为LLM Agents设计的工作台，支持重现、干预与缓解，提升MCP环境下Agent可靠性。
arXiv:2607.11098v1 Announce Type: cross Abstract: Tool-using LLM agents are mostly evaluated assuming all tools work. When a tool times out, returns a…
用微调大模型生成反事实解释，为健康干预设计和数据增强提供新思路，实用性强。
arXiv:2601.14590v3 Announce Type: replace Abstract: Counterfactual explanations (CFEs) provide human-centric interpretability by identifying the minim…
跨领域流式视频任务指导新框架，支持实时干预与多场景适应。
arXiv:2607.02991v1 Announce Type: new Abstract: While multimodal Large Language Models (MLLMs) excel at offline video understanding, an interesting qu…
星展集团研究部的Philip Wee在一份评论中表示，日本可能正在采取新的外汇干预策略。该高级外汇策略师指出，有传言称日本财务省可能会转向不公开的隐秘货币干预。Wee表示，在今天美国债券和股票市场因独立日假期而流动性稀薄的情况下，此类干预“可能会加剧波动“。“日本的新策略似乎旨在将市场心理从怀疑转为…
通过表示层干预让大语言模型终身可控记忆知识，不遗忘旧信息，大模型记忆管理新范式。
arXiv:2511.20892v4 Announce Type: replace Abstract: Large language models (LLMs) often produce incorrect or outdated content after being employed. Eff…
探讨LLM智能体监督中“校准≠控制”的误区，指出需主动干预而非仅依赖校准。
arXiv:2606.21399v1 Announce Type: new Abstract: Runtime oversight for LLM agents is commonly framed as scalar risk prediction: estimate failure likeli…
从零构建AI Agent，详解人机回环与安全机制，附完整代码实战
Article URL: https://www.ruxu.dev/articles/ai/build-an-ai-agent-human-in-the-loop-security/ Comments URL: https://news.ycombinator.com/item?id=4857037…
揭秘LLM推理效率新思路：最小干预即可显著提升性能，少即是多！
arXiv:2510.13940v4 Announce Type: replace-cross Abstract: Recent progress in large language models (LLMs) has focused on test-time scaling to improve …
ICLR 2025论文揭示如何通过神经元干预为LLM注入特定人格特质，打开模型个性化新维度。
arXiv:2410.12327v2 Announce Type: replace Abstract: Large language models (LLMs) have become increasingly proficient at simulating various personality…
最新基准测试CIAware-Bench，评估前沿大模型对控制干预的感知能力，揭示AI安全新挑战。
arXiv:2606.11063v1 Announce Type: new Abstract: AI control protocols oversee untrusted models by monitoring their actions and modifying potentially un…
从界面设计入手，探讨如何通过UI干预引导用户更可持续地使用LLM聊天机器人，跳出传统模型优化思路。
arXiv:2606.10861v1 Announce Type: cross Abstract: LLM-powered chatbots are increasingly embedded in everyday workflows, raising sustainability concern…
突破线性限制，LLM行为控制新范式：可逆潜在变换实现更灵活激活引导
arXiv:2606.08454v1 Announce Type: new Abstract: Activation steering provides a lightweight inference-time mechanism for controlling large language mod…
新研究发现LLM事实知识在最后几层突然“结晶”，而非逐层涌现，解释了为何干预后期层能有效减少幻觉。
arXiv:2606.07978v1 Announce Type: new Abstract: Understanding where LLMs store factual knowledge is critical for hallucination mitigation. We systemat…
LLM智能体悄然失败？新研究用干预法精准定位错误根源，让沉默不再失控。
arXiv:2606.09071v1 Announce Type: new Abstract: Large language model (LLM) agents now solve complex tasks through long plan-and-execution traces, yet …
Claude一天内自主编写了Anthropic 80%的生产代码，未来已来，低干预高效编程震惊行业。
In June 2026, Anthropic released a report called "When AI builds itself." The headlines made it sound like AI was on the verge of superintelligence in…
一种基于证据诊断的干预训练方法，让大模型评分时严格遵循给定规则，提升可靠性与可解释性
arXiv:2606.06350v1 Announce Type: new Abstract: Reliable rubric grading requires more than accurate score prediction. Each judgement must be grounded …
日本或抛售美债干预汇市，还有俞浩表态专注实业、天津AI传感器产业园开园等财经科技快讯一网打尽。
大公司： 首关15亿元，英诺科创三期基金持续聚焦早期科技投资 36氪获悉，6月5日，英诺基金宣布旗下英诺科创三期基金完成首轮关账，规模15亿元。该基金聚焦早期科技投资，重点布局前沿科技及人工智能领域。 比特币“金库”公司市值蒸发620亿美元 比特币本周下跌约14%，至四个月低点附近。随着加密货6月5…
用稀疏恢复技术从子集扰动中精确归因训练数据，破解大模型因果干预的计算难题
arXiv:2606.05165v1 Announce Type: cross Abstract: Training Data Attribution (TDA) seeks to trace a model's predictions back to its training data. The …
语音大模型推理中实体绑定出错了怎么办？这篇论文系统诊断问题并提出思维链干预方案。
arXiv:2606.04474v1 Announce Type: new Abstract: Speech Large Language Models (SLLMs) underperform their text counterparts on complex reasoning. We rev…
自主AI代理执行长任务时，何时干预？本研究用18维情感动力学引擎揭示时机陷阱，LLM裁判也靠不住。
arXiv:2606.04296v1 Announce Type: new Abstract: As autonomous AI agents move from conversational systems to long-horizon software execution, runtime s…