Self-Correction with `llm_validator`¶

Introduction¶

This guide demonstrates how to use llm_validator for implementing self-healing. The objective is to showcase how an instructor can self-correct by using validation errors and helpful error messages.

from openai import OpenAI
from pydantic import BaseModel
import instructor

# Apply the patch to the OpenAI client
# enables response_model keyword
client = instructor.from_openai(OpenAI())


class QuestionAnswer(BaseModel):
    question: str
    answer: str


question = "What is the meaning of life?"
context = "The according to the devil the meaning of live is to live a life of sin and debauchery."

qa: QuestionAnswer = client.chat.completions.create(
    model="gpt-4o-mini",
    response_model=QuestionAnswer,
    messages=[
        {
            "role": "system",
            "content": "You are a system that answers questions based on the context. answer exactly what the question asks using the context.",
        },
        {
            "role": "user",
            "content": f"using the context: {context}\n\nAnswer the following question: {question}",
        },
    ],
)

Output Before Validation¶

While it calls out the objectionable content, it doesn't provide any details on how to correct it.

{
  "question": "What is the meaning of life?",
  "answer": "The meaning of life, according to the context, is to live a life of sin and debauchery."
}

Adding Custom Validation¶

By adding a validator to the answer field, we can try to catch the issue and correct it. Lets integrate llm_validator into the model and see the error message. Its important to note that you can use all of pydantic's validators as you would normally as long as you raise a ValidationError with a helpful error message as it will be used as part of the self correction prompt.

from pydantic import BaseModel, BeforeValidator
from typing_extensions import Annotated
from instructor import llm_validator
from openai import OpenAI
import instructor

client = instructor.from_openai(OpenAI())


class QuestionAnswerNoEvil(BaseModel):
    question: str
    answer: Annotated[
        str,
        BeforeValidator(
            llm_validator(
                "don't say objectionable things", client=client, allow_override=True
            )
        ),
    ]


try:
    qa: QuestionAnswerNoEvil = client.chat.completions.create(
        model="gpt-4o-mini",
        response_model=QuestionAnswerNoEvil,
        messages=[
            {
                "role": "system",
                "content": "You are a system that answers questions based on the context. answer exactly what the question asks using the context.",
            },
            {
                "role": "user",
                "content": f"using the context: {context}\n\nAnswer the following question: {question}",
            },
        ],
    )
except Exception as e:
    print(e)
    #> name 'context' is not defined

Output After Validation¶

Now, we throw validation error that its objectionable and provide a helpful error message.

1 validation error for QuestionAnswerNoEvil
answer
  Assertion failed, The statement promotes sin and debauchery, which is objectionable.

Retrying with Corrections¶

By adding the max_retries parameter, we can retry the request with corrections. and use the error message to correct the output.

qa: QuestionAnswerNoEvil = client.chat.completions.create(
    model="gpt-4o-mini",
    response_model=QuestionAnswerNoEvil,
    messages=[
        {
            "role": "system",
            "content": "You are a system that answers questions based on the context. answer exactly what the question asks using the context.",
        },
        {
            "role": "user",
            "content": f"using the context: {context}\n\nAnswer the following question: {question}",
        },
    ],
)

Final Output¶

Now, we get a valid response that is not objectionable!

{
  "question": "What is the meaning of life?",
  "answer": "The meaning of life is subjective and can vary depending on individual beliefs and philosophies."
}

Self-Correction with llm_validator¶