langchainrb 0.4.1 → 0.5.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 100c9989f1a6787168064f6d810f009c6ded8129949f0040a7c1ed05ebff9d20
-  data.tar.gz: 0cf01f053841e9a45353d72b62df9a29c0daf7268fb69aff794aae830f7db448
+  metadata.gz: b0a2fe8026e861c9d97465bce7da08a0b077492d6f7cf8fb42c45dbfdfe6749f
+  data.tar.gz: c04099c44a847bd9c05e8594859f92ca1f54d338c463ce59a375c2cb9731b1ad
 SHA512:
-  metadata.gz: becea5f089e7455e8dd9437cb3647c921eebb3155a9685697190e0e757d31ec8d9f4572faa3e299911b44b371c097bd9b5edf3da35772562027d1b12cd743b9e
-  data.tar.gz: 44cfe6f8cb2056d6a0a5c619ded7828a9cdad0e5f4e8876242adad2f13251ea1c3dbb64c9feffea25b61e91f163245954c1f272fadb9f4f3a2a06cadd64bd9a0
+  metadata.gz: dec375b2b7cae377cf31f3f8ed0a6ac9d79215c945e7c0da78ed1fbad3c502ecfcc5ce5318c55a9a634db88fea1f5fbbeed7a0f7dc6ab8096c909e0a3ff02154
+  data.tar.gz: 558a0f6ddf90ad044f9e2cc7c6ca678958472748d2e430a3cbb4308290898b9b22a204a94a5b5943f06137f7da5238d038a65b8c79d96f8a3705499d95cfb597

data/.env.example

@@ -1,3 +1,4 @@
+AI21_API_KEY=
 CHROMA_URL=
 COHERE_API_KEY=
 HUGGING_FACE_API_KEY=
@@ -6,10 +7,10 @@ OPENAI_API_KEY=
 GOOGLE_PALM_API_KEY=
 PINECONE_API_KEY=
 PINECONE_ENVIRONMENT=
+POSTGRES_URL=
 REPLICATE_API_KEY=
 QDRANT_API_KEY=
 QDRANT_URL=
 SERPAPI_API_KEY=
 WEAVIATE_API_KEY=
 WEAVIATE_URL=
-POSTGRES_URL=

data/.rubocop.yml

@@ -0,0 +1,11 @@
+AllCops:
+  SuggestExtensions: false
+
+require:
+  - standard
+  - rubocop-performance
+
+inherit_gem:
+  standard: config/base.yml
+  standard-performance: config/base.yml
+  standard-custom: config/base.yml

data/CHANGELOG.md

@@ -1,5 +1,18 @@
 ## [Unreleased]
 
+## [0.5.0] - 2023-06-05
+- [BREAKING] LLMs are now passed as objects to Vectorsearch classes instead of `llm: :name, llm_api_key:` previously
+- 📋 Prompts
+  - YAML prompt templates are now supported
+- 🚚 Loaders
+  - Introduce `Langchain::Processors::Xlsx` to parse .xlsx files
+
+## [0.4.2] - 2023-06-03
+- 🗣️ LLMs
+  - Introducing support for AI21
+- Better docs generation
+- Refactors
+
 ## [0.4.1] - 2023-06-02
 - Beautiful colored log messages
 - 🛠️ Tools

data/Gemfile

@@ -10,3 +10,5 @@ gem "rake", "~> 13.0"
 gem "rspec", "~> 3.0"
 
 gem "standardrb"
+# Lets add rubocop explicitly here, we are using only standardrb rules in .rubocop.yml
+gem "rubocop"

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    langchainrb (0.4.1)
+    langchainrb (0.5.0)
       colorize (~> 0.8.1)
       tiktoken_ruby (~> 0.0.5)
 
