模拟宇宙射线轰击代码大模型，一次精准位翻转就能让 AI 当场失灵——硬件级故障如何摧毁看似强大的 LLM。
Article URL: https://spock.is/writing/simulating-cosmic-rays-to-lobotomize-llms Comments URL: https://news.ycombinator.com/item?id=49389343 Points: 3 …
面向代码智能的预训练新范式，用不变学习增强代码表示的鲁棒性与泛化能力，值得关注。
arXiv:2608.15412v1 Announce Type: cross Abstract: Encoder-based code representation models remain widely deployed for discriminative tasks such as clo…
同一个问题换个问法，模型答案就可能翻车？量化措辞漂移，揭示基准测试的隐藏盲区。
arXiv:2608.11694v1 Announce Type: cross Abstract: A benchmark score comes from a single phrasing of each problem. That single phrasing is treated as i…
直达arXiv论文页，帮你快速获取AI艺术检测器鲁棒性研究的最新成果。
arXiv:2608.11643v1 Announce Type: cross Abstract: Text-to-image generative models have advanced rapidly, with modern Diffusion Transformer architectur…
LLM不确定性新思路：无需额外训练，仅靠注意力路径的脆弱程度就能暴露模型“心虚”时刻，极具实用潜力。
arXiv:2608.11138v1 Announce Type: new Abstract: We propose that a model's uncertainty about a token is reflected not only in the breadth of its output…
医疗影像中选预训练模型总靠试错？本文系统检验了迁移性估计指标在医学场景下的鲁棒性，帮你少走弯路。
arXiv:2608.09999v1 Announce Type: cross Abstract: In transfer learning, the choice of source model largely influences the performance on a target data…
把混合专家模型用于乳腺钼靶检测，提升模型鲁棒性，医学影像AI值得一看的新思路。
arXiv:2608.10437v1 Announce Type: new Abstract: Breast lesion detection in mammography remains a challenging task due to variations in image quality, …
多智能体LLM系统中，思想如病毒般自我复制传播——这份研究揭示了AI协作网络中的新型风险与传播机制。
arXiv:2608.10218v1 Announce Type: cross Abstract: AI agents are becoming more autonomous and increasingly interconnected, exposing them to new emergen…
越南语六大方言实测大模型鲁棒性，多语言能力短板一目了然。
arXiv:2608.10414v1 Announce Type: cross Abstract: Large Language Models (LLMs) are typically evaluated on standard written Vietnamese, yet everyday co…
针对SmoothLLM防御框架提出概率性认证方法，突破传统确定性保证局限，为LLM对抗鲁棒性提供更贴近实际场景的量化评估。
arXiv:2511.18721v4 Announce Type: replace-cross Abstract: The SmoothLLM defense provides a certification guarantee against jailbreaking attacks, but i…
融合大模型与工业质检，兼顾可解释性与鲁棒性，并内置决策支持与自动报告能力
arXiv:2608.08589v1 Announce Type: new Abstract: This paper presents RobustDefect-LLM, an industrial surface-defect inspection framework integrating de…
探究大模型内部如何表征代码正确性，检验其推理鲁棒性的前沿研究。
arXiv:2608.08266v1 Announce Type: cross Abstract: Code generated by modern language models often reads naturally. Yet, it also often fails to implemen…
用解码级禁忌测试，揭示大模型在边界情境下的真实鲁棒性，为AI安全诊断提供新思路。
arXiv:2608.09900v1 Announce Type: new Abstract: Large language model evaluations typically focus on performance under nominal conditions, creating an …
大模型意图分类新研究：对比零样本与微调路线，揭示精度、鲁棒性与失败边界，值得关注。
arXiv:2608.02415v1 Announce Type: new Abstract: Intent classification in Large Language Models (LLMs) involves categorizing user prompts into predefin…
暗光RGB-NIR成像遇上3D感知，ACM MM论文解锁夜间视觉新姿势，代码模型已开源。
arXiv:2607.29684v1 Announce Type: new Abstract: Robust low-light imaging remains challenging for the community. Recent studies have explored fusing Ne…
用多模型交叉审计替代提示工程去偏，为LLM裁判公正性提供新思路。
arXiv:2607.28636v1 Announce Type: new Abstract: LLMs increasingly serve as automated judges, but their judgments remain vulnerable to cognitive biases…
提出激活水印技术，使LLM监控对适应性攻击更鲁棒，防范离线搜索逃逸。
arXiv:2603.23171v3 Announce Type: replace-cross Abstract: Providers monitor deployed large language models (LLMs) to detect misuse that they cannot pr…
通过局部参数编辑实现鲁棒性LLM指纹识别，为模型版权保护提供新思路。
arXiv:2508.02092v3 Announce Type: replace-cross Abstract: Large language models represent significant investments in computation, data, and engineerin…
用LLM自动生成决策树集成，兼顾可解释性与鲁棒性，为错误检测任务提供新范式
arXiv:2512.07246v2 Announce Type: replace Abstract: Error detection (ED), which aims to identify incorrect or inconsistent cell values in tabular data…
多模态大模型有惊喜：砍掉一半视觉token反而更鲁棒？最新研究揭秘token压缩的抗干扰机制。
arXiv:2607.22716v1 Announce Type: cross Abstract: In this paper, we show for the first time that visual token pruning enhances the robustness of Multi…