langchainrb 0.1.3 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1124e5c3ad459df63509c1682a713ecdf3952904df4685a3e0583c4e3c3a79dc
-  data.tar.gz: 0ebb4c9c0774101b32f2f679f95b1c99d425462407d7b28bd07273f9610c42fe
+  metadata.gz: c43d156af4f33487e653c0f72d0a27a327ebe07b1ecc7558409b8d877854318c
+  data.tar.gz: dba1509aa4494e1a35a08a40ca872bcb9f0b3ad03b7f17c6e8c3716c7d5ce805
 SHA512:
-  metadata.gz: f5c3fb67e7562df5b3e317f8f0c1d75e3948ce53812d10b5291236e4c068d8755781cc220add9eea686af189e77354b69daa45cb775353f3a3434dca1314920e
-  data.tar.gz: 0fb89fc2d4f1535b4af8ba5d3c9956da23ed3dd1174cb332f9c285838d9b1bfc3526552025ea1deca1b91e66275cef84b2c973c4b6118d3387074057ad192e67
+  metadata.gz: b800cf71cb8c9193082383d9994f1f40f61d7479e1db57ad970a089dd26e749be684f8f17b388a704aaba595909acb5babe9fa4ab37836953f0d11dbc55efa66
+  data.tar.gz: 9fdb6337593f3fa5902af104255a3e9f9a06214965b8dcb026d3b914094804b1d546c7b46c0e923d8327a216639ee4eb64e0febeafa9616c73d3706d74032037

data/CHANGELOG.md

@@ -1,5 +1,16 @@
 ## [Unreleased]
 
-## [0.1.0] - 2023-04-26
+## [0.2.0] - 2023-05-09
+
+- Prompt Templating
+  - Ability to create prompt templates and save them to JSON files
+  - Default `Prompt::FewShotPromptTemplate`
+  - New examples added to `examples/`
+
+## [0.1.4] - 2023-05-02
+
+- Backfilling missing specs
+
+## [0.1.3] - 2023-05-01
 
 - Initial release

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    langchainrb (0.1.3)
+    langchainrb (0.2.0)
       cohere-ruby (~> 0.9.1)
       milvus (~> 0.9.0)
       pinecone (~> 0.1.6)
@@ -111,6 +111,8 @@ GEM
     multipart-post (2.3.0)
     nokogiri (1.14.3-x86_64-darwin)
       racc (~> 1.4)
+    nokogiri (1.14.3-x86_64-linux)
+      racc (~> 1.4)
     pinecone (0.1.71)
       dry-struct (~> 1.6.0)
       dry-validation (~> 1.10.0)
@@ -167,6 +169,7 @@ GEM
 
 PLATFORMS
   x86_64-darwin-19
+  x86_64-linux
 
 DEPENDENCIES
   dotenv-rails (~> 2.7.6)

data/README.md

@@ -2,11 +2,13 @@
 --- 
 ⚡ Building applications with LLMs through composability ⚡
 
+👨‍💻👩‍💻 CURRENTLY SEEKING PEOPLE TO FORM THE CORE GROUP OF MAINTAINERS WITH
+
 :warning: UNDER ACTIVE AND RAPID DEVELOPMENT (MAY BE BUGGY AND UNTESTED)
 
 ![Tests status](https://github.com/andreibondarev/langchainrb/actions/workflows/ci.yml/badge.svg) [![Gem Version](https://badge.fury.io/rb/langchainrb.svg)](https://badge.fury.io/rb/langchainrb)
 
-Vectorsearch library is an abstraction layer on top of many popular vector search databases. It is a modern ORM that allows developers to easily chunk data, generate embeddings, store, search, query and retrieve data from vector search databases. Vectorsearch offers a straight-forward DSL and abstracts away overly complicated machine learning/data science-specific configurations and concepts
+Langchain.rb is a library that's an abstraction layer on top many emergent AI, ML and other DS tools. The goal is to abstract complexity and difficult concepts to make building AI/ML-supercharged applications approachable for traditional software engineers.
 
