AI可可AI生活

fly51fly
  • 5.0 (1)
  • TECHNOLOGY
  • UPDATED DAILY

来自 @爱可可-爱生活 的第一手AI快报，用最通俗的语言，聊最前沿的人工智能科研进展～ #人工智能# #科技前沿#

  1. 10H AGO

    [人人能懂] AI的“小灶、泥人与神级助理”

    00:00:29 解锁AI推理能力：少即是多的秘密 00:05:17 AI巨兽的“成长配方”：如何让小个子智慧指导大块头？ 00:10:03 AI作画的“临门一脚”：如何让快马跑得又快又好？ 00:14:05 AI 创造 3D 模型：从“盖房子”到“捏泥人” 00:07:44 AI界的“小灶”，如何喂出绝顶高手？ 本期介绍的五篇论文： [LG] Part I: Tricks or Traps? A Deep Dive into RL for LLM Reasoning   [Alibaba Group]   https://arxiv.org/abs/2508.08221  --- [LG] μ-Parametrization for Mixture of Experts   [University of Warsaw]   https://arxiv.org/abs/2508.09752  --- [LG] Noise Hypernetworks: Amortizing Test-Time Compute in Diffusion Models   [Technical University of Munich & Inceptive]   https://arxiv.org/abs/2508.09968  --- [CV] VertexRegen: Mesh Generation with Continuous Level of Detail   [Meta Reality Labs Research & UC San Diego]   https://arxiv.org/abs/2508.09062  --- [LG] Beyond Scaling Law: A Data-Efficient Distillation Framework for Reasoning   [Zhongxing Telecom Equipment (ZTE)]   https://arxiv.org/abs/2508.09883  --- [CL] Speed Always Wins: A Survey on Efficient Architectures for Large Language Models   [Unknown / Survey paper]   https://arxiv.org/abs/2508.09834

    23 min

  2. 22H AGO

    [人间有AI] AI与记忆幻觉：你所坚信的过去，也许从未发生

    真正的强大，不是拥有一个不会犯错的记忆，而是拥有重写故事的勇气。

    4 min

  3. 1D AGO

    [人人能懂] AI的内功心法：少即是多，巧胜于拙

    00:00:38 给AI的大脑做减法：一种聪明的“偷懒”智慧 00:04:34 你的数据正在“减肥”：人工智能时代的生存新法则 00:09:23 AI的“闭关修炼”：不给它网络，它能有多强？ 00:13:07 给AI装上一个“语法”导航仪 00:16:49 AI的“通感”：为什么“我写的”和“我是作者”是一回事？ 本期介绍的几篇论文： [LG] XQuant: Breaking the Memory Wall for LLM Inference with KV Cache Rematerialization   [UC Berkeley & FuriosaAI]   https://arxiv.org/abs/2508.10395 --- [LG] SoK: Data Minimization in Machine Learning   [ETH Zurich]   https://arxiv.org/abs/2508.10836 --- [CL] SSRL: Self-Search Reinforcement Learning   [Tsinghua University & Shanghai AI Laboratory]   https://arxiv.org/abs/2508.10874 --- [LG] Constrained Decoding of Diffusion LLMs with Context-Free Grammars   [ETH Zurich]   https://arxiv.org/abs/2508.10111 --- [CL] A Rose by Any Other Name Would Smell as Sweet: Categorical Homotopy Theory for Large Language Models   [Adobe Research]   https://arxiv.org/abs/2508.10018

    21 min

  4. 1D AGO

    [人间有AI] 你的“负面情绪”，是AI渴望的超能力

    当机器在逻辑、计算和效率上无限逼近甚至超越我们时，我们真正的“超能力”，是体验、是感受，是那份允许万物穿过身体的勇气和智慧。

    5 min

  5. 2D AGO

    [人人能懂] 给AI装上“学霸脑”与“专家外挂”

    00:00:32 让AI学会“学习”：不再死记硬背的秘密  00:04:30 给AI装个“外挂”：一个叫“记忆解码器”的U盘 00:08:23 开会吵不完？AI教你找共识 00:13:39 大海捞针？不如自己造根针 00:18:41 AI减肥记：如何让聪明的机器不说废话？ 本期介绍的五篇论文： [CL] Learning Facts at Scale with Active Reading   [Meta FAIR]   https://arxiv.org/abs/2508.09494   --- [CL] Memory Decoder: A Pretrained, Plug-and-Play Memory for Large Language Models   [Shanghai Jiao Tong University]   https://arxiv.org/abs/2508.09874   --- [CL] The Human-AI Hybrid Delphi Model: A Structured Framework for Context-Rich, Expert Consensus in Complex Domains   [Google Research]   https://arxiv.org/abs/2508.09349   --- [LG] SYNAPSE-G: Bridging Large Language Models and Graph Learning for Rare Event Classification   [Google Research]   https://arxiv.org/abs/2508.09544   --- [CL] Sample More to Think Less: Group Filtered Policy Optimization for Concise Reasoning   [Microsoft Research]   https://arxiv.org/abs/2508.09726

    23 min

  6. 2D AGO

    [人间有AI] 别再优化了，先「上线」再说：AI 给我们的第一课

    无论是人类学习一项新技能，还是AI模型的训练，第一步，永远不是追求“卓越”，而是确保“在场”。

    4 min

  7. 3D AGO

    [人人能懂] 从模仿到洞察：AI如何学会像侦探一样思考？

    00:00:34 你以为AI是高科技？它的底层逻辑可能几亿年前就有了 00:05:07 秒懂的秘密：如何看透行为背后的“算法” 00:09:28 AI大模型里的“快递小哥”：不只求快，还要求稳 00:14:22 AI的“双档切换”：如何用最聪明的大脑，干最划算的活？ 00:18:24 我们如何思考“如果”：一种全新的思维工具 本期介绍的五篇论文： [LG] Understanding Transformers through the Lens of Pavlovian Conditioning   [Meta Platforms, Inc]   https://arxiv.org/abs/2508.08289   --- [RO] Rational Inverse Reasoning   [MIT CSAIL]   https://arxiv.org/abs/2508.08983   --- [LG] Scaled-Dot-Product Attention as One-Sided Entropic Optimal Transport   [Stanford University]   https://arxiv.org/abs/2508.08369   --- [LG] OverFill: Two-Stage Models for Efficient Language Model Decoding   [Cornell University]   https://arxiv.org/abs/2508.08446   --- [LG] Topos Causal Models   [Adobe Research]   https://arxiv.org/abs/2508.08295

    23 min

  8. 4D AGO

    [人间有AI] 算法之外，寻找你的“同频共振”

    大大方方地做自己吧，带着你所有的“不完美”和“异常值”。

    4 min
See All (605)

About

来自 @爱可可-爱生活 的第一手AI快报，用最通俗的语言，聊最前沿的人工智能科研进展～ #人工智能# #科技前沿#

Information

  • Creator
    fly51fly
  • Years Active
    2024 - 2025
  • Episodes
    605
  • Rating
    Clean
  • Copyright
    © @爱可可-爱生活
  • Show Website
    AI可可AI生活

You Might Also Like