@@ -30,6 +30,7 @@ GEM
     addressable (2.8.4)
       public_suffix (>= 2.0.2, < 6.0)
     afm (0.2.2)
+    ai21 (0.2.0)
     ast (2.4.2)
     builder (3.2.4)
     byebug (11.1.3)
@@ -210,6 +211,7 @@ GEM
     rainbow (3.1.1)
     rake (13.0.6)
     rb_sys (0.9.78)
+    rdiscount (2.2.7)
     regexp_parser (2.8.0)
     replicate-ruby (0.2.2)
       addressable
@@ -217,6 +219,9 @@ GEM
       faraday-multipart
       faraday-retry
     rexml (3.2.5)
+    roo (2.10.0)
+      nokogiri (~> 1)
+      rubyzip (>= 1.3.0, < 3.0.0)
     rspec (3.12.0)
       rspec-core (~> 3.12.0)
       rspec-expectations (~> 3.12.0)
@@ -255,6 +260,7 @@ GEM
     rubyzip (2.3.2)
     safe_ruby (1.0.4)
       childprocess (>= 0.3.9)
+    sequel (5.68.0)
     standard (1.28.2)
       language_server-protocol (~> 3.17.0.2)
       lint_roller (~> 1.0)
@@ -286,6 +292,7 @@ GEM
       graphlient (~> 0.6.0)
     wikipedia-client (1.17.0)
       addressable (~> 2.7)
+    yard (0.9.34)
     zeitwerk (2.6.8)
 
 PLATFORMS
@@ -297,6 +304,7 @@ PLATFORMS
   x86_64-linux
 
 DEPENDENCIES
+  ai21 (~> 0.2.0)
   chroma-db (~> 0.3.0)
   cohere-ruby (~> 0.9.4)
   docx (~> 0.8.0)
@@ -315,13 +323,18 @@ DEPENDENCIES
   pry-byebug (~> 3.10.0)
   qdrant-ruby (~> 0.9.0)
   rake (~> 13.0)
+  rdiscount
   replicate-ruby (~> 0.2.2)
+  roo (~> 2.10.0)
   rspec (~> 3.0)
+  rubocop
   ruby-openai (~> 4.0.0)
   safe_ruby (~> 1.0.4)
+  sequel (~> 5.68.0)
   standardrb
   weaviate-ruby (~> 0.8.0)
   wikipedia-client (~> 1.17.0)
+  yard
 
 BUNDLED WITH
    2.4.0

data/README.md

@@ -47,8 +47,7 @@ Pick the vector search database you'll be using and instantiate the client:
 client = Langchain::Vectorsearch::Weaviate.new(
     url: ENV["WEAVIATE_URL"],
     api_key: ENV["WEAVIATE_API_KEY"],
-    llm: :openai, # or :cohere
-    llm_api_key: ENV["OPENAI_API_KEY"]
+    llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"])
 )
 
 # You can instantiate any other supported vector search database:
@@ -109,6 +108,10 @@ Add `gem "ruby-openai", "~> 4.0.0"` to your Gemfile.
 ```ruby
 openai = Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"])
 ```
+You can pass additional parameters to the constructor, it will be passed to the OpenAI client:
+```ruby
+openai = Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"], llm_options: {uri_base: "http://localhost:1234"}) )
+```
 ```ruby
 openai.embed(text: "foo bar")
 ```
@@ -147,6 +150,12 @@ Add `"google_palm_api", "~> 0.1.0"` to your Gemfile.
 google_palm = Langchain::LLM::GooglePalm.new(api_key: ENV["GOOGLE_PALM_API_KEY"])
 ```
 
+#### AI21
+Add `gem "ai21", "~> 0.2.0"` to your Gemfile.
+```ruby
+ai21 = Langchain::LLM::AI21.new(api_key: ENV["AI21_API_KEY"])
+```
+
 ### Using Prompts 📋
 
 #### Prompt Templates
@@ -168,9 +177,9 @@ prompt.format(adjective: "funny", content: "chickens") # "Tell me a funny joke a
 Creating a PromptTemplate using just a prompt and no input_variables:
 
