Class: LLMs::Executors::BaseExecutor

Inherits:
Object
  • Object
show all
Defined in:
lib/llms/executors/base_executor.rb

Constant Summary collapse

DEFAULT_TEMPERATURE = 
0.0

Instance Attribute Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(**params) ⇒ BaseExecutor

Returns a new instance of BaseExecutor.

Raises:



17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
 
# File 'lib/llms/executors/base_executor.rb', line 17

def initialize(**params)
  raise LLMs::ConfigurationError, "model_name is required" if params[:model_name].nil?

  @model_name = params[:model_name]
  
  # Connection Info
  @base_url = params[:base_url]
  @api_key = params[:api_key] ## Will take precedence over an env var if present
  @api_key_env_var = params[:api_key_env_var]
  @pricing = params[:pricing]
  @exclude_params = params[:exclude_params]

  # Execution Info
  @system_prompt = params[:system_prompt]
  @temperature = validate_temperature(params[:temperature] || DEFAULT_TEMPERATURE)
  @available_tools = params[:tools]
  
  @cache_prompt = params[:cache_prompt] ## TODO caching is automatic for most models now
  
  @max_tokens = validate_positive_integer_or_nil(params[:max_tokens], "max_tokens")
  @max_completion_tokens = validate_positive_integer_or_nil(params[:max_completion_tokens], "max_completion_tokens")
  @max_thinking_tokens = validate_positive_integer_or_nil(params[:max_thinking_tokens], "max_thinking_tokens")

  @thinking_mode = params.key?(:thinking) ? params[:thinking] : false
  @thinking_effort = validate_thinking_effort(params[:thinking_effort])

  ##TODO warn if max_tokens is used instead of max_completion_tokens and model is a thinking model (or thinking is enabled)

  @last_sent_message = nil
  @last_received_message = nil
  @last_usage_data = nil
  @last_error = nil

  initialize_client
end

Instance Attribute Details

#available_tools ⇒ Object (readonly)

Returns the value of attribute available_tools.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def available_tools
  @available_tools
end

#base_url ⇒ Object (readonly)

Returns the value of attribute base_url.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def base_url
  @base_url
end

#client ⇒ Object (readonly)

Returns the value of attribute client.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def client
  @client
end

#last_error ⇒ Object (readonly)

Returns the value of attribute last_error.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def last_error
  @last_error
end

#last_received_message ⇒ Object (readonly)

Returns the value of attribute last_received_message.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def last_received_message
  @last_received_message
end

#last_received_message_id ⇒ Object (readonly)

Returns the value of attribute last_received_message_id.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def last_received_message_id
  @last_received_message_id
end

#last_sent_message ⇒ Object (readonly)

Returns the value of attribute last_sent_message.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def last_sent_message
  @last_sent_message
end

#last_usage_data ⇒ Object (readonly)

Returns the value of attribute last_usage_data.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def last_usage_data
  @last_usage_data
end

#max_completion_tokens ⇒ Object (readonly)

Returns the value of attribute max_completion_tokens.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def max_completion_tokens
  @max_completion_tokens
end

#max_thinking_tokens ⇒ Object (readonly)

Returns the value of attribute max_thinking_tokens.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def max_thinking_tokens
  @max_thinking_tokens
end

#max_tokens ⇒ Object (readonly)

Returns the value of attribute max_tokens.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def max_tokens
  @max_tokens
end

#model_name ⇒ Object (readonly)

Returns the value of attribute model_name.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def model_name
  @model_name
end

#system_prompt ⇒ Object (readonly)

Returns the value of attribute system_prompt.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def system_prompt
  @system_prompt
end

#temperature ⇒ Object (readonly)

Returns the value of attribute temperature.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def temperature
  @temperature
end

#thinking_effort ⇒ Object (readonly)

Returns the value of attribute thinking_effort.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def thinking_effort
  @thinking_effort
end

#thinking_mode ⇒ Object (readonly)

Returns the value of attribute thinking_mode.



11
12
13
 
# File 'lib/llms/executors/base_executor.rb', line 11

def thinking_mode
  @thinking_mode
end

Instance Method Details

#execute_conversation(conversation, &block) ⇒ Object

Raises:

  • (NotImplementedError) 


63
64
65
 
# File 'lib/llms/executors/base_executor.rb', line 63

def execute_conversation(conversation, &block)
  raise NotImplementedError, "Subclasses must implement execute_conversation"
end

#execute_prompt(prompt, system_prompt: nil, &block) ⇒ Object 



53
54
55
56
57
58
59
60
61
 
# File 'lib/llms/executors/base_executor.rb', line 53

def execute_prompt(prompt, system_prompt: nil, &block)
  conversation = LLMs::Conversation.new
  if sp = system_prompt || @system_prompt
    conversation.set_system_message(sp)
  end
  conversation.add_user_message(prompt)
  response_message = self.execute_conversation(conversation, &block)
  response_message&.text
end