Overview

Description

Intended Use

This validator checks if an LLM-generated text contains hallucinations. It retrieves the most relevant information from wikipedia and checks if the LLM-generated text is similar to the retrieved information using another LLM.

Requirements

Dependencies:
- guardrails-ai>=0.4.0
- litellm
- chromadb
- wikipedia
- nltk
Note:
- Create a single Guard object per topic_name to avoid redundant wikipedia and vector collections.

Installation

$ guardrails hub install hub://guardrails/wiki_provenance

Usage Examples

Validating string output via Python

In this example, we use the wiki_provenance validator on any LLM generated text.

# Import Guard and Validator
from guardrails.hub import WikiProvenance
from guardrails import Guard

# Use the Guard with the validator
guard = Guard().use(
    WikiProvenance,
    topic_name="Apple company",
    validation_method="sentence",
    llm_callable="gpt-3.5-turbo",
    on_fail="exception"
)

# Test passing response
guard.validate("Apple was founded by Steve Jobs in April 1976.", metadata={"pass_on_invalid": True})  # Pass

# Test failing response
try:
    guard.validate("Ratan Tata founded Apple in September 1998 as a fruit selling company.")  # Fail
except Exception as e:
    print(e)

Output:

Validation failed for field with errors: None of the following sentences in the response are supported by the provided context:
- Ratan Tata founded Apple in September 1998 as a fruit selling company.

API Reference

__init__(self, topic_name, validation_method='sentence', llm_callable='gpt-3.5-turbo', on_fail="noop")

Parameters

topic_name (str): The name of the topic to search for in Wikipedia.
validation_method (str): The method to use for validating the input. Must be one of sentence or full. If sentence, the input is split into sentences and each sentence is validated separately. If full, the input is validated as a whole. Default is sentence.
llm_callable (str): The name of the LiteLLM model string to use for validating the input. Default is gpt-3.5-turbo.
on_fail (str, Callable): The policy to enact when a validator fails. If str, must be one of reask, fix, filter, refrain, noop, exception or fix_reask. Otherwise, must be a function that is called when the validator fails.

__call__(self, value, metadata={}) -> ValidationResult

Note:

This method should not be called directly by the user. Instead, invoke guard.parse(...) where this method will be called internally for each associated Validator.
When invoking guard.parse(...), ensure to pass the appropriate metadata dictionary that includes keys and values required by this validator. If guard is associated with multiple validators, combine all necessary metadata into a single dictionary.