omniai-llama 0.0.1 → 2.6.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 25c0a8a9ba448d407d0f8300d377d0e3a10a0aa1b18343540a0b32f3eb3baeb5
-  data.tar.gz: 67d4173043933e6314df901969eae71b0ab1137dfd8da4365377824eeed167dd
+  metadata.gz: b6c8f4322c62d60da848e490cf8815e518e50a9ca542c5fe37cf6efaf47db4cc
+  data.tar.gz: 713668fc41bb593495f60eac721b645bc4b5156ac6e2050c8939db8b10cc0582
 SHA512:
-  metadata.gz: 2e43f7b12d1dbcbe2b02c9a7bc206ee0c1980c231843cb7fc292774548fed69ebb492a17aa8d451374a106bcc7d24f9ec0af27cce34004459493174b74854243
-  data.tar.gz: b991b560116560bff895708f89cc28d01c20a1f477ab69625fb910249bdc7b853eb91f23666b8d1fedb7caaac5f715bc01c3cfa749ee91a0d4929d9e9ac1f3ec
+  metadata.gz: 0f39b5fab49e6aa7d7ec42dc404391c4319a8c5a59d97a43a187f611bb39940231684b07604cac1cc6caedbc0ecbe72f1088665eae0610f141db8db4ba9d5e51
+  data.tar.gz: ff42220c333e7458659d9610ba6ed39d57d08ec500674992271a52527779f0a3be7616a0f7210d6a9732cde9f54d5b6a37085ea5cc45777ae931505fd8014017

data/README.md

@@ -34,7 +34,7 @@ Global configuration is supported for the following options:
 
 ```ruby
 OmniAI::Llama.configure do |config|
-  config.api_key = 'sk-...' # default: ENV['LLAMA_API_KEY']
+  config.api_key = 'LLM|...' # default: ENV['LLAMA_API_KEY']
 end
 ```
 
@@ -59,15 +59,13 @@ completion.content # 'The capital of Canada is Ottawa.'
 
 #### Model
 
-`model` takes an optional string (default is `gpt-4o`):
+`model` takes an optional string (default is `Llama-4-Scout-17B-16E-Instruct-FP8`):
 
 ```ruby
-completion = client.chat('How fast is a cheetah?', model: OmniAI::Llama::Chat::Model::GPT_3_5_TURBO)
+completion = client.chat('How fast is a cheetah?', model: OmniAI::Llama::Chat::Model::LLAMA_4_SCOUT)
 completion.content # 'A cheetah can reach speeds over 100 km/h.'
 ```
 
-[OpenAI API Reference `model`](https://platform.openai.com/docs/api-reference/chat/create#chat-create-model)
-
 #### Temperature
 
 `temperature` takes an optional float between `0.0` and `2.0` (defaults is `0.7`):
@@ -77,8 +75,6 @@ completion = client.chat('Pick a number between 1 and 5', temperature: 2.0)
 completion.content # '3'
 ```
 
-[OpenAI API Reference `temperature`](https://platform.openai.com/docs/api-reference/chat/create#chat-create-temperature)
-
 #### Stream
 
 `stream` takes an optional a proc to stream responses in real-time chunks instead of waiting for a complete response:
@@ -90,8 +86,6 @@ end
 client.chat('Be poetic.', stream:)
 ```
 
-[OpenAI API Reference `stream`](https://platform.openai.com/docs/api-reference/chat/create#chat-create-stream)
-
 #### Format
 
