具有潜在推理的鲁棒高效防护栏
Robust and Efficient Guardrails with Latent Reasoning
AI总结：
提出COLAGUARD模型，通过阶段式训练将多步安全推理转移到连续潜在空间，在保持高安全性能的同时实现12.9倍加速和22.4倍令牌减少。
摘要：
随着大型语言模型（LLMs）在现实应用中的日益部署，维护其安全性至关重要。现有的安全防护栏通常依赖单次分类或更近期的蒸馏推理。基于推理的防护栏显著优于仅分类的基线，但会带来大量的查询延迟和令牌开销，使其不适用于高吞吐量部署。为了解决这一挑战，我们提出了COLAGUARD，一种通过阶段式训练课程将多步安全推理转移到连续潜在空间的防护栏模型，从而在推理时实现直接的隐藏状态传播。在涵盖八个安全基准的十个提示和响应审核设置上评估，COLAGUARD在宏观F1上比Llama Guard 3提高了8.24分，并与我们的显式推理基线GuardReasoner在宏观F1上相当，同时实现了12.9倍的加速和22.4倍的令牌使用减少。我们的结果表明，潜在推理为可部署的防护栏提供了一种实用的替代方案，以替代显式理由生成，共同提高安全鲁棒性和推理效率，而不是将它们视为竞争目标。
英文摘要：
Maintaining the safety of large language models (LLMs) is crucial as they are increasingly deployed in real-world applications. Existing safety guardrails typically rely on single-pass classification or, more recently, distilled reasoning. Reasoning-based guardrails significantly outperform classification-only baselines, but they incur substantial query latency and token overhead that make them impractical for highthroughput deployment. To address this challenge, we propose COLAGUARD, a guardrail model that transfers multi-step safety reasoning into a continuous latent space through a stage-wise training curriculum, enabling direct hidden-state propagation at inference. Evaluated on ten prompt- and response-moderation settings spanning eight safety benchmarks, COLAGUARD improves macro-F1 by 8.24 points over Llama Guard 3 and matches our explicit reasoning baseline, GuardReasoner, in macroF1 while delivering a 12.9X speedup and 22.4X reduction in token usage. Our results suggest that latent reasoning offers a practical alternative to explicit rationale generation for deployable guardrails, jointly improving safety robustness and inference efficiency rather than treating them as competing objectives.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
