Specifying structured JSON schema
You are part of a team working on an online education platform designing new interactive exercise where students are able to ask questions and their answer is displayed through a graphical view. This question-answering feature is powered by an LLM, but the graphical view requires a JSON input with the fields Question
and Answer
to correctly show the question and answer:
{
"Question": "...",
"Answer": "..."
}
A system prompt and example user question have been stored in the messages
variable.
This exercise is part of the course
Working with Llama 3
Exercise instructions
- Specify the
"Question"
and"Answer"
fields in the schema with the"string"
type.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
output = llm.create_chat_completion(
messages=messages,
response_format={
"type": "json_object",
"schema": {
"type": "object",
# Set the properties of the JSON fields and their data types
"properties": {"____": {"type": "____"}, "____": {"type": "____"}}
}
}
)
print(output['choices'][0]['message']['content'])