 `format` takes an optional symbol (`:json`) and that setes the `response_format` to `json_object`:
@@ -104,329 +98,4 @@ end
 JSON.parse(completion.content) # { "name": "Ringo" }
 ```
 
-[OpenAI API Reference `response_format`](https://platform.openai.com/docs/api-reference/chat/create#chat-create-stream)
-
 > When using JSON mode, you must also instruct the model to produce JSON yourself via a system or user message.
-
-### Transcribe
-
-A transcription is generated by passing in a path to a file:
-
-```ruby
-transcription = client.transcribe(file.path)
-transcription.text # '...'
-```
-
-#### Prompt
-
-`prompt` is optional and can provide additional context for transcribing:
-
-```ruby
-transcription = client.transcribe(file.path, prompt: '')
-transcription.text # '...'
-```
-
-[OpenAI API Reference `prompt`](https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-prompt)
-
-#### Format
-
-`format` is optional and supports `json`, `text`, `srt` or `vtt`:
-
-```ruby
-transcription = client.transcribe(file.path, format: OmniAI::Transcribe::Format::TEXT)
-transcription.text # '...'
-```
-
-[OpenAI API Reference `response_format`](https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-response_format)
-
-#### Language
-
-`language` is optional and may improve accuracy and latency:
-
-```ruby
-transcription = client.transcribe(file.path, language: OmniAI::Transcribe::Language::SPANISH)
-transcription.text
-```
-
-[OpenAI API Reference `language`](https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-language)
-
-#### Temperature
-
-`temperature` is optional and must be between 0.0 (more deterministic) and 1.0 (less deterministic):
-
-```ruby
-transcription = client.transcribe(file.path, temperature: 0.2)
-transcription.text
-```
-
-[OpenAI API Reference `temperature`](https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-temperature)
-
-### Speak
-
-Speech can be generated by passing text with a block:
-
-```ruby
-File.open('example.ogg', 'wb') do |file|
-  client.speak('How can a clam cram in a clean cream can?') do |chunk|
-    file << chunk
-  end
-end
-```
-
-If a block is not provided then a tempfile is returned:
-
-```ruby
-tempfile = client.speak('Can you can a can as a canner can can a can?')
-tempfile.close
-tempfile.unlink
-```
-
-#### Voice
-
-`voice` is optional and must be one of the supported voices:
-
-```ruby
-client.speak('She sells seashells by the seashore.', voice: OmniAI::Llama::Speak::Voice::SHIMMER)
-```
-
-[OpenAI API Reference `voice`](https://platform.openai.com/docs/api-reference/audio/createSpeech#audio-createspeech-voice)
-
-#### Model
-
-`model` is optional and must be either `tts-1` or `tts-1-hd` (default):
-
-```ruby
-client.speak('I saw a kitten eating chicken in the kitchen.', format: OmniAI::Llama::Speak::Model::TTS_1)
-```
-
-[OpenAI API Refernce `model`](https://platform.openai.com/docs/api-reference/audio/createSpeech#audio-createspeech-model)
-
-#### Speed
-
-`speed` is optional and must be between 0.25 and 0.40:
-
-```ruby
-client.speak('How much wood would a woodchuck chuck if a woodchuck could chuck wood?', speed: 4.0)
-```
-
-[OmniAI API Reference `speed`](https://platform.openai.com/docs/api-reference/audio/createSpeech#audio-createspeech-speed)
-
-#### Format
-
-`format` is optional and supports `MP3` (default), `OPUS`, `AAC`, `FLAC`, `WAV` or `PCM`:
-
-```ruby
-client.speak('A pessemistic pest exists amidst us.', format: OmniAI::Llama::Speak::Format::FLAC)
-```
-
-[OpenAI API Reference `format`](https://platform.openai.com/docs/api-reference/audio/createSpeech#audio-createspeech-response_format)
-
-## Files
-
-### Finding an File
-
-```ruby
-client.files.find(id: 'file_...')
-```
-
-### Listing all Files
-
-```ruby
-client.files.all
-```
-
-### Uploading a File
-
-#### Using a File
-
-```ruby
-file = client.files.build(io: File.open('demo.pdf', 'wb'))
-file.save!
-```
-
-#### Using a Path
-
-```ruby
-file = client.files.build(io: 'demo.pdf'))
-file.save!
-```
-
