ruby_llm 0.1.0.pre16 → 0.1.0.pre18

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 85ade36a78c5038ed31c362682ee1e81cba3b952e278ce092c93d3695740e2a1
-  data.tar.gz: 572e88e7f493f9d125873c9a3ba07d014302c480246d3b7b0b326e1d44924f30
+  metadata.gz: cb34b8e16d243dfdc95d7fc6a70a67ecb2afac895da5a646e57dbc915da0a8e9
+  data.tar.gz: 6e95fb39b3670a38422b0dd79ec97263c34e6143751fef2272a6401ccac79c95
 SHA512:
-  metadata.gz: 370a6e32a748d7119a3af729e0a43cdb9f4b80d60e09d959e425a28b0c4ecad8ca9e3a1f3aec6ba279039e40b2129c49e56641e041622f3268c72624d17ebc9d
-  data.tar.gz: 2295e3a850d514eb60964d3bdc47ee9682ba8b630010661efe8374ca468a81edb10617a10484a70ba47df2db8547856246c470160e8d87c157bdaccae2c8b5c6
+  metadata.gz: af33908a0ff937c4a7d0b074b4406c4211456dbb92c39d5c2237ca6463f3bd8129141a4b2c23270da85fbf90d26c0e8c6b295aaedb8574a23e0ab3eb107abd39
+  data.tar.gz: c6ce63d8d361fffcfaf6bfa2fbdea1df42ddf75e2969d68c5f789fbb14fccb61284c9a6b46fb0c0472c0e18cb31da150544b3cf2e9af4a1e53039a164b4d3e9f

data/README.md

@@ -76,6 +76,33 @@ last_message = chat.messages.last
 puts "Conversation used #{last_message.input_tokens} input tokens and #{last_message.output_tokens} output tokens"
 ```
 
+## Text Embeddings
+
+Need vector embeddings for your text? RubyLLM makes it simple:
+
+```ruby
+# Get embeddings with the default model
+vector = RubyLLM.embed "Hello, world!"
+
+# Use a specific model
+vector = RubyLLM.embed(
+  "Ruby is awesome!",
+  model: "text-embedding-3-large"
+)
+
+# Process multiple texts at once
+vectors = RubyLLM.embed([
+  "First document",
+  "Second document",
+  "Third document"
+])
+
+# Configure the default model
+RubyLLM.configure do |config|
+  config.default_embedding_model = 'text-embedding-3-large'
+end
+```
+
 ## Using Tools
 
 Give your AI assistants access to your Ruby code by creating tool classes that do one thing well:

data/lib/ruby_llm/chat.rb

@@ -97,8 +97,10 @@ module RubyLLM
 
     def handle_tool_calls(response, &block)
       response.tool_calls.each_value do |tool_call|
+        @on[:new_message]&.call
         result = execute_tool tool_call
-        add_tool_result tool_call.id, result if result
+        message = add_tool_result tool_call.id, result
+        @on[:end_message]&.call(message)
       end
 
       complete(&block)

data/lib/ruby_llm/configuration.rb

@@ -10,11 +10,16 @@ module RubyLLM
   #     config.anthropic_api_key = ENV['ANTHROPIC_API_KEY']
   #   end
   class Configuration
-    attr_accessor :openai_api_key, :anthropic_api_key, :default_model, :request_timeout
+    attr_accessor :openai_api_key,
+                  :anthropic_api_key,
+                  :default_model,
+                  :default_embedding_model,
+                  :request_timeout
 
     def initialize
       @request_timeout = 30
       @default_model = 'gpt-4o-mini'
+      @default_embedding_model = 'text-embedding-3-small'
     end
   end
 end

data/lib/ruby_llm/embedding.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  # Core embedding interface. Provides a clean way to generate embeddings
+  # from text using various provider models.
+  module Embedding
+    module_function
+
+    def embed(text, model: nil)
+      model_id = model || RubyLLM.config.default_embedding_model
+      Models.find(model_id)
+
+      provider = Provider.for(model_id)
+      provider.embed(text, model: model_id)
+    end
+  end
+end

data/lib/ruby_llm/provider.rb

@@ -30,17 +30,23 @@ module RubyLLM
         parse_list_models_response response
       end
 
+      def embed(text, model:)
+        payload = build_embedding_payload text, model: model
+        response = post embedding_url, payload
+        parse_embedding_response response
+      end
+
       private
 
       def sync_response(payload)
-        response = post payload
+        response = post completion_url, payload
         parse_completion_response response
       end
 
       def stream_response(payload, &block)
         accumulator = StreamAccumulator.new
 
-        post payload do |req|
+        post completion_url, payload do |req|
           req.options.on_data = handle_stream do |chunk|
             accumulator.add chunk
             block.call chunk
@@ -50,8 +56,8 @@ module RubyLLM
         accumulator.to_message
       end
 
-      def post(payload)
-        connection.post completion_url, payload do |req|
+      def post(url, payload)
+        connection.post url, payload do |req|
           req.headers.merge! headers
           yield req if block_given?
         end

data/lib/ruby_llm/providers/openai.rb

@@ -28,6 +28,10 @@ module RubyLLM
         '/v1/models'
       end
 
+      def embedding_url
+        '/v1/embeddings'
+      end
+
       def build_payload(messages, tools:, temperature:, model:, stream: false) # rubocop:disable Metrics/MethodLength
         {
           model: model,
@@ -53,6 +57,18 @@ module RubyLLM
         end
       end
 
+      def build_embedding_payload(text, model:)
+        {
+          model: model,
+          input: text
+        }
+      end
+
+      def parse_embedding_response(response)
+        embeddings = response.body['data'].map { |d| d['embedding'] }
+        embeddings.size == 1 ? embeddings.first : embeddings
+      end
+
       def format_tool_calls(tool_calls) # rubocop:disable Metrics/MethodLength
         return nil unless tool_calls&.any?
 

data/lib/ruby_llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RubyLLM
-  VERSION = '0.1.0.pre16'
+  VERSION = '0.1.0.pre18'
 end

data/lib/ruby_llm.rb

@@ -27,6 +27,10 @@ module RubyLLM
       Chat.new(model: model)
     end
 
+    def embed(...)
+      Embedding.embed(...)
+    end
+
     def models
       Models
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ruby_llm
 version: !ruby/object:Gem::Version
-  version: 0.1.0.pre16
+  version: 0.1.0.pre18
 platform: ruby
 authors:
 - Carmine Paolino
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2025-02-04 00:00:00.000000000 Z
+date: 2025-02-06 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: event_stream_parser
@@ -348,6 +348,7 @@ files:
 - lib/ruby_llm/chat.rb
 - lib/ruby_llm/chunk.rb
 - lib/ruby_llm/configuration.rb
+- lib/ruby_llm/embedding.rb
 - lib/ruby_llm/message.rb
 - lib/ruby_llm/model_capabilities/anthropic.rb
 - lib/ruby_llm/model_capabilities/openai.rb