Parse the response of a LLM

Source: R/ResponseParser.R

Parse the response of a LLM

parseLlmResponse(response, noMatchIsDontKnow = TRUE)

Arguments

response

The response of a LLM to the system prompt with prompt generated by createSystemPrompt() and createPrompt(), respectively.

noMatchIsDontKnow

If the response doesn't fit any predefined pattern, should we return "I don't know"?

Value

Returns a character string with one of the following values:

  • "yes": Yes, the patient has the disease.

  • "no": No, the patient does not have the disease.

  • "I don't know": The LLM cannot decide whether the patient has the disease.

  • NA: There was a problem parsing the LLM's response.

Examples

parseLlmResponse("Summary: yes")
#> [1] "yes"
parseLlmResponse("Summary: It is unclear whether the patient has the disease.")
#> [1] "I don't know"