This function is a R wrapper of `chatgpt_chaindoc.py`, a python script function aimed to elaborate large text documents using LangChain and ChatGPT. `chatgpt_chaindoc.py` uses the refine chain method to elaborate the text (see https://langchain.readthedocs.io/en/latest/modules/indexes/chain_examples/summarize.html#the-refine-chain for details).
Usage
chatgpt_wrapper(
document_path = NULL,
openaikey = NULL,
engine = "gpt-3.5-turbo",
temperature = 0.5,
refine_text = NULL
)Arguments
- document_path
The path of the txt document to summarise.
- openaikey
The OpenAI key token.
- engine
ChatGPT model (see https://platform.openai.com/docs/models).
- temperature
Level of randomness or "creativity" in the generated text (see https://platform.openai.com/docs/api-reference/completions/create).
- refine_text
A text useful to refine the original prompt (see https://langchain.readthedocs.io/en/latest/modules/indexes/chain_examples/summarize.html#the-refine-chain for details).
Examples
if (FALSE) {
output <-
summarise_chatgpt_wrapper(
document_path = "inst/docs_dataverse/5636634.txt",
openaikey = OPENAI_KEY,
engine = "gpt-3.5-turbo",
temperature = 0.7,
refine_text = pars$openai$refine_prompts$p1
)
cat(output$output_text)
}