This function allows you to chat with a chatbot that answers questions based on the provided context and chat history. It uses GPT-4 architecture to generate responses.
Usage
chat_with_context(
query,
service = "openai",
model = "gpt-4",
index = NULL,
add_context = TRUE,
chat_history = NULL,
history_name = "chat_history",
session_history = NULL,
add_history = TRUE,
task = "Context Only",
k_context = 4,
k_history = 4,
save_history = TRUE,
overwrite = FALSE,
local = FALSE,
embedding_model = NULL
)Arguments
- query
The input query to be processed.
- service
Name of the AI service to use, defaults to openai.
- model
Name of the openai model to use, defaults to gpt-3.5-turbo
- index
Index to look for context.
- add_context
Whether to add context to the query or not. Default is TRUE.
- chat_history
Chat history dataframe for reference.
- history_name
Name of the file where chat history is stored.
- session_history
Session history data for reference.
- add_history
Whether to add chat history to the query or not. Default is TRUE.
- task
Task type, either "Context Only" or "Permissive Chat". Default is "Context Only".
- k_context
Number of top context matches to consider. Default is 4.
- k_history
Number of top chat history matches to consider. Default is 4.
- save_history
Whether to save the chat history or not. Default is TRUE.
- overwrite
Whether to overwrite the history file or not. Default is FALSE.
- local
Whether to use the local model or not. Default is FALSE.
- embedding_model
A model object to use for embedding. Only needed if local is TRUE. Default is NULL.
Examples
if (FALSE) { # rlang::is_interactive()
rlang::is_interactive()
query <- "What is the capital of France?"
result <- chat_with_context(query = query, context = context)
}