NVIDIA (Jensen Huang / AI infrastructure)
你收到的每一条AI回复，最初都始于一个电子⚡
黄仁勋近期接受红杉资本采访时将AI基础设施比作五层蛋糕：在芯片、数据中心、模型之下的最底层是能源。智能时代的核心约束并非算力、软件，而是电力。英伟达将AI工厂称作当代发电机，输入电子、输出智能token，这场人类史上最大规模基建已投数千亿美元，仍需数万亿跟进，核心问题是电网能否跟上AI扩容节奏⬆️
AI Signal Daily Digest
AI Signal 日报
追踪 AI 一线声音：做事的人、写代码的人、下注的人。今天重点从 GitHub 源池拉取 X / Twitter、播客和 arXiv，保留原文链接，页面只放可快速阅读的卡片摘要。
26X / Twitter 内容
18播客条目
0含全文字幕播客
30arXiv 论文
提示：summaries feed generated_at is older than 30 hours。
X / Twitter
保留原文链接，按信号强度筛选播客
优先使用 transcript，有链接才纳入今天暂无可展示的新内容
源池已读取，但这一栏没有通过筛选的高质量中文摘要。
论文
arXiv feed 可能略旧，保留论文链接SPADE: Self-Play in Adaptive Synthetic Executable Environments
Continuous self-improvement requires an ever-expanding pool of self-generated, diverse, adaptive goals.
ADEPT: Accelerating Dexterity via Pre-Training and Post-Training using Reinforce...
We introduce Accelerating Dexterity via Pre-Training (ADEPT), a large-scale reinforcement learning (RL) framework for learning sim-to-real transferable dexterity across high degree-of-freedom (DoF) robot embodiments that can solve long-horizon tasks directly from raw visuo-tactile perception. ADEPT pretrains a dexterous policy on a generic object reposing ta...
Beyond Teacher Likelihood: Group-Calibrated On-Policy Distillation for Long-Cont...
On-policy distillation (OPD) trains a student on its own responses using dense token-level guidance from a stronger teacher.
Finetuning Strategies for Querying Sounds by Vocal Imitation
This technical report describes our winning submission to the AES AIMLA 2025 Challenge on querying sound effects by vocal imitation.
Lévy Attention: Single-Pass Predictive Uncertainty for Continuous-Time Attention
Deep models for irregularly-sampled time series answer queries at arbitrary continuous timestamps, yet report nothing about how far each answer should be trusted.
Learned, Then Lost: A Measured Single-Example Counterfactual in Pre-training
A single training example's contribution to a finished model is normally estimated rather than measured, because measuring it takes two expensive full pre-training runs that differ in one row of one batch.