 ## Installation
 
@@ -112,12 +114,101 @@ cohere.embed(text: "foo bar")
 cohere.complete(prompt: "What is the meaning of life?")
 ```
 
+### Using Prompts
+
+#### Prompt Templates
+
+Create a prompt with one input variable:
+
+```ruby
+prompt = Prompt::PromptTemplate.new(template: "Tell me a {adjective} joke.", input_variables: ["adjective"])
+prompt.format(adjective: "funny") # "Tell me a funny joke."
+```
+
+Create a prompt with multiple input variables:
+
+```ruby
+prompt = Prompt::PromptTemplate.new(template: "Tell me a {adjective} joke about {content}.", input_variables: ["adjective", "content"])
+prompt.format(adjective: "funny", content: "chickens") # "Tell me a funny joke about chickens."
+```
+
+Creating a PromptTemplate using just a prompt and no input_variables:
+
+```ruby
+prompt = Prompt::PromptTemplate.from_template("Tell me a {adjective} joke about {content}.")
+prompt.input_variables # ["adjective", "content"]
+prompt.format(adjective: "funny", content: "chickens") # "Tell me a funny joke about chickens."
+```
+
+Save prompt template to JSON file:
+
+```ruby
+prompt.save(file_path: "spec/fixtures/prompt/prompt_template.json")
+```
+
+Loading a new prompt template using a JSON file:
+
+```ruby
+prompt = Prompt.load_from_path(file_path: "spec/fixtures/prompt/prompt_template.json")
+prompt.input_variables # ["adjective", "content"]
+```
+
+#### Few Shot Prompt Templates
+
+Create a prompt with a few shot examples:
+
+```ruby
+prompt = Prompt::FewShotPromptTemplate.new(
+  prefix: "Write antonyms for the following words.",
+  suffix: "Input: {adjective}\nOutput:",
+  example_prompt: Prompt::PromptTemplate.new(
+    input_variables: ["input", "output"],
+    template: "Input: {input}\nOutput: {output}"
+  ),
+  examples: [
+    { "input": "happy", "output": "sad" },
+    { "input": "tall", "output": "short" }
+  ],
+   input_variables: ["adjective"]
+)
+
+prompt.format(adjective: "good")
+
+# Write antonyms for the following words.
+#
+# Input: happy
+# Output: sad
+#
+# Input: tall
+# Output: short
+#
+# Input: good
+# Output:
+```
+
+Save prompt template to JSON file:
+
+```ruby
+prompt.save(file_path: "spec/fixtures/prompt/few_shot_prompt_template.json")
+```
+
+Loading a new prompt template using a JSON file:
+
+```ruby
+prompt = Prompt.load_from_path(file_path: "spec/fixtures/prompt/few_shot_prompt_template.json")
+prompt.prefix # "Write antonyms for the following words."
+```
+
 ## Development
 
 After checking out the repo, run `bin/setup` to install dependencies. Then, run `rake spec` to run the tests. You can also run `bin/console` for an interactive prompt that will allow you to experiment.
 
 To install this gem onto your local machine, run `bundle exec rake install`. To release a new version, update the version number in `version.rb`, and then run `bundle exec rake release`, which will create a git tag for the version, push git commits and the created tag, and push the `.gem` file to [rubygems.org](https://rubygems.org).
 
+## Honorary Contributors
+[<img style="border-radius:50%" alt="Andrei Bondarev" src="https://avatars.githubusercontent.com/u/541665?v=4" width="80" height="80" class="avatar">](https://github.com/andreibondarev)
+[<img style="border-radius:50%" alt="Rafael Figueiredo" src="https://avatars.githubusercontent.com/u/35845775?v=4" width="80" height="80" class="avatar">](https://github.com/rafaelqfigueiredo)
+
 ## Contributing
 
 Bug reports and pull requests are welcome on GitHub at https://github.com/andreibondarev/langchain.

data/examples/create_and_manage_few_shot_prompt_templates.rb

@@ -0,0 +1,36 @@
+require "langchain"
+
+# Create a prompt with a few shot examples
+prompt = Prompt::FewShotPromptTemplate.new(
+  prefix: "Write antonyms for the following words.",
