello. What is the answer to 5 + 2 ?", :user, messages:)
print "[#{res.choices[0].role}]", res.choices[0].content, "\n"

Parameters:

  • prompt (String) —

    The input prompt to be completed

  • role (Symbol) (defaults to: :user) —

    The role of the prompt (e.g. :user, :system)

  • model (String) (defaults to: default_model) —

    The model to use for the completion

  • schema (#to_json, nil) (defaults to: nil) —

    The schema that describes the expected response format

  • params (Hash) —

    Other completion parameters

Returns:

Raises:

See Also:



77
78
79
80
81
82
83
84
85
86
 
# File 'lib/llm/providers/gemini.rb', line 77

def complete(prompt, role = :user, model: default_model, schema: nil, **params)
  model.respond_to?(:id) ? model.id : model
  path = ["/v1beta/models/#{model}", "generateContent?key=#{@secret}"].join(":")
  req  = Net::HTTP::Post.new(path, headers)
  messages = [*(params.delete(:messages) || []), LLM::Message.new(role, prompt)]
  body = JSON.dump({contents: format(messages)}.merge!(expand_schema(schema)))
  set_body_stream(req, StringIO.new(body))
  res = request(@http, req)
  Response::Completion.new(res).extend(response_parser)
end

#default_model ⇒ String

Returns the default model for chat completions

Returns:

  • (String)

See Also:



127
128
129
 
# File 'lib/llm/providers/gemini.rb', line 127

def default_model
  "gemini-1.5-flash"
end

#embed(input, model: "text-embedding-004", **params) ⇒ LLM::Response::Embedding

Provides an embedding

Parameters:

  • input (String, Array<String>) —

    The input to embed

  • model (String) (defaults to: "text-embedding-004") —

    The embedding model to use

  • params (Hash) —

    Other embedding parameters

Returns:

Raises:



55
56
57
58
59
60
61
62
 
# File 'lib/llm/providers/gemini.rb', line 55

def embed(input, model: "text-embedding-004", **params)
  model = model.respond_to?(:id) ? model.id : model
  path = ["/v1beta/models/#{model}", "embedContent?key=#{@secret}"].join(":")
  req = Net::HTTP::Post.new(path, headers)
  req.body = JSON.dump({content: {parts: [{text: input}]}})
  res = request(@http, req)
  Response::Embedding.new(res).extend(response_parser)
end

#files ⇒ Object

Provides an interface to Gemini’s file management API

See Also:



106
107
108
 
# File 'lib/llm/providers/gemini.rb', line 106

def files
  LLM::Gemini::Files.new(self)
end

#images ⇒ see LLM::Gemini::Images

Provides an interface to Gemini’s image generation API

Returns:

See Also:



99
100
101
 
# File 'lib/llm/providers/gemini.rb', line 99

def images
  LLM::Gemini::Images.new(self)
end

#models ⇒ Object

Provides an interface to Gemini’s models API

See Also:



113
114
115
 
# File 'lib/llm/providers/gemini.rb', line 113

def models
  LLM::Gemini::Models.new(self)
end