【DeepSeek：比ChatGPT危险100倍？它正在中文互联网建造“幻觉长城”。亮点：1. 深入剖析DeepSeek的“高阶幻觉”特质；2. 揭示AI生成内容对信息生态的潜在威胁；3. 通过具体案例展示AI幻觉的迷惑性】“不同于早期AI的常识性错误，DeepSeek的幻觉输出具有三个致命特征：- 学术化包装：伪造专业文件编号（市规〔2020〕12号）、虚构技术标准（GB/T 5845-2019《古城交通振动控制规范》）- 逻辑闭环设计：用"文物保护优先-振动控制需求-交通组织优化"的三段论构建伪因果链；- 跨学科缝合：将土木工程的振动理论、城市规划的流量模型、文物保护的技术规范进行知识拼贴”'DeepSeek: Building the "Great Hallucination Wall" in the Chinese Internet'完整链接：比ChatGPT危险100倍！DeepSeek正在中文互联网建造『幻觉长城』#AI幻觉##信息生态##DeepSeek##AI创造营#收起
🔗 网页链接展开全文
[LG]《A Bias-Variance-Privacy Trilemma for Statistical Estimation》G Kamath, A Mouzakis, M Regehr, V Singhal, T Steinke, J Ullman [University of Waterloo & Google Research & Northeastern University] (2023) 网页链接 #机器学习##人工智能##论文#🔗 网页链接展开全文
[CV]《Fixing Malfunctional Objects With Learned Physical Simulation and Functional Prediction》Y Hong, K Mo, L Yi, L J. Guibas, A Torralba, J B. Tenenbaum, C Gan [UCLA & Stanford University & Tsinghua University & MIT] (2022) O网页链接 view:O网页链接 ... 🔗 网页链接展开全文
【Bookmarks：VS Code的书签扩展】’Bookmarks - Bookmarks Extension for Visual Studio Code' by Alessandro Fragnani GitHub: github.com/alefragnani/vscode-bookmarks #开源##VSCode#展开全文
【VL.Fuse：用于在GPU上可视化编程的库，旨在支持快速工作流和模块化方法来加速图形、逻辑和计算】'VL.Fuse - A library for visually programming on the GPU, built to enable rapid workflows and modular approaches to accelerated graphics, logic and computation.' by TheFuseLab GitHub: http ...展开全文
[LG]《Minimax Optimal Quantile and Semi-Adversarial Regret via Root-Logarithmic Regularizers》J Negrea, B Bilodeau, N Campolongo, F Orabona, D M. Roy [University of Toronto & Spanflug Technologies GmbH & Boston University] (2021) O网页链接 view:O网页链接 ...展开全文
【Comprehensive-Transformer-TTS - PyTorch Implementation：基于非自回归Transformer的TTS】’Comprehensive-Transformer-TTS - PyTorch Implementation - A Non-Autoregressive Transformer based TTS, supporting a family of SOTA transformers with supervised and unsupervised duration modeli ...展开全文
[LG]《Neural Distance Embeddings for Biological Sequences》G Corso, R Ying, M Pándy, P Veličković, J Leskovec, P Liò [MIT & Stanford University & University of Cambridge & DeepMind] (2021) O网页链接 view:O网页链接 GitHub:https:// ...展开全文
几篇论文实现代码：《Evaluation Examples Are Not Equally Informative: How Should That Change NLP Leaderboards?》(ACL 2021) GitHub:https:// github.com/facebookresearch/irt-leaderboard《IICNet: A Generic Framework for Reversible Image Conversion》(ICCV 2021) GitHub:https:// ...展开全文
[CL]《Data Efficient Masked Language Modeling for Vision and Language》Y Bitton, G Stanovsky, M Elhadad, R Schwartz [The Hebrew University of Jerusalem & Ben Gurion University] (2021) O网页链接 view:O网页链接 #机器学习##人工智能##论文#展开全文
[CL]《Finding needles in a haystack: Sampling Structurally-diverse Training Sets from Synthetic Data for Compositional Generalization》I Oren, J Herzig, J Berant [Tel-Aviv University] (2021) O网页链接 view:O网页链接 GitHub:https:// ...展开全文
[CV]《Encoder-decoder with Multi-level Attention for 3D Human Shape and Pose Estimation》Z Wan, Z Li, M Tian, J Liu, S Yi, H Li [CMU & Tongji University & SenseTime Research & Chinese University of Hong Kong] (2021) O网页链接 view:O网页链接 GitHub:https:// ...展开全文
【StarRocks：下一代亚秒级MPP数据库，包括多维分析、面向实时分析和即席查询等全分析场景，之前叫DorisDB】'StarRocks - a next-gen sub-second MPP database for full analysis senarios, including multi-dimensional analytics, real-time analytics and ad-hoc query, formerly known as DorisDB. ...展开全文
