reasoning-active-training

DT-Guard introduces a reasoning-active training, reasoning-free inference paradigm for LLM safety guardrails, achieving strong F1 scores on safety benchmarks with a 4B backbone model, outperforming larger baselines.