 ```ruby
-prompt = Langchain::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."
+prompt = Langchain::Prompt::PromptTemplate.from_template("Tell me a funny joke about chickens.")
+prompt.input_variables # []
+prompt.format # "Tell me a funny joke about chickens."
 ```
 
 Save prompt template to JSON file:
@@ -232,6 +241,13 @@ prompt = Langchain::Prompt.load_from_path(file_path: "spec/fixtures/prompt/few_s
 prompt.prefix # "Write antonyms for the following words."
 ```
 
+Loading a new prompt template using a YAML file:
+
+```ruby
+prompt = Langchain::Prompt.load_from_path(file_path: "spec/fixtures/prompt/prompt_template.yaml")
+prompt.input_variables #=> ["adjective", "content"]
+```
+
 ### Using Agents 🤖
 Agents are semi-autonomous bots that can respond to user questions and use available to them Tools to provide informed replies. They break down problems into series of steps and define Actions (and Action Inputs) along the way that are executed and fed back to them as additional information. Once an Agent decides that it has the Final Answer it responds with it.
 
@@ -240,7 +256,7 @@ Agents are semi-autonomous bots that can respond to user questions and use avail
 Add `gem "ruby-openai"`, `gem "eqn"`, and `gem "google_search_results"` to your Gemfile
 
 ```ruby
-agent = Langchain::Agent::ChainOfThoughtAgent.new(llm: :openai, llm_api_key: ENV["OPENAI_API_KEY"], tools: ['search', 'calculator'])
+agent = Langchain::Agent::ChainOfThoughtAgent.new(llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"]), tools: ['search', 'calculator'])
 
 agent.tools
 # => ["search", "calculator"]
@@ -250,6 +266,19 @@ agent.run(question: "How many full soccer fields would be needed to cover the di
 #=> "Approximately 2,945 soccer fields would be needed to cover the distance between NYC and DC in a straight line."
 ```
 
+#### SQL-Query Agent
+
+Add `gem "sequel"` to your Gemfile
+
+```ruby
+agent = Langchain::Agent::SQLQueryAgent.new(llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"]), db_connection_string: "postgres://user:password@localhost:5432/db_name")
+
+```
+```ruby
+agent.ask(question: "How many users have a name with length greater than 5 in the users table?")
+#=> "14 users have a name with length greater than 5 in the users table."
+```
+
 #### Demo
 ![May-12-2023 13-09-13](https://github.com/andreibondarev/langchainrb/assets/541665/6bad4cd9-976c-420f-9cf9-b85bf84f7eaf)
 
@@ -260,6 +289,7 @@ agent.run(question: "How many full soccer fields would be needed to cover the di
 | Name         | Description                                        | ENV Requirements                                              | Gem Requirements                          |
 | ------------ | :------------------------------------------------: | :-----------------------------------------------------------: | :---------------------------------------: |
 | "calculator" | Useful for getting the result of a math expression |                                                               | `gem "eqn", "~> 1.6.5"`                   |
+| "database"   | Useful for querying a SQL database |                                                               | `gem "sequel", "~> 5.68.0"`                   |
 | "ruby_code_interpreter" | Interprets Ruby expressions             |                                                               | `gem "safe_ruby", "~> 1.0.4"`             |
 | "search"     | A wrapper around Google Search                     | `ENV["SERPAPI_API_KEY"]` (https://serpapi.com/manage-api-key) | `gem "google_search_results", "~> 2.0.0"` |
 | "wikipedia"  | Calls Wikipedia API to retrieve the summary        |                                                               | `gem "wikipedia-client", "~> 1.17.0"`     |
@@ -294,13 +324,14 @@ Langchain::Loader.load('https://www.example.com/file.pdf')
 | JSON   | Langchain::Processors::JSON  |                              |
 | JSONL  | Langchain::Processors::JSONL |                              |
 | csv    | Langchain::Processors::CSV   |                              |
+| xlsx   | Langchain::Processors::Xlsx  |   `gem "roo", "~> 2.10.0"`   |
 