-### Downloading a File
-
-```ruby
-file = client.files.find(id: 'file_...')
-File.open('...', 'wb') do |file|
-  file.content do |chunk|
-    file << chunk
-  end
-end
-```
-
-### Destroying a File
-
-```ruby
-client.files.destroy!('file_...')
-```
-
-## Assistants
-
-### Finding an Assistant
-
-```ruby
-client.assistants.find(id: 'asst_...')
-```
-
-### Listing all Assistants
-
-```ruby
-client.assistants.all
-```
-
-### Creating an Assistant
-
-```ruby
-assistant = client.assistants.build
-assistant.name = 'Ringo'
-assistant.model = OmniAI::Llama::Chat::Model::GPT_4
-assistant.description = 'The drummer for the Beatles.'
-assistant.save!
-```
-
-### Updating an Assistant
-
-```ruby
-assistant = client.assistants.find(id: 'asst_...')
-assistant.name = 'George'
-assistant.model = OmniAI::Llama::Chat::Model::GPT_4
-assistant.description = 'A guitarist for the Beatles.'
-assistant.save!
-```
-
-### Destroying an Assistant
-
-```ruby
-client.assistants.destroy!('asst_...')
-```
-
-## Threads
-
-### Finding a Thread
-
-```ruby
-client.threads.find(id: 'thread_...')
-```
-
-### Creating a Thread
-
-```ruby
-thread = client.threads.build
-thread.metadata = { user: 'Ringo' }
-thread.save!
-```
-
-### Updating a Thread
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-thread.metadata = { user: 'Ringo' }
-thread.save!
-```
-
-### Destroying a Threads
-
-```ruby
-client.threads.destroy!('thread_...')
-```
-
-### Messages
-
-#### Finding a Message
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-message = thread.messages.find(id: 'msg_...')
-message.save!
-```
-
-#### Listing all Messages
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-thread.messages.all
-```
-
-#### Creating a Message
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-message = thread.messages.build(role: 'user', content: 'Hello?')
-message.save!
-```
-
-#### Updating a Message
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-message = thread.messages.build(role: 'user', content: 'Hello?')
-message.save!
-```
-
-### Runs
-
-#### Finding a Run
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-run = thread.runs.find(id: 'run_...')
-run.save!
-```
-
-#### Listing all Runs
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-thread.runs.all
-```
-
-#### Creating a Run
-
-```ruby
-run = client.runs.find(id: 'thread_...')
-run = thread.runs.build
-run.metadata = { user: 'Ringo' }
-run.save!
-```
-
-#### Updating a Run
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-run = thread.messages.find(id: 'run_...')
-run.metadata = { user: 'Ringo' }
-run.save!
-```
-
-#### Polling a Run
-
-```ruby
-run.terminated? # false
-run.poll!
-run.terminated? # true
-run.status # 'cancelled' / 'failed' / 'completed' / 'expired'
-```
-
-#### Cancelling a Run
-
-```ruby
-thread = client.threads.find(id: 'thread_...')
-run = thread.runs.cancel!(id: 'run_...')
-```
-
-### Embed
-
-Text can be converted into a vector embedding for similarity comparison usage via:
-
-```ruby
-response = client.embed('The quick brown fox jumps over a lazy dog.')
-response.embedding # [0.0, ...]
-```

data/lib/omniai/llama/chat/content_serializer.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+module OmniAI
+  module Llama
+    class Chat
+      # Overrides content serialize / deserialize.
+      module ContentSerializer
+        # @param data [Hash]
+        # @param context [Context]
+        # @return [OmniAI::Chat::Text, OmniAI::Chat::ToolCall]
+        def self.deserialize(data, context:)
+          if data["tool_call"]
+            OmniAI::Chat::ToolCall.deserialize(data, context:)
+          else
+            data["text"]
+          end
+        end
+      end
+    end
+  end
+end

data/lib/omniai/llama/chat/response_serializer.rb

@@ -18,17 +18,17 @@ module OmniAI
       #     metrics: [
       #       {
       #         metric: "num_completion_tokens",
-      #         value: 25,
+      #         value: 2,
       #         unit: "tokens",
       #       },
       #       {
