Class: Langchain::LLM::AwsBedrock

Inherits:
Base
  • Object
show all
Defined in:
lib/langchain/llm/aws_bedrock.rb

Overview

LLM interface for Aws Bedrock APIs: docs.aws.amazon.com/bedrock/

Gem requirements:

gem 'aws-sdk-bedrockruntime', '~> 1.1'

Usage:

bedrock = Langchain::LLM::AwsBedrock.new(llm_options: {})

Constant Summary collapse

DEFAULTS = 
{
  completion_model_name: "anthropic.claude-v2",
  embedding_model_name: "amazon.titan-embed-text-v1",
  max_tokens_to_sample: 300,
  temperature: 1,
  top_k: 250,
  top_p: 0.999,
  stop_sequences: ["\n\nHuman:"],
  anthropic_version: "bedrock-2023-05-31",
  return_likelihoods: "NONE",
  count_penalty: {
    scale: 0,
    apply_to_whitespaces: false,
    apply_to_punctuations: false,
    apply_to_numbers: false,
    apply_to_stopwords: false,
    apply_to_emojis: false
  },
  presence_penalty: {
    scale: 0,
    apply_to_whitespaces: false,
    apply_to_punctuations: false,
    apply_to_numbers: false,
    apply_to_stopwords: false,
    apply_to_emojis: false
  },
  frequency_penalty: {
    scale: 0,
    apply_to_whitespaces: false,
    apply_to_punctuations: false,
    apply_to_numbers: false,
    apply_to_stopwords: false,
    apply_to_emojis: false
  }
}.freeze
SUPPORTED_COMPLETION_PROVIDERS = 
%i[anthropic cohere ai21].freeze
SUPPORTED_CHAT_COMPLETION_PROVIDERS = 
%i[anthropic].freeze
SUPPORTED_EMBEDDING_PROVIDERS = 
%i[amazon].freeze

Instance Attribute Summary collapse

Instance Method Summary collapse

Methods inherited from Base

#default_dimensions, #summarize

Methods included from DependencyHelper

#depends_on

Constructor Details

#initialize(completion_model: , embedding_model: , aws_client_options: {}, default_options: {}) ⇒ AwsBedrock

Returns a new instance of AwsBedrock.



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

def initialize(completion_model: DEFAULTS[:completion_model_name], embedding_model: DEFAULTS[:embedding_model_name], aws_client_options: {}, default_options: {})
  depends_on "aws-sdk-bedrockruntime", req: "aws-sdk-bedrockruntime"

  @client = ::Aws::BedrockRuntime::Client.new(**aws_client_options)
  @defaults = DEFAULTS.merge(default_options)
    .merge(completion_model_name: completion_model)
    .merge(embedding_model_name: embedding_model)
end

Instance Attribute Details

#client ⇒ Object (readonly)

Returns the value of attribute client.



49
50
51
 
# File 'lib/langchain/llm/aws_bedrock.rb', line 49

def client
  @client
end

#defaults ⇒ Object (readonly)

Returns the value of attribute defaults.



49
50
51
 
# File 'lib/langchain/llm/aws_bedrock.rb', line 49

def defaults
  @defaults
end

Instance Method Details

#chat(messages: [], system: nil, model: , max_tokens: , stop_sequences: nil, temperature: nil, top_p: nil, top_k: nil) ⇒ Langchain::LLM::AnthropicMessagesResponse

Generate a chat completion for a given prompt Currently only configured to work with the Anthropic provider and the claude-3 model family

Parameters:

  • messages (Array) (defaults to: []) —

    The messages to generate a completion for

  • system (String) (defaults to: nil) —

    The system prompt to provide instructions

  • model (String) (defaults to: ) —

    The model to use for completion defaults to @defaults

  • max_tokens (Integer) (defaults to: ) —

    The maximum number of tokens to generate

  • stop_sequences (Array) (defaults to: nil) —

    The stop sequences to use for completion

  • temperature (Float) (defaults to: nil) —

    The temperature to use for completion

  • top_p (Float) (defaults to: nil) —

    The top p to use for completion

  • top_k (Integer) (defaults to: nil) —

    The top k to use for completion

Returns:

  • (Langchain::LLM::AnthropicMessagesResponse) —

    Response object

Raises:

  • (ArgumentError) 


125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
 
# File 'lib/langchain/llm/aws_bedrock.rb', line 125

def chat(
  messages: [],
  system: nil,
  model: defaults[:completion_model_name],
  max_tokens: defaults[:max_tokens_to_sample],
  stop_sequences: nil,
  temperature: nil,
  top_p: nil,
  top_k: nil
)
  raise ArgumentError.new("messages argument is required") if messages.empty?

  raise "Model #{model} does not support chat completions." unless Langchain::LLM::AwsBedrock::SUPPORTED_CHAT_COMPLETION_PROVIDERS.include?(completion_provider)

  inference_parameters = {
    messages: messages,
    max_tokens: max_tokens,
    anthropic_version: @defaults[:anthropic_version]
  }
  inference_parameters[:system] = system if system
  inference_parameters[:stop_sequences] = stop_sequences if stop_sequences
  inference_parameters[:temperature] = temperature if temperature
  inference_parameters[:top_p] = top_p if top_p
  inference_parameters[:top_k] = top_k if top_k

  response = client.invoke_model({
    model_id: model,
    body: inference_parameters.to_json,
    content_type: "application/json",
    accept: "application/json"
  })

  parse_response response
end

#complete(prompt:, **params) ⇒ Langchain::LLM::AnthropicResponse

Generate a completion for a given prompt

Parameters:

  • prompt (String) —

    The prompt to generate a completion for

  • params —

    extra parameters passed to Aws::BedrockRuntime::Client#invoke_model

Returns:



94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
 
# File 'lib/langchain/llm/aws_bedrock.rb', line 94

def complete(prompt:, **params)
  raise "Completion provider #{completion_provider} is not supported." unless SUPPORTED_COMPLETION_PROVIDERS.include?(completion_provider)

  raise "Model #{@defaults[:completion_model_name]} only supports #chat." if @defaults[:completion_model_name].include?("claude-3")

  parameters = compose_parameters params

  parameters[:prompt] = wrap_prompt prompt

  response = client.invoke_model({
    model_id: @defaults[:completion_model_name],
    body: parameters.to_json,
    content_type: "application/json",
    accept: "application/json"
  })

  parse_response response
end

#embed(text:, **params) ⇒ Langchain::LLM::AwsTitanResponse

Generate an embedding for a given text

Parameters:

  • text (String) —

    The text to generate an embedding for

  • params —

    extra parameters passed to Aws::BedrockRuntime::Client#invoke_model

Returns:



71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
 
# File 'lib/langchain/llm/aws_bedrock.rb', line 71

def embed(text:, **params)
  raise "Completion provider #{embedding_provider} is not supported." unless SUPPORTED_EMBEDDING_PROVIDERS.include?(embedding_provider)

  parameters = {inputText: text}
  parameters = parameters.merge(params)

  response = client.invoke_model({
    model_id: @defaults[:embedding_model_name],
    body: parameters.to_json,
    content_type: "application/json",
    accept: "application/json"
  })

  Langchain::LLM::AwsTitanResponse.new(JSON.parse(response.body.string))
end