 ## Examples
 Additional examples available: [/examples](https://github.com/andreibondarev/langchainrb/tree/main/examples)
 
 ## Logging
 
-LangChain.rb uses standard logging mechanisms and defaults to `:debug` level. Most messages are at info level, but we will add debug or warn statements as needed.
+LangChain.rb uses standard logging mechanisms and defaults to `:warn` level. Most messages are at info level, but we will add debug or warn statements as needed.
 To show all log messages:
 
 ```ruby
@@ -313,6 +344,10 @@ Langchain.logger.level = :info
 2. `cp .env.example .env`, then fill out the environment variables in `.env`
 3. `bundle exec rake` to ensure that the tests pass and to run standardrb
 4. `bin/console` to load the gem in a REPL session. Feel free to add your own instances of LLMs, Tools, Agents, etc. and experiment with them.
+5. Optionally, install lefthook git hooks for pre-commit to auto lint: `gem install lefthook && lefthook install -f`
+
+## Community
+Join us in the [Ruby AI Builders](https://discord.gg/SBmjAnKT) Discord community in #langchainrb
 
 ## Core 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)

data/Rakefile

@@ -3,6 +3,7 @@
 require "bundler/gem_tasks"
 require "rspec/core/rake_task"
 require "standard/rake"
+require "yard"
 
 RSpec::Core::RakeTask.new(:spec)
 
@@ -11,3 +12,7 @@ task default: :spec
 Rake::Task["spec"].enhance do
   Rake::Task["standard:fix"].invoke
 end
+
+YARD::Rake::YardocTask.new do |t|
+  t.options = ["--fail-on-warning"]
+end

data/examples/pdf_store_and_query_with_chroma.rb

@@ -7,8 +7,7 @@ require "langchain"
 chroma = Vectorsearch::Chroma.new(
   url: ENV["CHROMA_URL"],
   index_name: "documents",
-  llm: :openai,
-  llm_api_key: ENV["OPENAI_API_KEY"]
+  llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"])
 )
 
 # Create the default schema.

data/examples/store_and_query_with_pinecone.rb

@@ -8,8 +8,7 @@ pinecone = Vectorsearch::Pinecone.new(
   environment: ENV["PINECONE_ENVIRONMENT"],
   api_key: ENV["PINECONE_API_KEY"],
   index_name: "recipes",
-  llm: :openai,
-  llm_api_key: ENV["OPENAI_API_KEY"]
+  llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"])
 )
 
 # Create the default schema.

data/examples/store_and_query_with_qdrant.rb

@@ -8,8 +8,7 @@ qdrant = Vectorsearch::Qdrant.new(
   url: ENV["QDRANT_URL"],
   api_key: ENV["QDRANT_API_KEY"],
   index_name: "recipes",
-  llm: :cohere,
-  llm_api_key: ENV["COHERE_API_KEY"]
+  llm: Langchain::LLM::Cohere.new(api_key: ENV["COHERE_API_KEY"])
 )
 
 # Create the default schema.

data/examples/store_and_query_with_weaviate.rb

@@ -8,8 +8,7 @@ weaviate = Vectorsearch::Weaviate.new(
   url: ENV["WEAVIATE_URL"],
   api_key: ENV["WEAVIATE_API_KEY"],
   index_name: "Recipes",
-  llm: :openai,
-  llm_api_key: ENV["OPENAI_API_KEY"]
+  llm: Langchain::LLM::OpenAI.new(api_key: ENV["OPENAI_API_KEY"])
 )
 
 # Create the default schema. A text field `content` will be used.

data/lefthook.yml

@@ -0,0 +1,5 @@
+pre-commit:
+  commands:
+    lint:
+      run: standardrb --fix
+      stage_fixed: true

data/lib/langchain/agent/chain_of_thought_agent/chain_of_thought_agent.rb

@@ -2,23 +2,19 @@
 
 module Langchain::Agent
   class ChainOfThoughtAgent < Base
-    attr_reader :llm, :llm_api_key, :llm_client, :tools
+    attr_reader :llm, :tools
 
     # Initializes the Agent
     #
-    # @param llm [Symbol] The LLM to use
-    # @param llm_api_key [String] The API key for the LLM
