› Async::Ollama › Source › Async › Ollama › Client

`class Client`

Represents a connection to the Ollama service, providing methods to generate completions, chat, and list models.

Definitions

`ENDPOINT = Async::HTTP::Endpoint.parse("http://localhost:11434")`

The default endpoint to connect to.

`def generate(prompt, **options, &block)`

Generates a response from the given prompt using Ollama.

Signature

parameter　prompt　String: The prompt to generate a response from.
parameter　options　Hash: Additional options for the request.
returns　Generate: The generated response representation.

Implementation

def generate(prompt, **options, &block)
	options[:prompt] = prompt
	options[:model] ||= MODEL
	
	Generate.post(self.with(path: "/api/generate"), options) do |resource, response|
		if block_given?
			yield response
		end
		
		Generate.new(resource, value: response.read, metadata: response.headers)
	end
end

`def chat(messages, **options, &block)`

Sends a chat request with the given messages to Ollama.

Signature

parameter　messages　Array(Hash): The chat messages to send.
parameter　options　Hash: Additional options for the request.
returns　Chat: The chat response representation.

Implementation

def chat(messages, **options, &block)
	options[:model] ||= MODEL
	options[:messages] = messages
	
	Chat.post(self.with(path: "/api/chat"), options) do |resource, response|
		if block_given?
			yield response
		end
		
		Chat.new(resource, value: response.read, metadata: response.headers)
	end
end

`def models`

Retrieves the list of available models from Ollama.

Signature

returns　Models: The models response representation.

Implementation

def models
	Models.get(self.with(path: "/api/tags"))
end