NVIDIA's guardrail toolkit
Programmable rails that keep LLM chatbots on-topic, safe and secure.
NVIDIA's Colang-based guardrails: programmable rails that keep LLM dialogs on-topic, safe and factual, with out-of-the-box topics and jailbreak detection. The enterprise option among guardrail libraries.
Who it's for
Python teams adding conversational guardrails (topic control, hallucination blocking) to LLM apps.
Official links
Keep exploring
Concepts in the glossary
1 in catalog
NeMo Guardrails is written in
2 in catalog