Skip to contents

Create and Manage LLM Prompt Settings

new_LlmPromptConfig.Rd

The new_LlmPromptConfig() function constructs an S3 object that stores the parameters required for making requests to Large Language Models (LLMs) such as OpenAI's GPT models and DeepSeek models.

Usage

new_LlmPromptConfig(
  prompt_content,
  model,
  max_tokens = 100,
  temperature = 1,
  prompt_role = "user",
  seed = NULL,
  top_p = 1,
  n = 1,
  stop = NULL,
  presence_penalty = 0,
  frequency_penalty = 0,
  logprobs = FALSE
)

Arguments

prompt_content

character string containing the primary instruction or query for the model. This serves as the main input to the LLM.

model

Character string specifying the model to use (e.g., `'gpt-4.1'` for OpenAI or `'deepseek-chat'` for DeepSeek). To retrieve a list of valid models for each LLM, use the get_llm_models() method

See the following documentation for valid models: - OpenAI model list - DeepSeek model list

max_tokens

numeric (default: 100) defining the maximum number of tokens to be generated in the response.

temperature

numeric (default: 1.0) controlling randomness in responses. Lower values (e.g., 0.2) make responses deterministic, while higher values (e.g., 1.5) increase creativity.

prompt_role

character (default: 'user') specifying the role of the message. Common values include 'system', 'assistant', and 'user'.

seed

numeric (optional) for controlling reproducibility. If NULL, no seed is set.

top_p

numeric (default: 1) alternative to temperature, specifying nucleus sampling probability. A value of 0.1 considers only the top 10% probability mass.

n

numeric (default: 1) defining the number of responses to generate per request. If temperature is 0, n is automatically set to 1.

stop

character or character vector (default: NULL) defining stop sequences for response termination. Up to 4 sequences can be specified.

presence_penalty

numeric (default: 0) between -2.0 and +2.0, influencing model inclination to introduce new topics.

frequency_penalty

numeric (default: 0) between -2.0 and +2.0, influencing model tendency to repeat words or phrases.

logprobs

boolean (default: FALSE) specifying whether to return log probabilities for output tokens.

Value

An object of class LlmPromptConfig, containing all specified parameters in a structured format.

Examples

if (FALSE) { # \dontrun{
# Retrieve available models
api <- new_RemoteLlmApi(api_key_path = "path/to/openai_key.txt", provider = "OpenAI")
models <- get_llm_models(api)
} # }

# Create a parameter object for OpenAI GPT-4.1
params <- new_LlmPromptConfig(
  prompt_content = 'Explain entropy in simple terms.',
  model = 'gpt-4.1',
  temperature = 0.7,
  max_tokens = 150
)

# Create a parameter object for DeepSeek
params <- new_LlmPromptConfig(
  prompt_content = 'What are three innovative AI research topics?',
  model = 'deepseek-chat',
  temperature = 0.9,
  n = 3
)

# Print the parameter object
print(params)
#> LLM Promp Settings
#> Model: deepseek-chat 
#> Prompt Role: user 
#> Prompt Content: What are three innovative AI research topics? 
#> Max Tokens: 100 
#> Temperature: 0.9 
#> Top-P: 1 
#> N: 3