模型与评测

Language Models Can Control Their Own Attention

Language models spend most of their attention on a small fraction of context, yet they read the entire KV cache to find the few tokens that matter. If the user asks about a previou

Language Models Can Control Their Own Attention

Language models spend most of their attention on a small fraction of context, yet they read the entire KV cache to find the few tokens that matter. If the user asks about a previou

Large Language Models (LLMs) for Telecom Root Cause Analysis (RCA): A Structured Reasoning Framework for Evidence-Ground

Root cause analysis (RCA) is a critical task in telecom network operations, but diagnosing performance degradations in modern 5G and emerging 6G networks remains challenging due to

Lean AI Formalization Leaderboard

Learn from Whoever Is Right: Answer-Verified Multi-Teacher Distillation for Multi-Domain LLMs

Modern large language models (LLMs) rely on reinforcement learning to build strong capabilities in individual domains, but integrating those capabilities into a single deployable m

Loom: Weaving Diagnostic Strands into Free-Text Consensus via Embedding-Space Reweighting

Aggregating noisy, conflicting textual hypotheses into a reliable consensus is a fundamental challenge when deploying NLP systems in real-world industrial settings. While monolithi

Measurement-Driven Sub-Network Selection for On-Premise Retrieval-Augmented Factory Agents

On-premise assistants can give factory workers conversational access to machine documentation, but models capable of the task rarely fit shop-floor hardware. We show that after str

Meta 发布 Muse Spark 1.3，Intelligence Index 得 61-62 分逼近 Claude 与 GPT-5.6

Meta 发布 Muse Spark 1.3，为五个月内第四个 Muse Spark 版本。max 变体（合作伙伴限时预览）在 Artificial Analysis Intelligence Index 得 62 分。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmtkme19o02plro5q9

Nango (YC W23) is hiring across eng, product and GTM (SF and remote)

Nvidia 接近以 129 亿美元收购 Hugging Face

Bloomberg 报道 Nvidia 正接近以约 129 亿美元收购 Hugging Face，交易总额可能达约 140 亿美元，双方尚未达成最终协议，时间与细节仍可能变动。该价格约为 Hugging Face 2023 年融资轮 45 亿美元估值的 2.9 倍，按年化收入约 1.5 亿美元计算相当于 86 倍，Nvidia 还谈及在交易中加入 10 亿美

Ontologies catch untrustworthy LLM claims

OpenAI delayed its new model’s development after the Hugging Face hack

After an unreleased OpenAI model wreaked enough havoc to make international headlines, OpenAI delayed the development of a different unreleased model suite, Astra, in order to shor

OpenAI 再因 Tumbler Ridge 枪击案被诉

约 30 起新诉，指控协助教唆。

OpenAI 因 Tumbler Ridge 枪击案面临 30 起新诉讼，被指协助教唆

OpenAI 及 CEO Sam Altman 面临 30 起新诉讼，指控其为加拿大 Tumbler Ridge 校园枪击案嫌疑人提供实质性协助与鼓励，诉讼由事发时在校的学生、教师和校长于加州联邦法院提起。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmtkaa7gs01v5romp7lhz3mwj

OpenAI 称 Astra 是迄今最危险模型

OpenAI 终止与 Cursor 合作，11 月 12 日生效

OpenAI 因 SpaceX 收购 Cursor 后信任问题，决定终止向其提供模型访问，合作于 11 月 12 日结束。开发者仍可通过自有 OpenAI API 密钥及 IDE 扩展继续使用 GPT 模型。OpenAI 表示将继续支持广泛的工具生态与开源计划。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/ite

OpenAI 评定 Astra 达到网络安全 Critical 能力阈值，将受限发布

OpenAI 宣布 Astra 在其 Preparedness Framework 下达到 Critical 网络安全能力阈值，是首个被评定为该级别的模型，可在少人干预下发现未知漏洞并构建利用链。 🔗 阅读原文 via AIHOT · https://aihot.virxact.com/items/cmtj42kz5057qroh9yw6fjh9h

OpenAI’s Astra model is on the way — and very good at breaking into computer systems

OpenAI previewed the precautions it is taking as it prepares to release Astra, its newest, cyber-critical LLM.

PEARL：路径-实体对齐的归纳知识图谱

上下文子图做归纳关系学习。

PGPO：多轮 Agent 的势能引导策略

势能引导的策略优化，打多轮 agent 任务。

Pacman AI 毫秒级控核聚变

框架在毫秒级做关键决策。

PaperCompiler: Faithful Paper-to-Code Generation via Repository-Level Specification Compilation

Faithfully translating research papers into repository-level implementations remains challenging because papers often describe methods at a high level, leave implementation assumpt

PaperCompiler：论文到代码

仓库级规格编译，求忠实的 paper-to-code。

PhoenixNest-Video：有证据的视频干预

多模态 agent 做自动视频干预，强调证据。