+  suffix: "Input: {adjective}\nOutput:",
+  example_prompt: Prompt::PromptTemplate.new(
+    input_variables: ["input", "output"],
+    template: "Input: {input}\nOutput: {output}"
+  ),
+  examples: [
+    { "input": "happy", "output": "sad" },
+    { "input": "tall", "output": "short" }
+  ],
+   input_variables: ["adjective"]
+)
+
+prompt.format(adjective: "good")
+
+# Write antonyms for the following words.
+#
+# Input: happy
+# Output: sad
+#
+# Input: tall
+# Output: short
+#
+# Input: good
+# Output:
+
+# Save prompt template to JSON file
+prompt.save(file_path: "spec/fixtures/prompt/few_shot_prompt_template.json")
+
+# Loading a new prompt template using a JSON file
+prompt = Prompt.load_from_path(file_path: "spec/fixtures/prompt/few_shot_prompt_template.json")
+prompt.prefix # "Write antonyms for the following words."

data/examples/create_and_manage_prompt_templates.rb

@@ -0,0 +1,21 @@
+require "langchain"
+
+# Create a prompt with one input variable
+prompt = Prompt::PromptTemplate.new(template: "Tell me a {adjective} joke.", input_variables: ["adjective"])
+prompt.format(adjective: "funny") # "Tell me a funny joke."
+
+# Create a prompt with multiple input variables
+prompt = Prompt::PromptTemplate.new(template: "Tell me a {adjective} joke about {content}.", input_variables: ["adjective", "content"])
+prompt.format(adjective: "funny", content: "chickens") # "Tell me a funny joke about chickens."
+
+# Creating a PromptTemplate using just a prompt and no input_variables
+prompt = Prompt::PromptTemplate.from_template("Tell me a {adjective} joke about {content}.")
+prompt.input_variables # ["adjective", "content"]
+prompt.format(adjective: "funny", content: "chickens") # "Tell me a funny joke about chickens."
+
+# Save prompt template to JSON file
+prompt.save(file_path: "spec/fixtures/prompt/prompt_template.json")
+
+# Loading a new prompt template using a JSON file
+prompt = Prompt.load_from_path(file_path: "spec/fixtures/prompt/prompt_template.json")
+prompt.input_variables # ["adjective", "content"]

data/lib/langchain.rb

@@ -14,4 +14,12 @@ module LLM
   autoload :Base, "llm/base"
   autoload :Cohere, "llm/cohere"
   autoload :OpenAI, "llm/openai"
+end
+
+module Prompt
+  require_relative "prompt/loading"
+
+  autoload :Base, "prompt/base"
+  autoload :PromptTemplate, "prompt/prompt_template"
+  autoload :FewShotPromptTemplate, "prompt/few_shot_prompt_template"
 end

data/lib/prompt/base.rb

@@ -0,0 +1,86 @@
+# frozen_string_literal: true
+
+require 'strscan'
+
+module Prompt
+  class Base
+    def format(**kwargs)
+      raise NotImplementedError
+    end
+
+    def prompt_type
+      raise NotImplementedError
+    end
+
+    def to_h
+      raise NotImplementedError
+    end
+
+    #
+    # Validate the input variables against the template.
+    #
+    # @param template [String] The template to validate against.
+    # @param input_variables [Array<String>] The input variables to validate.
+    #
+    # @raise [ArgumentError] If there are missing or extra variables.
+    #
+    # @return [void]
+    #
+    def validate(template:, input_variables:)
+      input_variables_set = @input_variables.uniq
+      variables_from_template = Prompt::Base.extract_variables_from_template(template)
+
+      missing_variables = variables_from_template - input_variables_set
+      extra_variables = input_variables_set - variables_from_template
+
+      raise ArgumentError, "Missing variables: #{missing_variables}" if missing_variables.any?
+      raise ArgumentError, "Extra variables: #{extra_variables}" if extra_variables.any?
+    end
+
+    #
+    # Save the object to a file in JSON format.
+    #
+    # @param file_path [String, Pathname] The path to the file to save the object to
