Research question
How can application developers add programmable safety rails to LLM apps?
Key findings
- Colang dialogue rails, input/output checks and fact-checking rails composed around any model - Open-sourced toolkit adopted widely
NeMo Guardrails
NVIDIA's programmable guardrail toolkit, the reference architecture for application-level rails around LLMs.
Contribute
Spot an error, or know a record we should add? Members can propose additions and submit corrections.