Guard Vector: Beyond English LLM Guardrails with Task-Vector Composition and Streaming-Aware Prefix SFT
This work addresses safety and efficiency challenges in multilingual AI systems, though it appears incremental by building on existing guardrail methods.
The authors tackled the problem of extending safety guardrails beyond English to other languages and improving efficiency under streaming conditions, achieving improved classification quality across safety suites and enabling language extensibility to Chinese, Japanese, and Korean without additional training or labels.
We introduce Guard Vector, a safety task vector computed as the parameter difference between a guardrail model (Guard Model) and a same-architecture pretrained language model. Composing this vector with a target language model yields a Target Guard Model (TGM). We then adapt TGM with a streaming-aware approach that combines prefix-based training and evaluation with a classifier that produces a single-token output. With this composition alone, TGM improves classification quality over established Guard Models across standard safety suites and enables language extensibility to Chinese, Japanese, and Korean, requiring neither additional training nor target language labels. It also demonstrates model portability across two widely used public guardrail backbones, Llama and Gemma. With prefix SFT (supervised fine-tuning), TGM preserves classification quality under streaming by aligning the behavior between prefix inputs and full-text inputs. The single-token output design increases throughput and reduces latency. Together, these components reduce data and compute requirements while promoting streaming-aware evaluation practices, thereby contributing to a more responsible AI ecosystem.