+    #
+    # @raise [ArgumentError] If file_path doesn't end with .json
+    #
+    # @return [void]
+    #
+    def save(file_path:)
+      save_path = file_path.is_a?(String) ? Pathname.new(file_path) : file_path
+      directory_path = save_path.dirname
+      FileUtils.mkdir_p(directory_path) unless directory_path.directory?
+
+      if save_path.extname == ".json"
+        File.open(file_path, "w") { |f| f.write(to_h.to_json) }
+      else
+        raise ArgumentError, "#{file_path} must be json"
+      end
+    end
+
+    private
+
+    #
+    # Extracts variables from a template string.
+    #
+    # This method takes a template string and returns an array of input variable names
+    # contained within the template. Input variables are defined as text enclosed in
+    # curly braces (e.g. "{variable_name}").
+    #
+    # @param template [String] The template string to extract variables from.
+    #
+    # @return [Array<String>] An array of input variable names.
+    #
+    def self.extract_variables_from_template(template)
+      input_variables = []
+      scanner = StringScanner.new(template)
+
+      while scanner.scan_until(/\{([^{}]*)\}/)
+        variable = scanner[1].strip
+        input_variables << variable unless variable.empty?
+      end
+
+      input_variables
+    end
+  end
+end

data/lib/prompt/few_shot_prompt_template.rb

@@ -0,0 +1,73 @@
+# frozen_string_literal: true
+
+module Prompt
+  class FewShotPromptTemplate < Base
+    attr_reader :examples, :example_prompt, :input_variables, :prefix, :suffix, :example_separator
+
+    #
+    # Initializes a new instance of the class.
+    #
+    # @param examples [Array<Hash>] Examples to format into the prompt.
+    # @param example_prompt [PromptTemplate] PromptTemplate used to format an individual example.
+    # @param suffix [String] A prompt template string to put after the examples.
+    # @param input_variables [Array<String>] A list of the names of the variables the prompt template expects.
+    # @param example_separator [String] String separator used to join the prefix, the examples, and suffix.
+    # @param prefix [String] A prompt template string to put before the examples.
+    # @param validate_template [Boolean] Whether or not to try validating the template.
+    #
+    def initialize(
+      examples:,
+      example_prompt:,
+      input_variables:,
+      suffix:,
+      prefix: "",
+      example_separator: "\n\n",
+      validate_template: true
+    )
+      @examples = examples
+      @example_prompt = example_prompt
+      @input_variables = input_variables
+      @prefix = prefix
+      @suffix = suffix
+      @example_separator = example_separator
+
+      validate(template: @prefix + @suffix, input_variables: @input_variables) if @validate_template
+    end
+
+    #
+    # Format the prompt with the inputs.
+    #
+    # @param kwargs [Hash] Any arguments to be passed to the prompt template.
+    #
+    # @return [String] A formatted string.
+    #
+    def format(**kwargs)
+      example_string = @examples.map { |example| @example_prompt.format(**example) }
+
+      suffix_string = @suffix
+      kwargs.each { |key, value| suffix_string = suffix_string.gsub(/\{#{key}\}/, value.to_s) }
+
+      [@prefix, *example_string, suffix_string].join(@example_separator)
+    end
+
+    #
+    # Returns the key type of prompt as a string.
+    #
+    # @return [String] the prompt type key
+    #
+    def prompt_type
+      "few_shot"
+    end
+
+    def to_h
+      {
+        _type: prompt_type,
+        input_variables: @input_variables,
+        prefix: @prefix,
+        example_prompt: @example_prompt.to_h,
+        examples: @examples,
+        suffix: @suffix
+      }