+    # @param llm [Object] The LLM client to use
     # @param tools [Array] The tools to use
     # @return [ChainOfThoughtAgent] The Agent::ChainOfThoughtAgent instance
-    def initialize(llm:, llm_api_key:, tools: [])
-      Langchain::LLM::Base.validate_llm!(llm: llm)
+    def initialize(llm:, tools: [])
       Langchain::Tool::Base.validate_tools!(tools: tools)
 
-      @llm = llm
-      @llm_api_key = llm_api_key
       @tools = tools
 
-      @llm_client = Langchain::LLM.const_get(Langchain::LLM::Base::LLMS.fetch(llm)).new(api_key: llm_api_key)
+      @llm = llm
     end
 
     # Validate tools when they're re-assigned
@@ -42,8 +38,8 @@ module Langchain::Agent
       )
 
       loop do
-        Langchain.logger.info("[#{self.class.name}]".red + ": Sending the prompt to the #{llm} LLM")
-        response = llm_client.complete(
+        Langchain.logger.info("[#{self.class.name}]".red + ": Sending the prompt to the #{llm.class} LLM")
+        response = llm.complete(
           prompt: prompt,
           stop_sequences: ["Observation:"],
           max_tokens: 500

data/lib/langchain/agent/sql_query_agent/sql_query_agent.rb

@@ -0,0 +1,78 @@
+# frozen_string_literal: true
+
+module Langchain::Agent
+  class SQLQueryAgent < Base
+    attr_reader :llm, :db, :schema
+
+    # Initializes the Agent
+    #
+    # @param llm [Object] The LLM client to use
+    # @param db_connection_string [String] Database connection info
+    def initialize(llm:, db_connection_string:)
+      @llm = llm
+      @db = Langchain::Tool::Database.new(db_connection_string)
+      @schema = @db.schema
+    end
+
+    # Ask a question and get an answer
+    #
+    # @param question [String] Question to ask the LLM/Database
+    # @return [String] Answer to the question
+    def ask(question:)
+      prompt = create_prompt_for_sql(question: question)
+
+      # Get the SQL string to execute
+      Langchain.logger.info("[#{self.class.name}]".red + ":  Passing the inital prompt to the #{llm.class} LLM")
+      sql_string = llm.complete(prompt: prompt, max_tokens: 500)
+
+      # Execute the SQL string and collect the results
+      Langchain.logger.info("[#{self.class.name}]".red + ":  Passing the SQL to the Database: #{sql_string}")
+      results = db.execute(input: sql_string)
+
+      # Pass the results and get the LLM to synthesize the answer to the question
+      Langchain.logger.info("[#{self.class.name}]".red + ":  Passing the synthesize prompt to the #{llm.class} LLM with results: #{results}")
+      prompt2 = create_prompt_for_answer(question: question, sql_query: sql_string, results: results)
+      llm.complete(prompt: prompt2, max_tokens: 500)
+    end
+
+    private
+
+    # Create the initial prompt to pass to the LLM
+    # @param question[String] Question to ask
+    # @return [String] Prompt
+    def create_prompt_for_sql(question:)
+      prompt_template_sql.format(
+        dialect: "standard SQL",
+        schema: schema,
+        question: question
+      )
+    end
+
+    # Load the PromptTemplate from the JSON file
+    # @return [PromptTemplate] PromptTemplate instance
+    def prompt_template_sql
+      Langchain::Prompt.load_from_path(
+        file_path: Langchain.root.join("langchain/agent/sql_query_agent/sql_query_agent_sql_prompt.json")
+      )
+    end
+
+    # Create the second prompt to pass to the LLM
+    # @param question [String] Question to ask
+    # @return [String] Prompt
