AI可可AI生活

fly51fly
  • 5.0(1개의 평가)
  • 과학 기술
  • 매일 업데이트

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. 4시간 전

    [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

    23분

  2. 14시간 전

    [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”。

    4분

  3. 1일 전

    [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

    23분

  4. 1일 전

    [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”。

    4분

  5. 2일 전

    [人人能懂] AI思考的五种新姿势：它如何学会了“死磕”、“变通”与“偷懒”？

    00:00:34 AI侦探养成记：如何让机器学会“死磕到底”？ 00:04:22 AI也需要“元学习”：如何打造一把能开万能锁的钥匙？ 00:07:56 拆解AI大脑：它如何学会“绕个弯”解决问题？ 00:12:21 AI学会“举一反三”的秘密：两层楼就够了？ 00:16:34 AI思考的秘密：为什么“少”就是“多”？ 本期介绍的五篇论文： [CL] Beyond Ten Turns: Unlocking Long-Horizon Agentic Search with Large-Scale Asynchronous RL   [Tsinghua University]   https://arxiv.org/abs/2508.079   --- [LG] AdaptFlow: Adaptive Workflow Optimization via Meta-Learning   [Peking University & University of Chinese Academy of Sciences]   https://arxiv.org/abs/2508.08053   --- [LG] Multi-head Transformers Provably Learn Symbolic Multi-step Reasoning via Gradient Descent   [CMU & UPenn & OSU]   https://arxiv.org/abs/2508.08222   --- [LG] What One Cannot, Two Can: Two-Layer Transformers Provably Represent Induction Heads on Any-Order Markov Chains   [MIT & EPFL & UC Berkeley]   https://arxiv.org/abs/2508.07208   --- [CL] Less Is More: Training-Free Sparse Attention with Global Locality for Efficient Reasoning   [Princeton University & CMU]   https://arxiv.org/abs/2508.07101

    21분

  6. 2일 전

    [人间有AI] 一个顶尖AI工程师的秘密：你不是在思考，而是在“编程”你的现实

    那个能一键“扭转乾坤”的开发者，不是别人，就是你自己。

    5분

  7. 3일 전

    [人人能懂] 解锁AI新维度：思考、行动、创造、自知

    00:00:30 AI界的“学霸”是怎么炼成的？ 00:04:45 你的下一个AI，为什么必须是个“行动派”？ 00:08:35 AI当上帝：从零开始创造一门语言 00:13:09 如何给AI大模型“摸骨”，看透它的知识边界？ 本期介绍的四篇论文： [CL] UR²: Unify RAG and Reasoning through Reinforcement Learning   [Tsinghua University & Hebei University of Economics and Business]   https://arxiv.org/abs/2508.06165  --- [CL] GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models   [Zhipu AI & Tsinghua University]   https://arxiv.org/abs/2508.06471  --- [CL] ConlangCrafter: Constructing Languages with a Multi-Hop LLM Pipeline   [Tel Aviv University & UC Berkeley]   https://arxiv.org/abs/2508.06094  --- [CL] Efficient Knowledge Probing of Large Language Models by Adapting Pre-trained Embeddings   [Georgia Institute of Technology & MIT]   https://arxiv.org/abs/2508.06030

    18분

  8. 3일 전

    [人间有AI] 你的对手，是你亲手训练的AI

    你最强大的对手，和你最伟大的盟友，都是你。

    4분
모두 보기(601개)

소개

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#

정보

  • 제작진
    fly51fly
  • 방송 연도
    2024년 - 2025년
  • 에피소드
    601
  • 등급
    전체 연령 사용가
  • 저작권
    © @爱可可-爱生活
  • 웹사이트 보기
    AI可可AI生活

좋아할 만한 다른 항목