Metraipolis › archive › dc_da04ce5210b0
COGNIT-Guard: Calibrated Standalone Direct-Decision Guardrails with Heterogeneous CPU-NPU Confidence Cascading under Explicit Latency and False-Positive Constraints — arXiv:2609.33671
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.33671
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T22:34:40.691Z
- SHA-256 of the archived text
8c362f8e5efd5ec367b2223aee4656bd8f1b33886d2de78ea80e9f884e7d502a- Stable URL of this page
https://metraipolis.com/documento/dc_da04ce5210b0
How to cite
COGNIT-Guard: Calibrated Standalone Direct-Decision Guardrails with Heterogeneous CPU-NPU Confidence Cascading under Explicit Latency and False-Positive Constraints — arXiv:2609.33671 — arXiv (en), https://arxiv.org/abs/2609.33671 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_da04ce5210b0 (Metraipolis archive dc_da04ce5210b0, sha256 8c362f8e5efd5ec367b2223aee4656bd8f1b33886d2de78ea80e9f884e7d502a)
BibTeX
@misc{dcda04ce5210b0,
title = {COGNIT-Guard: Calibrated Standalone Direct-Decision Guardrails with Heterogeneous CPU-NPU Confidence Cascading under Explicit Latency and False-Positive Constraints — arXiv:2609.33671},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.33671},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_da04ce5210b0; sha256 8c362f8e5efd5ec367b2223aee4656bd8f1b33886d2de78ea80e9f884e7d502a},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33671. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
When must a foundation-model safety gateway generate tokens, and when should it directly output a calibrated decision? We study calibrated standalone direct-decision foundation models for real-time pre-ingestion safety guardrails, jointly addressing probability calibration, dual-use false-positive control, and heterogeneous CPU-NPU routing under explicit latency SLOs. Pre-ingestion guardrails must screen prompts prior to target-LLM prefill with low false alarms on benign compliance inquiries; however, shallow classifiers are brittle to phrasing shifts, hidden-state probes require coupling to a target LLM, and generative guards incur high decoding latency and dual-use false positives. We present COGNIT-Guard, coupling a validation-calibrated CPU fast gatekeeper with confidence-gated escalation to an NPU-resident 322M bidirectional direct-decision model (Laya-322M) under an asymmetric false-positive penalty. On the clean unseen DUCS-Bench test split ($N=607$), COGNIT-Guard achieves 98.85% accuracy (McNemar $p = 1.19 \times 10^{-4}$ vs. ML), reduces benign FPR to 0.42% ($1/238$; Fisher's exact $p = 8.23 \times 10^{-4}$ vs. ML), and attains 1.12% ECE and 0.0104 Brier score. On Huawei Ascend 910C NPUs, pure NPU inference runs in 21.77 ms mean latency (45.90 QPS), while the live serial CPU-NPU cascade ($\theta^*_{\mathrm{deploy}}=0.70$) achieves 41.63 ms mean latency (P50: 39.47 ms, 99.23% accuracy, 0.00% FPR). Evaluation on SafetyBench-ZH ($N=2,100$) and comparison against a bi-encoder direct-decision baseline (CLM-8B) disentangle in-domain gains, OOD alignment tax (60.33% $\to$ 56.81% on Laya; 55.10% on domain CLM-8B), and experience replay recovery, restoring OOD accuracy to 64.10%-65.05% and reaching 99.67%-99.84% in-domain accuracy with 0.00%-0.42% FPR.
Tema degli esploratori: fiducia digitale e crittografia (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CR).
Categorie arXiv: Cryptography and Security (cs.CR) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.33671 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T22:13:03.695Z dalla pagina https://arxiv.org/catchup/cs.CR/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).