Guardrails AI
Adding guardrails to large language models
About Guardrails AI
Guardrails is a Python package that provides users with the ability to add structure, type, and quality assurance to the output generated by large language models (LLMs). Guardrails AI:
- uses pydantic-style validation to thoroughly check the output of LLMs, including the detection of any bias, errors, or other irregularities.
- has the capability to take corrective measures when validation fails,
- ensures that the output generated adheres to specific structure and type requirements, such as a JSON format.