Check out our EMNLP paper! We show how to build a state-of-the-art safety guardrail using just a fraction of the training data, by focusing on quality over quantity.
1/4 We trained an 8B safety classifier on just 1,275 carefully annotated prompts.
It beat GPT-5.4, Claude Sonnet 4.6, and leading open-weight guardrails trained on far more data, on average across 5 safety benchmarks.
Quality > quantity.View quoted post on X
