From Scores to Evidence: Auditable Decisions Can Improve Speech Deepfake Detection
可审计决策记录改善语音深伪…
作者 Mengzhe Geng
Intel · 情报源
Intel 情报源
AI 开发圈现在正在聊什么，直接从来源抓取——最新的在最前面。
已收集 2,033 条 · 第 5 页，共 34 页
Evolution of Multimodal Question Answering: From Modality-Adaptive Extraction to Unified Language Representation
比较多模态QA框架，显示转向统一…
作者 Abdullah Al ShafiQ2D-Web: A Large-Scale Benchmark for Retrieval in Agentic RAG Systems
大型基准测试用于代理式RAG检索，含1.9亿…
作者 Maximilian SchallI Rewrote My Back End in Go in 72 Hours and Cut My AWS Bill by 94%
用Go重写后端，72小时内AWS成本降94%。
Disappointed Optimists
反思未达期望与乐观。
OntoKG-EQ: A provenance-grounded, competency-question-governed knowledge graph for auditable analyst querying
知识图谱用于新兴领域中可审计的分析师查询…
作者 Furqan NasirAPI Benchmark Scores Do Not Reliably Transfer to Chatbot Interfaces
API基准分数无法可靠转移至聊天机器人…
作者 Jennifer WangReverse engineering the 2008 LogiCola, a logic drill program
逆向工程2008年逻辑钻孔程序。
Google DeepMind Releases AlphaGenome Atlas
DeepMind发布AlphaGenome Atlas用于基因组学。
ZX Spectrum: Experimenting with 1-Bit Sound
ZX Spectrum上1位元声音实验。
Closing the Consistency Gap: Self-Evolving Agents That Learn to Stay on Course
自演化代理缩小任务成功的一致性差距。
作者 Evelyn DuesterwaldBenchmarking Qwen3.8 27B quantizations: 4-bit holds up, 1-bit collapses
Qwen3.8 27B：4位元可用，1位元失败。
A Closed-Form Estimator and Diagnostic Battery for Anchor-Judge Error Correlation, Under a Single-Common-Factor Model
锚定评审误差相关性的闭式估计器。
作者 Veerendra Kumar SunkavalliThe two Christian saints who are the Buddha
两位基督教圣人关联佛陀起源。
DHS 'Predictive Policing' Unit Is Analyzing Americans' Financial Habits
DHS分析财务习惯用于预测性警务。
Paramount Caught Using 'Astroturf' Group to Drum Up Fake Support for Merger
Paramount使用假团体支持合并。
The far right's win in Germany is sending a chill through Europe's establishment
德国极右翼胜选令欧洲建制派担忧。
Apple unfolds its decade-long iPhone
苹果公布十年iPhone计划；澳大利亚推动…
Ostrich: Taking Large Strides Through Stiff Contact in Differentiable Dynamics
可微分模拟器处理大规模刚性接触…
作者 Aleš KučeraThe Rater Ising-Potts Model with LLM-Derived Weights: An Application to Multi-Category Scoring Reliability
Ising-Potts模型用于多类别评分可靠性。
作者 Matthias von DavierHi-FLoop: Hierarchical State-Feedback Loops for Multi-Timescale World Modeling
多时间尺度世界模型的分层状态反馈回路。
作者 Rx FanEvidence-Grounded Retrieval for Investigation Hunt Lead Generation from CTI Reports
基于证据的检索用于CTI报告的猎捕线索。
作者 Akash PrakashSilent Revision: Measuring Undisclosed Change in the Safety Frameworks of Frontier AI Developers
衡量AI安全框架中未披露的变更。
作者 Louis Yiven ZhuLatest open artifacts (#24): Motif-3, GLM-5.3, Hy4-preview and open model licenses
新开放模型与许可扩展生态系统。
作者 Florian BrandAlphaGenome Atlas predictive map of every DNA letter change in the human genome
地图预测所有人类DNA变化的影响。
I-have-ADHD: A skill to stop coding agents from burying the answer
防止编码代理隐藏答案的技巧。
It's All in the Way You Say It: The Role of Information Representation in LLM-Based Glycemic-Event Prediction
信息表示影响LLM血糖事件…
作者 Andrea ApicellaLibreOffice breaks download records after declaring it has no AI features
LibreOffice下载创纪录，无AI功能。
Benchmark Scores Are Pipeline-Dependent: A Reliability Audit of Cybersecurity LLM Benchmarks
网络安全LLM基准分数依赖管道。
作者 Aymene BerricheKairos: A Dataset for Fine-Grained Video-Language Modeling over Space, Time, and Dynamics
用于时间上细粒度视频语言建模的数据集。
