chat_completion: Generate Text Using the OpenAI API's Chat Endpoint

Description

This function generates natural language text in a conversational style using the OpenAI API's chat endpoint. It takes a series of chat messages as input, either as a data.frame or a chatlog object, and generates a text completion based on the conversation history and the specified model parameters.

Usage

chat_completion(
  msgs,
  model = "gpt-3.5-turbo",
  temperature = NULL,
  max_tokens = NULL,
  n = NULL,
  stop = NULL,
  presence_penalty = NULL,
  frequency_penalty = NULL,
  best_of = NULL,
  logit_bias = NULL,
  stream = FALSE,
  top_p = NULL,
  user = NULL
)

Value

A character vector containing the generated text(s).

Arguments

msgs: A data.frame containing the chat history to generate text from or a chatlog object.
model: A character string specifying the ID of the model to use. The default value is "gpt-3.5-turbo".
temperature: An optional numeric scalar specifying the sampling temperature to use.
max_tokens: An optional integer scalar specifying the maximum number of tokens to generate in the text.
n: An optional integer scalar specifying the number of text completions to generate.
stop: An optional character string or character vector specifying one or more stop sequences to use when generating the text.
presence_penalty: An optional numeric scalar specifying the presence penalty to use when generating the text. The default value is NULL.
frequency_penalty: An optional numeric scalar specifying the frequency penalty to use when generating the text. The default value is NULL.
best_of: An optional integer scalar specifying the number of completions to generate and return the best one. The default value is NULL.
logit_bias: An optional named numeric vector specifying the logit bias to use for each token in the generated text.
stream: An optional logical scalar specifying whether to use the streaming API. The default value is FALSE.
top_p: An optional numeric scalar specifying the top p sampling ratio. The default value is NULL.
user: A unique identifier representing your end-user, which can help OpenAI to monitor and detect abuse.

Author

Ulrich Matter umatter@protonmail.com

Examples

Run this code

if (FALSE) {
openai_api_key("your_api_key_here")
msgs_df <- data.frame(role=c("system",
"user",
"assistant",
"user"),
content=c("You are a helpful assistant",
"Who won the world series in 2020?",
"The Los Angeles Dodgers won the World Series in 2020.",
"Where was it played?"))
chat_completion(msgs_df)
}