       #         metric: "num_prompt_tokens",
-      #         value: 25,
+      #         value: 3,
       #         unit: "tokens",
       #       },
       #       {
       #         metric: "num_total_tokens",
-      #         value: 50,
+      #         value: 4,
       #         unit: "tokens",
       #       },
       #     ],

data/lib/omniai/llama/chat/stream.rb

@@ -0,0 +1,92 @@
+# frozen_string_literal: true
+
+module OmniAI
+  module Llama
+    class Chat
+      # A stream is used to process a series of chunks of data. It converts the following into a combined payload.
+      class Stream < OmniAI::Chat::Stream
+        # @yield [delta]
+        # @yieldparam delta [OmniAI::Chat::Delta]
+        #
+        # @return [Hash]
+        def stream!(&block)
+          @message = { "role" => "assistant" }
+          @metrics = []
+
+          @chunks.map do |chunk|
+            parser.feed(chunk) do |type, data, id|
+              process!(type, data, id, &block)
+            end
+          end
+
+          {
+            "completion_message" => @message,
+            "metrics" => @metrics,
+          }
+        end
+
+      protected
+
+        #
+        # @param data [Hash]
+        #
+        # @yield [delta]
+        # @yieldparam delta [OmniAI::Chat::Delta]
+        def process_data!(data:, &)
+          event = data["event"]
+
+          process_metrics(metrics: event["metrics"]) if event["metrics"]
+          process_delta(delta: event["delta"], &) if event["delta"]
+        end
+
+        # @param delta [Hash]
+        #
+        # @yield [delta]
+        # @yieldparam delta [OmniAI::Chat::Delta]
+        def process_delta(delta:, &block)
+          block&.call(OmniAI::Chat::Delta.new(text: delta["text"])) if delta["text"] && !delta["text"].empty?
+
+          case delta["type"]
+          when "text" then process_delta_text(delta:)
+          when "tool_call" then process_delta_tool_call(delta:)
+          end
+        end
+
+        # @param delta [Hash]
+        def process_delta_text(delta:)
+          return if delta["text"].empty?
+
+          if @message["content"]
+            @message["content"]["text"] += delta["text"]
+          else
+            @message["content"] = delta
+          end
+        end
+
+        # @param delta [Hash]
+        def process_delta_tool_call(delta:)
+          @message["tool_calls"] ||= []
+
+          latest = @message["tool_calls"][-1]
+
+          if delta["id"]
+            @message["tool_calls"] << {
+              "id" => delta["id"],
+              "function" => delta["function"],
+            }
+          else
+            latest["function"]["arguments"] ||= ""
+            latest["function"]["arguments"] += delta["function"]["arguments"]
+          end
+        end
+
+        # @param metrics [Array<Hash>]
+        def process_metrics(metrics:)
+          return unless metrics
+
+          @metrics = metrics
+        end
+      end
+    end
+  end
+end

data/lib/omniai/llama/chat/usage_serializer.rb

@@ -8,17 +8,17 @@ module OmniAI
       #   [
       #     {
       #       metric: "num_completion_tokens",
-      #       value: 25,
+      #       value: 2,
       #       unit: "tokens",
       #     },
       #     {
       #       metric: "num_prompt_tokens",
-      #       value: 25,
+      #       value: 3,
       #       unit: "tokens",
       #     },
       #     {
       #       metric: "num_total_tokens",
-      #       value: 50,
+      #       value: 4,
       #       unit: "tokens",
       #     },
       #   ]

data/lib/omniai/llama/chat.rb

@@ -2,7 +2,7 @@
 
 module OmniAI
   module Llama
-    # An OpenAI chat implementation.
+    # An Llama chat implementation.
     #
     # Usage:
     #
@@ -13,7 +13,6 @@ module OmniAI
     #   completion.choice.message.content # '...'
     class Chat < OmniAI::Chat
       JSON_RESPONSE_FORMAT = { type: "json_object" }.freeze
-      DEFAULT_STREAM_OPTIONS = { include_usage: ENV.fetch("OMNIAI_STREAM_USAGE", "on").eql?("on") }.freeze
 