作者 Ruibo MingWe Must Return to the Office to Use AI in Person
主张回办公室以使用AI的论点。
DaVinci Resolve 21.1
DaVinci Resolve 21.1发布，含更新。
Show HN: Copperhead – Hardware as Fast as Software
Copperhead目标是硬件速度的软件。
GoAnt: Quality-Diversity Multi-Agent Search for Alpha Factor Discovery in Market Microstructure Data
用于alpha因子的质量多样性多智能体搜索…
作者 Stella ZhaoChimaera: A Mixture-of-Graph-Experts Architecture for Cross-Task and Cross-Dataset Graph Learning
用于跨任务/数据集的专家混合图模型…
作者 Jonathan FrankThe first GPT-6 model
首个GPT-6模型及其预期用途。
TontaubeV1: Streaming Text-to-Speech with Hierarchical Codec Modeling and Bounded Context
具有分层编解码器和有界上下文的流式TTS。
作者 Fritz CremerRecord Grouping Controls Evidence Weight in Language Models
记录分组控制语言模型中的证据权重。
作者 Zhongxuan LiuGlobal Divergence, Local Convergence: Representation Geometry in SSMs and Transformers
SSM和Transformer全局不同但局部收敛。
作者 Amit Ben-ArtzyWhen Victorian Becomes a Prompt: Literary Periodization as a Generative Constraint in 100 AI-Generated Novels
AI生成小说中的维多利亚提示改变语法。
作者 Mehdy Sedaghat PayamCausalChapter: Improving Long-Video Chaptering with Interventional Dependency Modeling
CausalChapter通过依赖改善长视频分章…
作者 Xinran DuanBIFTA: Brain-Inspired Few-Shot Tactile Adaptation for Unknown Sensors
脑启发式少样本适应于未知触觉…
作者 Boheng LiuX2Streaming-ASR: wait when uncertain, emit when ready for streaming ASR
流式ASR就绪时提交，不确定时等待。
作者 Zhiwei LinMoEMB: Scaling Universal Multimodal Embeddings with Efficient Mixture-of-Experts Models
MoE扩展用于通用多模态嵌入…
作者 Xuanming CuiTriCCOT: Tri-part Convolutional Conformal Transformer for Onboard Space Object Detection
三部分CNN-共形-变压器用于机载物体…
作者 Adrien DoriseCharts Are Beyond Pixels: Probing for Layer-Wise Chart Understanding and Editing
基准测试探测分层图表理解…
作者 Xiaochuan ZhongDifficulty-Adaptive Tree-Structured Policy Optimization for Expanding Reasoning Coverage in RLVR
难度自适应树搜索提升推理覆盖…
作者 Youngjun YuCombating Instruction Conflict via Energy-Driven Latent Conflict Detection
能量驱动潜在检测器用于指令冲突…
作者 Mingyu MaSUN: Reaching for Novelty in Reinforcement Learning
可达性感知目标选择用于RL新奇性。
作者 Wenyan YangCASD: Chunk-Aligned Semantic Distillation for Multi-StageRobot Manipulation
块对齐语义蒸馏用于多阶段机器人…
作者 Tinghe DingNavigating the digital spectrum: Assessing political bias, stability, and downstream fairness in Large Language Models
稳健评估LLM中的政治偏见与公平性。
作者 Luka DebevcSuan: Rectifying Direct Preference Safety Alignment in Large Language Models
梯度级偏好优化用于安全对齐。
作者 Oleksandr CherednichenkoFreeBSD 14.5-Release
FreeBSD 14.5发布，含更新与修复。
SynthRCT: Scalable Conditional Deformation Synthesis for Synthetic Repeat CT Generation
可扩展条件形变合成用于重复CT…
作者 Tomas Guija-ValienteLeveraging contextual events on structure-aware next activity prediction
情境事件通过…改善下一活动预测
作者 Alessandro MeleTarget-Independent Micro-Interventions for Predicting Training Response Across Language-Model Families
微干预预测跨模型的训练反应…
作者 Zhongxuan LiuWhy shared attention vectors fail: a case for outcome-indexed tuning
共享注意力向量在多结果下不稳定…
作者 Lenard DomeNeptune: An AI model for Global Ocean Subseasonal Prediction
Neptune是数据驱动的海洋模拟器，用于次季节…
作者 Davide DonnoCLAMP: Constrained Decoding for Vision-Language Embodied Planning
CLAMP通过约束…将VLM计划基于场景证据
作者 Tianyi MaGraph-Based Personalized Memory for LLM Agents: Representation, Evolution, Retrieval, and Evaluation
基于图的LLM代理个性化记忆调查…
作者 Dac Duy Anh Nguyen