+    end
+  end
+end

data/lib/prompt/loading.rb

@@ -0,0 +1,87 @@
+# frozen_string_literal: true
+
+require 'strscan'
+require 'pathname'
+
+module Prompt
+  TYPE_TO_LOADER = {
+    "prompt" => ->(config) { Prompt.load_prompt(config) },
+    "few_shot" => ->(config) { Prompt.load_few_shot_prompt(config) }
+  }
+
+  class << self
+    #
+    # Load prompt from file.
+    #
+    # @param file_path [String, Pathname] The path of the file to read the configuration data from.
+    #
+    # @return [Object] The loaded prompt loaded.
+    #
+    # @raise [ArgumentError] If the file type of the specified file path is not supported.
+    #
+    def load_from_path(file_path:)
+      file_path = file_path.is_a?(String) ? Pathname.new(file_path) : file_path
+
+      if file_path.extname == ".json"
+        config = JSON.parse(File.read(file_path))
+      else
+        raise ArgumentError, "Got unsupported file type #{file_path.extname}"
+      end
+
+      load_from_config(config)
+    end
+
+    #
+    # Loads a prompt template with the given configuration.
+    #
+    # @param config [Hash] A hash containing the configuration for the prompt.
+    #
+    # @return [PromptTemplate] The loaded prompt loaded.
+    #
+    def load_prompt(config)
+      template, input_variables = config.values_at("template", "input_variables")
+      PromptTemplate.new(template: template, input_variables: input_variables)
+    end
+
+    #
+    # Loads a prompt template with the given configuration.
+    #
+    # @param config [Hash] A hash containing the configuration for the prompt.
+    #
+    # @return [FewShotPromptTemplate] The loaded prompt loaded.
+    #
+    def load_few_shot_prompt(config)
+      prefix, suffix, example_prompt, examples, input_variables = config.values_at("prefix", "suffix", "example_prompt", "examples", "input_variables")
+      example_prompt = load_prompt(example_prompt)
+      FewShotPromptTemplate.new(prefix: prefix, suffix: suffix, example_prompt: example_prompt, examples: examples, input_variables: input_variables)
+    end
+
+    private
+
+    #
+    # Loads the prompt from the given configuration hash
+    #
+    # @param config [Hash] the configuration hash to load from
+    #
+    # @return [Object] the loaded prompt
+    #
+    # @raise [ArgumentError] if the prompt type specified in the config is not supported
+    #
+    def load_from_config(config)
+      # If `_type` key is not present in the configuration hash, add it with a default value of `prompt`
+      unless config.key?("_type")
+        puts "[WARN] No `_type` key found, defaulting to `prompt`"
+        config["_type"] = "prompt"
+      end
+
+      # If the prompt type specified in the configuration hash is not supported, raise an exception
+      unless TYPE_TO_LOADER.key?(config["_type"])
+        raise ArgumentError, "Loading #{config["_type"]} prompt not supported"
+      end
+
+      # Load the prompt using the corresponding loader function from the `TYPE_TO_LOADER` hash
+      prompt_loader = TYPE_TO_LOADER[config["_type"]]
+      prompt_loader.call(config)
+    end
+  end
+end

data/lib/prompt/prompt_template.rb

@@ -0,0 +1,62 @@
+# frozen_string_literal: true
+
+module Prompt
+  class PromptTemplate < Base
+    attr_reader :template, :input_variables, :validate_template
+
+    #
+    # Initializes a new instance of the class.
+    #
+    # @param template [String] The prompt template.
+    # @param input_variables [Array<String>] A list of the names of the variables the prompt template expects.
+    # @param validate_template [Boolean] Whether or not to try validating the template.
+    #
+    def initialize(template:, input_variables:, validate_template: true)
+      @template = template