+    def create_prompt_for_answer(question:, sql_query:, results:)
+      prompt_template_answer.format(
+        question: question,
+        sql_query: sql_query,
+        results: results
+      )
+    end
+
+    # Load the PromptTemplate from the JSON file
+    # @return [PromptTemplate] PromptTemplate instance
+    def prompt_template_answer
+      Langchain::Prompt.load_from_path(
+        file_path: Langchain.root.join("langchain/agent/sql_query_agent/sql_query_agent_answer_prompt.json")
+      )
+    end
+  end
+end

data/lib/langchain/agent/sql_query_agent/sql_query_agent_answer_prompt.json

@@ -0,0 +1,10 @@
+{
+  "_type": "prompt",
+  "template":
+    "Given an input question and results of a SQL query, look at the results and return the answer. Use the following format:\nQuestion: {question}\nThe SQL query: {sql_query}\nResult of the SQLQuery: {results}\nFinal answer: Final answer here",
+  "input_variables": [
+    "question",
+    "sql_query",
+    "results"
+  ]
+}

data/lib/langchain/agent/sql_query_agent/sql_query_agent_sql_prompt.json

@@ -0,0 +1,10 @@
+{
+  "_type": "prompt",
+   "template":
+    "Given an input question, create a syntactically correct {dialect} query to run, then return the query in valid SQL.\nNever query for all the columns from a specific table, only ask for a the few relevant columns given the question.\nPay attention to use only the column names that you can see in the schema description. Be careful to not query for columns that do not exist. Pay attention to which column is in which table. Also, qualify column names with the table name when needed.\nOnly use the tables listed below.\n{schema}\nUse the following format:\nQuestion: {question}\nSQLQuery:",
+  "input_variables": [
+    "dialect",
+    "schema",
+    "question"
+  ]
+}

data/lib/langchain/dependency_helper.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module Langchain
+  module DependencyHelper
+    class VersionError < ScriptError; end
+
+    # This method requires and loads the given gem, and then checks to see if the version of the gem meets the requirements listed in `langchain.gemspec`
+    # This solution was built to avoid auto-loading every single gem in the Gemfile when the developer will mostly likely be only using a few of them.
+    #
+    # @param gem_name [String] The name of the gem to load
+    # @return [Boolean] Whether or not the gem was loaded successfully
+    # @raise [LoadError] If the gem is not installed
+    # @raise [VersionError] If the gem is installed, but the version does not meet the requirements
+    #
+    def depends_on(gem_name)
+      gem(gem_name) # require the gem
+
+      return(true) unless defined?(Bundler) # If we're in a non-bundler environment, we're no longer able to determine if we'll meet requirements
+
+      gem_version = Gem.loaded_specs[gem_name].version
+      gem_requirement = Bundler.load.dependencies.find { |g| g.name == gem_name }&.requirement
+
+      raise LoadError unless gem_requirement
+
+      unless gem_requirement.satisfied_by?(gem_version)
+        raise VersionError, "The #{gem_name} gem is installed, but version #{gem_requirement} is required. You have #{gem_version}."
+      end
+
+      true
+    rescue LoadError
+      raise LoadError, "Could not load #{gem_name}. Please ensure that the #{gem_name} gem is installed."
+    end
+  end
+end

data/lib/langchain/llm/ai21.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+module Langchain::LLM