       module Model
         LLAMA_4_SCOUT_17B_16E_INSTRUCT_FP8 = "Llama-4-Scout-17B-16E-Instruct-FP8"
@@ -30,12 +29,22 @@ module OmniAI
       CONTEXT = Context.build do |context|
         context.deserializers[:response] = ResponseSerializer.method(:deserialize)
         context.deserializers[:choice] = ChoiceSerializer.method(:deserialize)
-        context.deserializers[:message] = MessageSerializer.method(:deserialize)
+        context.deserializers[:content] = ContentSerializer.method(:deserialize)
         context.deserializers[:usage] = UsageSerializer.method(:deserialize)
       end
 
     protected
 
+      # @return [HTTP::Response]
+      def request!
+        logger&.debug("Chat#request! payload=#{payload.inspect}")
+
+        @client
+          .connection
+          .accept(stream? ? "text/event-stream" : :json)
+          .post(path, json: payload)
+      end
+
       # @return [Context]
       def context
         CONTEXT
@@ -48,7 +57,6 @@ module OmniAI
           model: @model,
           response_format: (JSON_RESPONSE_FORMAT if @format.eql?(:json)),
           stream: stream? || nil,
-          stream_options: (DEFAULT_STREAM_OPTIONS if stream?),
           temperature: @temperature,
           tools: (@tools.map(&:serialize) if @tools&.any?),
         }).compact

data/lib/omniai/llama/config.rb

@@ -2,7 +2,7 @@
 
 module OmniAI
   module Llama
-    # Configuration for OpenAI.
+    # Configuration for Llama.
     class Config < OmniAI::Config
       DEFAULT_HOST = "https://api.llama.com"
 

data/lib/omniai/llama/version.rb

@@ -2,6 +2,6 @@
 
 module OmniAI
   module Llama
-    VERSION = "0.0.1"
+    VERSION = "2.6.0"
   end
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: omniai-llama
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 2.6.0
 platform: ruby
 authors:
 - Kevin Sylvestre
 bindir: exe
 cert_chain: []
-date: 2025-05-01 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: event_stream_parser
@@ -29,14 +29,14 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '2.2'
+        version: '2.6'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '2.2'
+        version: '2.6'
 - !ruby/object:Gem::Dependency
   name: zeitwerk
   requirement: !ruby/object:Gem::Requirement
@@ -51,7 +51,7 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0'
-description: An implementation of OmniAI for OpenAI
+description: An implementation of OmniAI for Llama
 email:
 - kevin@ksylvest.com
 executables: []
@@ -63,8 +63,9 @@ files:
 - lib/omniai/llama.rb
 - lib/omniai/llama/chat.rb
 - lib/omniai/llama/chat/choice_serializer.rb
-- lib/omniai/llama/chat/message_serializer.rb
+- lib/omniai/llama/chat/content_serializer.rb
 - lib/omniai/llama/chat/response_serializer.rb
+- lib/omniai/llama/chat/stream.rb
 - lib/omniai/llama/chat/usage_serializer.rb
 - lib/omniai/llama/client.rb
 - lib/omniai/llama/config.rb
@@ -90,7 +91,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.6
+rubygems_version: 3.6.9
 specification_version: 4
-summary: A generalized framework for interacting with OpenAI
+summary: A generalized framework for interacting with Llama
 test_files: []

data/lib/omniai/llama/chat/message_serializer.rb

@@ -1,31 +0,0 @@
-# frozen_string_literal: true
-
-module OmniAI
-  module Llama
-    class Chat
-      # Overrides choice serialize / deserialize for the following payload:
-      #
-      #   {
-      #     content: {
-      #       type: "text",
-      #       text: "Hello!",
-      #     },
-      #     role: "assistant",
-      #     stop_reason: "stop",
-      #     tool_calls: [],
-      #   }
-      module MessageSerializer
-        # @param data [Hash]
-        # @param context [OmniAI::Context]
-        #
-        # @return [OmniAI::Chat::Message]
-        def self.deserialize(data, context:)
-          role = data["role"]
-          content = OmniAI::Chat::Content.deserialize(data["content"], context:)
-
-          OmniAI::Chat::Message.new(content:, role:)
-        end
-      end
-    end
-  end
-end