+      @input_variables = input_variables
+      @validate_template = validate_template
+
+      validate(template: @template, input_variables: @input_variables) if @validate_template
+    end
+
+    #
+    # Format the prompt with the inputs.
+    #
+    # @param kwargs [Hash] Any arguments to be passed to the prompt template.
+    # @return [String] A formatted string.
+    #
+    def format(**kwargs)
+      result = @template
+      kwargs.each { |key, value| result = result.gsub(/\{#{key}\}/, value.to_s) }
+      result
+    end
+
+    #
+    # Returns the key type of prompt as a string.
+    #
+    # @return [String] the prompt type key
+    #
+    def prompt_type
+      "prompt"
+    end
+
+    def to_h
+      {
+        _type: prompt_type,
+        input_variables: @input_variables,
+        template: @template
+      }
+    end
+
+    #
+    # Creates a new instance of the class using the given template.
+    #
+    # @param template [String] The template to use
+    #
+    # @return [Object] A new instance of the class
+    #
+    def self.from_template(template)
+      new(template: template, input_variables: extract_variables_from_template(template))
+    end
+  end
+end

data/lib/vectorsearch/base.rb

@@ -37,33 +37,30 @@ module Vectorsearch
       :generate_completion,
       :default_dimension
 
-    # def generate_embedding(text:)
-    #   llm_client.embed(text: text)
-    # end
-
-    # def generate_completion(prompt:)
-    #   llm_client.complete(prompt: prompt)
-    # end
-
-    # def default_dimension
-    #   llm_client.default_dimension
-    # end
-
     def generate_prompt(question:, context:)
-      "Context:\n" +
-      "#{context}\n" +
-      "---\n" +
-      "Question: #{question}\n" +
-      "---\n" +
-      "Answer:"
+      prompt_template = Prompt::FewShotPromptTemplate.new(
+        prefix: "Context:",
+        suffix: "---\nQuestion: {question}\n---\nAnswer:",
+        example_prompt: Prompt::PromptTemplate.new(
+          template: "{context}",
+          input_variables: ["context"]
+        ),
+        examples: [
+          { context: context }
+        ],
+        input_variables: ["question"],
+        example_separator: "\n"
+      )
+
+      prompt_template.format(question: question)
     end
 
     private
 
     def validate_llm!(llm:)
-      # TODO: Fix that this works is string `llm` value is passed in instead of symbol
+      # TODO: Fix so this works when `llm` value is a string instead of a symbol
       unless LLM::Base::LLMS.keys.include?(llm)
-        raise ArgumentError, "LLM must be one of #{LLMS}"
+        raise ArgumentError, "LLM must be one of #{LLM::Base::LLMS.keys}"
       end
     end
   end

data/lib/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Langchain
-  VERSION = "0.1.3"
+  VERSION = "0.2.0"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: langchainrb
 version: !ruby/object:Gem::Version
-  version: 0.1.3
+  version: 0.2.0
 platform: ruby
 authors:
 - Andrei Bondarev
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-05-01 00:00:00.000000000 Z
+date: 2023-05-09 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: pry-byebug
@@ -137,6 +137,8 @@ files:
 - README.md
 - Rakefile
 - examples/.keep
+- examples/create_and_manage_few_shot_prompt_templates.rb
+- examples/create_and_manage_prompt_templates.rb
 - examples/store_and_query_with_pinecone.rb
 - examples/store_and_query_with_qdrant.rb
 - examples/store_and_query_with_weaviate.rb
@@ -144,6 +146,10 @@ files:
 - lib/llm/base.rb
 - lib/llm/cohere.rb
 - lib/llm/openai.rb
+- lib/prompt/base.rb
+- lib/prompt/few_shot_prompt_template.rb
+- lib/prompt/loading.rb
+- lib/prompt/prompt_template.rb
 - lib/vectorsearch/base.rb
 - lib/vectorsearch/milvus.rb
 - lib/vectorsearch/pinecone.rb