AI资讯日报 - 2025/9/8

👨‍🔬 Yuqi Pan, Yupeng Feng, Jinghao Zhuang, Siyu Ding, Zehao Liu, Bohan Sun, Yuhong Chou, Han Xu, Xuerui Qiu, Anlin Deng, Anjie Hu, Peng Zhou, Man Yao, Jibin Wu, Jian Yang, Guoliang Sun, Bo Xu, Guoqi Li

Scaling Performance of Large Language Model Pretraining

学术论文 ArXiv 重要度: 8

系统分析大语言模型预训练的扩展性能，提供分布式训练和大规模数据集管理的实用建议。

👨‍🔬 Alexander Interrante-Grant, Carla Varela-Rosa, Suhaas Narayan, Chris Connelly, Albert Reuther

RapidGNN: Energy and Communication-Efficient Distributed Training on Large-Scale Graph Neural Networks

学术论文 ArXiv 重要度: 8

提出高效分布式图神经网络训练框架，显著提升训练吞吐量和能效，减少通信开销。

👨‍🔬 Arefin Niam, Tevfik Kosar, M S Q Zulkar Nine

CURE: Controlled Unlearning for Robust Embeddings -- Mitigating Conceptual Shortcuts in Pre-Trained Language Models

学术论文 ArXiv 重要度: 8

提出轻量级框架，系统性消除预训练语言模型中的概念捷径偏差，提升模型鲁棒性和公平性。

👨‍🔬 Aysenur Kocak, Shuo Yang, Bardh Prenkaj, Gjergji Kasneci

WinT3R: Window-Based Streaming Reconstruction with Camera Token Pool

学术论文 ArXiv 重要度: 7

提出在线3D重建模型，通过滑动窗口机制和相机令牌池实现高质量实时重建和相机姿态估计。

👨‍🔬 Zizun Li, Jianjun Zhou, Yifan Wang, Haoyu Guo, Wenzheng Chang, Yang Zhou, Haoyi Zhu, Junyi Chen, Chunhua Shen, Tong He

Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining

学术论文 ArXiv 重要度: 7

使用稀疏交叉编码器追踪大语言模型预训练中语言特征的演化过程，提升模型可解释性分析。

👨‍🔬 Deniz Bayazit, Aaron Mueller, Antoine Bosselut

LatticeWorld: A Multimodal Large Language Model-Empowered Framework for Interactive Complex World Generation

学术论文 ArXiv 重要度: 7

提出多模态大语言模型驱动的3D世界生成框架，大幅提升工业级环境生成效率和质量。

👨‍🔬 Yinglin Duan, Zhengxia Zou, Tongwei Gu, Wei Jia, Zhan Zhao, Luyi Xu, Xinzhu Liu, Hao Jiang, Kang Chen, Shuang Qiu

Enhancing 3D Point Cloud Classification with ModelNet-R and Point-SkipNet

学术论文 ArXiv 重要度: 7

提出改进版ModelNet-R数据集和轻量级Point-SkipNet网络，显著提升3D点云分类精度和效率。

👨‍🔬 Mohammad Saeid, Amir Salarpour, Pedram MohajerAnsari

COGITAO: A Visual Reasoning Framework To Study Compositionality & Generalization

学术论文 ArXiv 重要度: 6

构建模块化视觉推理框架，系统研究组合性和泛化能力，揭示现有视觉模型的组合泛化缺陷。

👨‍🔬 Yassine Taoudi-Benchekroun, Klim Troyan, Pascal Sager, Stefan Gerber, Lukas Tuggener, Benjamin Grewe

Recomposer: Event-roll-guided generative audio editing

学术论文 ArXiv 重要度: 6

开发基于事件滚动转录的生成式音频编辑系统，支持复杂场景中单个声音事件的删除、插入和增强。

👨‍🔬 Daniel P. W. Ellis, Eduardo Fonseca, Ron J. Weiss, Kevin Wilson, Scott Wisdom, Hakan Erdogan, John R. Hershey, Aren Jansen, R. Channing Moore, Manoj Plakal

Uncertain but Useful: Leveraging CNN Variability into Data Augmentation

学术论文 ArXiv 重要度: 6

研究CNN训练变异性，将其转化为数据增强策略，提升神经影像分析的鲁棒性和下游应用性能。

👨‍🔬 Inés Gonzalez-Pepe, Vinuyan Sivakolunthu, Yohan Chatelain, Tristan Glatard

🤖 AI资讯日报

📊 今日趋势总结

Why Boring Businesses Outlast AI Hype Cycles

Ask HN: Anyone concerned about NYC Local Law 144?

NLP, AI, ML, bots – a passing trend or much more? What's your take on this?

Ask HN: What's the pain using current AI algorithms?

Ask HN: Is the rate of progress in AI exponential?

The AI Crackpot Index

The Next Bill Gates or Albert Einstein in AI “Chris Clark” – Yourobot

Ask HN: What would you read to learn about "artificial intelligence"?

Ask HN: Dipping my toes with artificial intelligence and what to expect? (CS)

Common Lisp + Machine Learning Internship at Google (Mountain View, CA)

Bioinformatician

Show HN: Startup Raising capital through Book Sales

HoPE: Hyperbolic Rotary Positional Encoding for Stable Long-Range Dependency Modeling in Large Language Models

SpikingBrain Technical Report: Spiking Brain-inspired Large Models

Scaling Performance of Large Language Model Pretraining

RapidGNN: Energy and Communication-Efficient Distributed Training on Large-Scale Graph Neural Networks

CURE: Controlled Unlearning for Robust Embeddings -- Mitigating Conceptual Shortcuts in Pre-Trained Language Models

WinT3R: Window-Based Streaming Reconstruction with Camera Token Pool

Crosscoding Through Time: Tracking Emergence & Consolidation Of Linguistic Representations Throughout LLM Pretraining

LatticeWorld: A Multimodal Large Language Model-Empowered Framework for Interactive Complex World Generation

Enhancing 3D Point Cloud Classification with ModelNet-R and Point-SkipNet

COGITAO: A Visual Reasoning Framework To Study Compositionality & Generalization

Recomposer: Event-roll-guided generative audio editing

Uncertain but Useful: Leveraging CNN Variability into Data Augmentation

📅 历史日报目录