+  class AI21 < Base
+    #
+    # Wrapper around AI21 Studio APIs.
+    #
+    # Gem requirements: gem "ai21", "~> 0.2.0"
+    #
+    # Usage:
+    # ai21 = Langchain::LLM::AI21.new(api_key:)
+    #
+
+    def initialize(api_key:)
+      depends_on "ai21"
+      require "ai21"
+
+      @client = ::AI21::Client.new(api_key)
+    end
+
+    #
+    # Generate a completion for a given prompt
+    #
+    # @param prompt [String] The prompt to generate a completion for
+    # @param params [Hash] The parameters to pass to the API
+    # @return [String] The completion
+    #
+    def complete(prompt:, **params)
+      response = client.complete(prompt, params)
+      response.dig(:completions, 0, :data, :text)
+    end
+
+    #
+    # Generate a summary for a given text
+    #
+    # @param text [String] The text to generate a summary for
+    # @param params [Hash] The parameters to pass to the API
+    # @return [String] The summary
+    #
+    def summarize(text:, **params)
+      response = client.summarize(text, "TEXT", params)
+      response.dig(:summary)
+    end
+  end
+end

data/lib/langchain/llm/base.rb

@@ -2,17 +2,9 @@
 
 module Langchain::LLM
   class Base
-    attr_reader :client
+    include Langchain::DependencyHelper
 
-    # Currently supported LLMs
-    # TODO: Add support for HuggingFace and other LLMs
-    LLMS = {
-      cohere: "Cohere",
-      google_palm: "GooglePalm",
-      huggingface: "HuggingFace",
-      openai: "OpenAI",
-      replicate: "Replicate"
-    }.freeze
+    attr_reader :client
 
     def default_dimension
       self.class.const_get(:DEFAULTS).dig(:dimension)
@@ -37,14 +29,5 @@ module Langchain::LLM
     def summarize(...)
       raise NotImplementedError, "#{self.class.name} does not support summarization"
     end
-
-    # Ensure that the LLM value passed in is supported
-    # @param llm [Symbol] The LLM to use
-    def self.validate_llm!(llm:)
-      # TODO: Fix so this works when `llm` value is a string instead of a symbol
-      unless Langchain::LLM::Base::LLMS.key?(llm)
-        raise ArgumentError, "LLM must be one of #{Langchain::LLM::Base::LLMS.keys}"
-      end
-    end
   end
 end

data/lib/langchain/llm/cohere.rb

@@ -2,6 +2,15 @@
 
 module Langchain::LLM
   class Cohere < Base
+    #
+    # Wrapper around the Cohere API.
+    #
+    # Gem requirements: gem "cohere-ruby", "~> 0.9.4"
+    #
+    # Usage:
+    # cohere = Langchain::LLM::Cohere.new(api_key: "YOUR_API_KEY")
+    #
+
     DEFAULTS = {
       temperature: 0.0,
       completion_model_name: "base",

data/lib/langchain/llm/google_palm.rb

@@ -2,7 +2,14 @@
 
 module Langchain::LLM
   class GooglePalm < Base
+    #
     # Wrapper around the Google PaLM (Pathways Language Model) APIs.
+    #
+    # Gem requirements: gem "google_palm_api", "~> 0.1.0"
+    #
+    # Usage:
+    # google_palm = Langchain::LLM::GooglePalm.new(api_key: "YOUR_API_KEY")
+    #
 
     DEFAULTS = {
       temperature: 0.0,

data/lib/langchain/llm/hugging_face.rb

@@ -2,6 +2,15 @@
 
 module Langchain::LLM
   class HuggingFace < Base
+    #
+    # Wrapper around the HuggingFace Inference API.
+    #
+    # Gem requirements: gem "hugging-face", "~> 0.3.4"
+    #
+    # Usage:
+    # hf = Langchain::LLM::HuggingFace.new(api_key: "YOUR_API_KEY")
+    #
+
     # The gem does not currently accept other models:
     # https://github.com/alchaplinsky/hugging-face/blob/main/lib/hugging_face/inference_api.rb#L32-L34
     DEFAULTS = {