guardrails and circle-guard-bench
The former project provides a framework for adding guardrails to LLMs, while the latter offers a benchmark for evaluating the effectiveness of such guard systems, making them complements.
About guardrails
guardrails-ai/guardrails
Adding guardrails to large language models.
This tool helps developers build reliable AI applications by ensuring the output from large language models (LLMs) is safe, compliant, and correctly formatted. It takes an LLM's raw output and applies predefined 'guards' or validation rules to it, flagging or correcting issues like toxic language, competitor mentions, or incorrect data formats. The end user is an AI developer or engineer responsible for integrating LLMs into applications and maintaining their quality and safety.
About circle-guard-bench
whitecircle-ai/circle-guard-bench
First-of-its-kind AI benchmark for evaluating the protection capabilities of large language model (LLM) guard systems (guardrails and safeguards)
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work