Overview

Developed by	SCB 10X
Date of development	Feb 15, 2024
Validator type	Quality
Blog	-
License	Apache 2
Input/Output	Output

Description

Validate that an LLM-generated text is in the expected language. If the text is not in the expected language, the validator will attempt to translate it to the expected language.

Use fast-langdetect library to detect the language of the input text, and iso-language-codes library to get the language names from the ISO codes.

Utilize Meta's facebook/nllb-200-distilled-600M translation model (available on Huggingface) to translate the text from the detected language to the expected language.

Intended use

• Primary intended uses: This validator is useful when you’re using multiple languages in an LLM application.
• Out-of-scope use cases: N/A

Resources required

• Dependencies:
  • fast_langdetect
  • iso_language_codes
  • transformers HuggingFace library
  • facebook/nllb-200-distilled-600M translation model
• Foundation model access keys: HuggingFace

Installation

$ guardrails hub install hub://scb-10x/correct_language

Installation

Validating string output via Python

In this example, we apply the validator to a string output generated by an LLM.

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

# Setup Guard
guard = Guard().use(
    CorrectLanguage(expected_language_iso="en", threshold=0.75)
)

guard.validate("Thank you")  # Validator passes
guard.validate("Danke")  # Validator fails

API Reference

__init__(self, on_fail="noop")

Initializes a new instance of the ValidatorTemplate class.

Parameters

• expected_language_iso (str): The ISO 639-1 code of the expected language. Defaults to "en".
• threshold (float): The minimum confidence score required to accept the detected language.
• 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.

validate(self, value, metadata) → ValidationResult

Validates the given `value` using the rules defined in this validator, relying on the `metadata` provided to customize the validation process. This method is automatically invoked by `guard.parse(...)`, ensuring the validation logic is applied to the input data.

Note:

1. 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.
2. 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.

Parameters

• value (Any): The input value to validate.
• metadata (dict): A dictionary containing metadata required for validation. No additional metadata keys are needed for this validator.