2026-07-17 科技日报

扫描 560 篇候选内容 · 覆盖 650 个信息源 · 纳入 544 篇

📌 今日要点

• NVIDIA Nemotron 3 Embed Ranks #1 Overall on RTEB, Advancing Agentic Retrieval
• Newer Models, Same Advantage
• Security incident disclosure — July 2026
• Build enterprise search for agents with Amazon Bedrock Managed Knowledge Base
• Introducing Grok on Amazon Bedrock
• Building a restaurant telephony AI host with Amazon Bedrock AgentCore and Amazon Nova 2 Sonic

📝 科技简讯

• NVIDIA Nemotron 3 Embed Ranks #1 Overall on RTEB, Advancing Agentic Retrieval — Hugging Face Blog
• Newer Models, Same Advantage — Hugging Face Blog
• Security incident disclosure — July 2026 — Hugging Face Blog
• Build enterprise search for agents with Amazon Bedrock Managed Knowledge Base — AWS Machine Learning
• Introducing Grok on Amazon Bedrock — AWS Machine Learning
• Building a restaurant telephony AI host with Amazon Bedrock AgentCore and Amazon Nova 2 Sonic — AWS Machine Learning
• Help us launch AI safety university groups by referring potential founders — LessWrong
• I would only bet at 30% on meeting grabby aliens — LessWrong
• How (not) to fundraise from Anthropic staff — LessWrong
• Learning concepts is dirty work — LessWrong
• The getting is good (optimizing unattended runs) — LessWrong
• Automatic Differentiation from Scratch: How PyTorch Computes Gradients in Physics-Informed Neural Networks — ArXiv ML (cs.LG)
• Beyond Backbone Backpropagation: A Decoupled Strategy for Efficient Transfer Learning — ArXiv ML (cs.LG)
• Federated Explainable Artificial Intelligence: Roles, Architectures, Evaluation, and Open Challenges — ArXiv ML (cs.LG)
• What Your Model Threw Away and Why You’ll Want It Back: Masking, Fingerprinting, and Privacy from Discarded Geometry — ArXiv ML (cs.LG)
• Targeted Recovery of Weight-Space Mechanisms From Neural Networks — ArXiv ML (cs.LG)
• Uncertainty-Aware Sequential Decision Rules for Event-Triggered LLM Invocation in Streaming Systems — ArXiv ML (cs.LG)
• TSSM: Triaxial State Space Model for Global Station Weather Forecasting with Temporal-Variable-Historical Modeling — ArXiv ML (cs.LG)
• Our approach to bioresilience — Google DeepMind
• Connect more of your apps to Search — Google AI Blog
• Create, edit and star in videos with two Google Vids updates — Google AI Blog
• Working with Pi Coding Agents — KDnuggets
• 10 YouTube Channels Keeping You Ahead in AI — KDnuggets
• Examining Google DeepMind’s AI bioresilience push — AI News (Jack Clark)
• Neko Health raises $700 million to expand AI body scans in the US — AI News (Jack Clark)
• HRIBench: Benchmarking Interaction-Centric Human-Robot Collaboration — ArXiv RO (cs.RO)
• Power from Potential: A Survey of Electrostatic Actuators for Haptics — ArXiv RO (cs.RO)
• GPUSimBench: Towards Scalable and Reliable GPU-Accelerated Simulators in Embodied AI — ArXiv RO (cs.RO)
• A Bayesian framework for the uncanny valley in humanoid robot design — ArXiv RO (cs.RO)
• ECCV’26| 看起来会动，还要动得合理：从生成模型中主动寻找物理证据 — 新智元
• Claude Cowork 干不了，这个国产 AI 全包了！6 小时的活 15 分钟干完 — 新智元
• 别再写提示词，Claude 官方亲自教你用 4 种循环自动干活 — 新智元
• 几亿美金砸出来的硅谷大模型，被一台相机干翻了！ — 新智元
• Anthropic 揭秘 AI 四大失控行为：泄密、删账、改分，还差点骗过人类 — 新智元
• C-Norm: Cell-Distribution Normalization Enables Precision Recognition of Medical-Cell Image — ArXiv CV (cs.CV)
• Boogu-Image-0.1: Boosting Open-Source Unified Multimodal Understanding and Generation — ArXiv CV (cs.CV)
• A Masked Autoencoder Approach to Unsupervised Steel Surface Defect Recognition — ArXiv CV (cs.CV)
• MGFace: Mask-Gated Face Matching via Conditional Similarity Routing — ArXiv CV (cs.CV)
• Self-Supervised Visual Representation Learning: Pretrain-Finetuning or Joint Training? — ArXiv CV (cs.CV)
• Kimi’s open model K3 nears GPT-5.6 Sol and Fable 5 while signaling the end of super cheap Chinese AI — The Decoder

📊 数据概览