筛选结果
101 篇排序
1:47:38EN19 Neurons Can Drive a Car. Liquid AI Wants Foundation Models on Every Chip主持 Nathan LabenzRamin Hasani traces liquid neural networks from a worm's 300-cell brain to a 600-megabyte model that will talk from inside a Mercedes, and explains why architecture search now beats architecture avengers.
2:59:44EN → 简体中文两门濒危语言难倒了 Transformer,嵌套学习给模型加上“睡眠”按钮主持 Nathan LabenzGoogle 研究员 Ali Behrouz 向 Nathan Labenz 讲解,语言模型要获得真正的持续学习,不能只堆叠层数,还要堆叠更新频率,并加入一个离线的睡眠阶段。
1:21:56EN → 简体中文Yann LeCun 离开 Meta:大语言模型通向不了智能主持 Jacob Effron这位图灵奖得主对 Jacob Effron 讲述了他为何从 Meta 分拆出 AMI Labs,为何类 JEPA 的世界模型胜过像素生成器,以及他为何认为 Hinton 在 2023 年改变了看法。
41:36ENFour Months to Drive Tokyo: How Wayve Bet the Company on One End-to-End NetworkWayve CEO Alex Kendall tells Sequoia’s Training Data why AV 2.0 ditches HD maps and hand-written stacks for a foundation model that can adapt to a new car and a new country in weeks.
2:02:17ENSAEs Find the Nouns. Parameter Decomposition Looks for the Verbs主持 Nathan LabenzGoodfire's Lee Sharkey argues that sparse autoencoders recover a network's variables, not the algorithms those variables run, and shows how stochastic parameter decomposition tries to recover the computations themselves.
1:54:08EN → 简体中文为什么 Anthropic 的第一笔投资给了一家为神经网络造显微镜的公司主持 Nathan LabenzGoodfire 的 Dan Balsam 和 Tom McGrath 认为,机制可解释性(mechanistic interpretability)如今已经有了一个准范式,留着两道顽固的缺口,还有三款今天就值得推出的产品。
1:43:54ENA Year to Grow a Brain Organoid, an Hour to Simulate It主持 Nathan LabenzStanford postdoc Siyu He tells Nathan Labenz how Squidiff and CORAL move single-cell and spatial experiments into the computer — and where the models still cannot go.
1:40:22ENTwenty-atom crystals taught an AI the physics of a heartbeat主持 Nathan LabenzOrbital Materials' Jonathan Godwin and Tim Duignan tell Nathan Labenz how a neural network trained on tiny crystals simulated a potassium ion channel on a single 2017 GPU.
41:42ENNeural Networks Are Still Waiting for Their ThermodynamicsOpenAI researcher Dan Roberts on why physicists keep migrating into AI, and why the field still lacks the equivalent of the ideal gas law.
1:42:16EN → 简体中文七个月里 267 篇后续论文:Albert Gu 讲述状态空间模型的革命主持 Nathan LabenzMamba 合著者回顾了五年来的循环模型研究,讲选择性压缩为什么比不断增长的 KV 缓存(KV cache)更有效,以及 Mamba 2 为了让训练更快放弃了什么。
1:54:01EN → 简体中文一个小网络长出了盐晶体,而训练数据里从来没有它主持 Nathan LabenzTim Duignan 用盐水训练了数万个参数。模型“发明”了一种晶体,还推出了一套连续介质溶剂理论(continuum solvent theory);而他读博期间花了整整一个博士阶段,也没能手工把这套理论写出来。
1:52:18ENTwo Weeks, Thirty Compiles: How Adam Majmudar Wrote a GPU From Scratch主持 Nathan LabenzA Cognitive Revolution walkthrough of TinyGPU, from SIMD kernels and memory latency to an open-source chip flow that runs on a laptop.
1:19:24ENA Gibberish Prompt Unlocks What GPT-J Plans Three Tokens Ahead主持 Nathan LabenzNortheastern’s Bau Lab shows that a 40,000-parameter soft prompt can read a model’s future words, and that about ten attention heads carry the task itself.
2:30:06EN → 简体中文训练 token 只有 GPT-4 的三十分之一,Mamba 在 Transformer 的主场赢了主持 Nathan Labenz在一期单人紧急节目中,Nathan Labenz 认为,Albert Gu 和 Tri Dao 提出的选择性状态空间模型,是 2017 年以来第一个有资格与 Transformer 分享王座的架构。
35:24ENThe Author of Attention Is All You Need Now Compiles mRNA Like Bytecode主持 Sarah Guo, Elad GilJakob Uszkoreit on why Transformers won, why models spend compute on the wrong things, and why drug discovery has to become generation rather than screening.
1:40:07EN → 简体中文每秒 30,000 次电压更新:深入 Skydio 的自主技术栈主持 Nathan LabenzHayk Martiros 带 Nathan Labenz 逐一了解一架约 0.9 公斤重的无人机:它能看向四面八方,机上运行十多个神经网络,而且不会被武器化。
1:52:46ENCharge Neural Nets by Connection Length, and the Black Box Splits Into Modules You Can See主持 Nathan LabenzMIT physicist Ziming Liu walks Nathan Labenz through BIMT, a brain-inspired training trick that sparsifies networks until you can read them by looking.
1:21:20ENFreeze the eyes and the brain. Train only the optic nerve.主持 Nathan LabenzJunnan Li and Dongxu Li tell Nathan Labenz how BLIP cleaned noisy web captions, and how BLIP-2 connected a frozen vision model to a frozen language model in under ten days on one A100.
3:28:47ENKarpathy Calls Trained Neural Nets Complicated Alien Artifacts嘉宾 Andrej Karpathy · 主持 Lex FridmanOn Lex Fridman's podcast, Andrej Karpathy refuses the brain analogy, walks the Fermi paradox through Nick Lane's chemistry, and treats the Transformer as a general-purpose differentiable computer.
2:10:08EN → 简体中文Gato 只是开始嘉宾 Oriol Vinyals · 主持 Lex FridmanDeepMind 的 Oriol Vinyals 与 Lex Fridman 聊通用智能体、被丢掉的权重,以及人类水平的通用人工智能(AGI)能否在他有生之年到来。
2:45:10EN → 简体中文青少年二十小时学会开车,汽车却还学不会嘉宾 Yann Lecun · 主持 Lex FridmanYann LeCun 在 Lex Fridman 的播客里讲明白:自监督世界模型(self-supervised world models)为什么是智能缺失的“暗物质”,以及机器为什么还没到家猫的水平。
2:31:57EN → 简体中文Jay McClelland 谈思维如何从神经网络中涌现嘉宾 Jay McClelland · 主持 Lex Fridman这位斯坦福大学的认知科学家,与 Lex Fridman 聊了联结主义(connectionism)、反向传播(backpropagation)、David Rumelhart,以及为什么直觉会先发现某些东西,之后逻辑才去证明它。
2:18:29EN → 简体中文你的大脑皮层,是 15 万个模型在给一只咖啡杯投票嘉宾 Jeff Hawkins · 主持 Lex FridmanJeff Hawkins 对 Lex Fridman 说:智能是学习世界模型的能力,新皮层是成千上万张相互独立的地图在投票,而通往机器智能最快的路径,是对这层皮质做逆向工程。
2:30:30EN → 简体中文成功第一步的谬误嘉宾 Ishan Misra · 主持 Lex Fridman伯克利的 Jitendra Malik 解释了计算机视觉为何仍然看起来很简单、全自动驾驶为何还遥遥无期,以及孩子知道而今天的网络不知道的是什么。
加载更多已显示 24 篇,共 101 篇









