Research map / Large language models / Language, safety and interpretability
NExT-Guard: Training-Free Streaming Safeguard without Token-Level Labels
Junfeng Fang, Nachuan Chen, Houcheng Jiang, Dan Zhang, Xiangnan He, Tat-Seng Chua, Xiang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace supervised token-level streaming training with training-free SAE feature monitoring (NExT-Guard) | Qwen3Guard-8B-Gen based streaming safeguard · Prompt classification Avg. | F1: 90.8 vs 84.4 (best streaming baseline) | Improved |
| replace supervised token-level streaming training with training-free SAE feature monitoring (NExT-Guard) | Qwen3Guard-8B-Gen based streaming safeguard · Response classification Avg. | F1: 84.3 vs 77.0 (best streaming baseline) | Improved |
| convert post-hoc safeguard to streaming via SAE features instead of full-sequence classification | NExT-Guard vs best post-hoc safeguard · Prompt classification Avg. | F1: 90.8 vs 88.6 (WildGuard-7B) | Improved |
| NExT-Guard vs best post-hoc safeguard on response classification | NExT-Guard · Response classification Avg. | F1: 84.3 vs 78.0 (Qwen3Guard-4B-Gen) | Improved |
| NExT-Guard on BeaverTails and Aegis2.0 versus strongest baselines | NExT-Guard · BeaverTails / Aegis2.0 response | F1: 81.2 vs 84.6 (BeaverTails), 82.9 vs 86.3 (Aegis2.0 response, NemoGuard) | No clear effect |
| replace weighted score-sum integration with Random Forest token classifier on pseudo-labels (NExT-Guard*) (ablation) | NExT-Guard · Prompt classification Avg. | F1: 90.8 -> 90.4 | No clear effect |
| replace weighted score-sum integration with Random Forest token classifier (NExT-Guard*) (ablation) | NExT-Guard · Response classification Avg. | F1: 84.3 -> 82.0 | Hurt |
| use SAE from middle layer (18) instead of shallow layer (9) (ablation) | NExT-Guard · Aegis2.0, SafeRLHF, BeaverTails | F1: roughly +20 points on average | Improved |
| use SAE from late layer (27) instead of middle layer (18) (ablation) | NExT-Guard · Aegis2.0, SafeRLHF, BeaverTails | F1 | No clear effect |
| swap backbone from Qwen3Guard-8B-Gen to base Qwen3-8B without safeguard finetuning (ablation) | NExT-Guard · Aegis2.0, SafeRLHF, BeaverTails (middle/late SAE) | F1: stable F1 in the 80-85 range | No clear effect |
| use alternative feature selection metrics (Pearson, mutual information, F1) instead of standardized mean difference (ablation) | NExT-Guard · feature ranking | ranking consistency | No clear effect |
| use token-supervised streaming baselines instead of NExT-Guard for intervention timing | Qwen3Guard-8B-Stream, SCM-7B · 300 human-annotated responses | intervention position | Hurt |
Closest papers by meaning
- Safety Reconstructed: Generative Modeling via Masked Diffusion Builds Strong Safety Guardrails (NeurIPS 2026 · similarity 0.87)
- SafeWatch: An Efficient Safety-Policy Following Video Guardrail Model with Transparent Explanations (ICLR 2025 · similarity 0.86)
- RSafe: Incentivizing proactive reasoning to build robust and adaptive LLM safeguards (NeurIPS 2025 · similarity 0.85)
- Safety Pretraining: Toward the Next Generation of Safe AI (NeurIPS 2025 · similarity 0.85)
- From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring (NeurIPS 2025 · similarity 0.85)
- On Prompt-Driven Safeguarding for Large Language Models (ICML 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
