Results
88 articlesSort
1:16:29EN
Jane Street· Aug 26, 2026Grant Sanderson and Alok Puranik: Positional Encodings Through Group TheoryGrant Sanderson, Alok Puranik · 3Blue1BrownStarting with assumptions about attention, they derive matrix exponentials and connect rotations, decay and Jordan blocks to RoPE and ALiBi.
1:13:33EN
Unsupervised Learning: With Jacob Effron· Jun 3, 2026PodcastA Trillion Tokens Later, the Concept Finally Lands: Łukasz Kaiser on What Transformers Still MissHost Jacob EffronThe Attention Is All You Need co-author tells Unsupervised Learning that transformers will probably keep winning — and that they still do not learn the way people do.
2:02:17EN
Cognitive Revolution "How AI Changes Everything"· Aug 27, 2025PodcastSAEs Find the Nouns. Parameter Decomposition Looks for the VerbsHost 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.
1:37:05ENCognitive Revolution "How AI Changes Everything"· Jun 19, 2025PodcastAn Algebraic Geometer Quit Tenure Because the Bowl-Shaped Loss Plot Is a LieHost Nathan LabenzJesse Hoogland and Daniel Murfet of Timaeus argue that singularities in the loss landscape — not the smooth valleys in every slide deck — are what actually govern how neural networks learn, generalize, and go wrong.
2:12:00ENCognitive Revolution "How AI Changes Everything"· May 15, 2025PodcastGradient Descent at Test Time: Titans Pushes LLM Context to 10 Million TokensHost Nathan LabenzA Cornell PhD student's neural memory module holds about 70% accuracy where GPT-4's scores fall off a cliff.
2:33:39ENCognitive Revolution "How AI Changes Everything"· Feb 27, 2025PodcastThe Attacker Goes Second: Why Nicholas Carlini Keeps Breaking AI DefensesHost Nathan LabenzA Google DeepMind security researcher on simple loss functions, 70% robust accuracy, and why the fix has to live in the system, not the model.
1:40:22ENCognitive Revolution "How AI Changes Everything"· Jan 22, 2025PodcastTwenty-atom crystals taught an AI the physics of a heartbeatHost 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.
1:06:23ENJane Street· Oct 14, 2024Sylvain Gugger on Accelerating ML Models and Reproducing ResultsYaron Minsky, Sylvain Gugger · Jane StreetFrom learning-rate schedules to GPU bottlenecks and noisy market data, Sylvain Gugger explains how he approaches practical machine-learning performance.
1:42:16ENCognitive Revolution "How AI Changes Everything"· Jul 4, 2024Podcast267 Papers in Seven Months: Albert Gu Explains the State Space Model RevolutionHost Nathan LabenzThe Mamba co-author walks through five years of recurrent-model work, why selective compression beats a growing KV cache, and what Mamba 2 trades away to train faster.
1:54:01ENCognitive Revolution "How AI Changes Everything"· Jun 26, 2024PodcastA Tiny Network Grew a Salt Crystal the Training Data Never ShowedHost Nathan LabenzTim Duignan trained tens of thousands of parameters on salt water. The model invented a crystal — and a continuum solvent theory he had spent a PhD failing to write by hand.
1:58:43ENCognitive Revolution "How AI Changes Everything"· Feb 7, 2024PodcastA Worm With 302 Neurons Already Makes Trade-offs. AI Is Still on the Same Ladder.Host Nathan LabenzMax Bennett maps 600 million years of brain evolution onto five breakthroughs — and asks which rungs today's models have actually reached.
1:19:24ENCognitive Revolution "How AI Changes Everything"· Nov 17, 2023PodcastA Toy Paper Did Not Prove LLMs Cannot Generalize. The Discourse Still Ran With It.Host Nathan LabenzNathan Labenz and Erik Torenberg unpack a viral DeepMind result, a weekend reproduction, and a premature war over voluntary AI commitments.
43:45EN
No Priors: AI, Machine Learning, Tech, & Startups· Aug 3, 2023Podcast · 2 speakersEight Years of No Growth, Then OpenAI and NVIDIA Became His CustomersHost Sarah Guo, Elad GilWeights & Biases CEO Lukas Biewald tells No Priors how a labeling company with no chasm to cross funded the ML developer tools that trained most of the large language models.
1:33:56ENCognitive Revolution "How AI Changes Everything"· Jul 6, 2023PodcastThe AI Scouting Report: 90 Minutes From GPT-4's Exam Scores to Whether Models UnderstandHost Nathan LabenzNathan Labenz spends a solo episode walking through what general-purpose models can do, how they are trained, how capabilities unlock, and whether they understand anything at all.
1:32:22ENCognitive Revolution "How AI Changes Everything"· Jun 29, 2023PodcastA Million Bytes and 256 Symbols: Meta's MEGABYTE Wants to Throw the Tokenizer AwayHost Nathan LabenzLili Yu of Meta AI explains a multiscale transformer that predicts million-byte sequences, skips tokenization, and aims to model any file as raw bytes.
1:52:46ENCognitive Revolution "How AI Changes Everything"· Jun 27, 2023PodcastCharge Neural Nets by Connection Length, and the Black Box Splits Into Modules You Can SeeHost 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.
1:25:50ENCognitive Revolution "How AI Changes Everything"· Apr 18, 2023PodcastPrompt Engineering Became Programming, and Riley Goodside Is Fine With ThatHost Nathan LabenzScale AI’s first staff prompt engineer on tool use, fine-tuning without amnesia, generate-then-rank, and why the 2023 job-of-the-future story never arrived.
46:57ENNo Priors: AI, Machine Learning, Tech, & Startups· Mar 2, 2023Podcast · 2 speakers95% of Drug Programs Fail. Daphne Koller Is Betting Machine Learning on the First DecisionHost Sarah Guo, Elad GilA new medicine still takes more than a decade and well over a billion dollars — and Insitro's founder argues the expensive failures start with the wrong target, indication, or patient.
2:31:57EN
Lex Fridman· Sep 20, 2021Podcast · 2 speakersJay McClelland on How Thought Emerges From Neural NetworksGuest Jay McClelland · Host Lex FridmanThe Stanford cognitive scientist talks with Lex Fridman about connectionism, backpropagation, David Rumelhart, and why intuition discovers what logic later proves.
2:51:33ENLex Fridman· Aug 29, 2021Podcast · 2 speakersThe Self Is a GPT Prompt: Wojciech Zaremba on Consciousness, Codex, and Fermi's SilenceGuest Wojciech Zaremba · Host Lex FridmanOpenAI co-founder Wojciech Zaremba tells Lex Fridman he thinks we might be alone in the universe — and that GPT-3, Codex, and a one-handed robot already hint at how intelligence is built.
2:18:29ENLex Fridman· Aug 8, 2021Podcast · 2 speakersYour Cortex Is 150,000 Brains Voting on a Coffee CupGuest Jeff Hawkins · Host Lex FridmanJeff Hawkins tells Lex Fridman that intelligence is learning a model of the world, the neocortex is tens of thousands of independent maps that vote, and the fastest path to machine intelligence is reverse-engineering that sheet.
2:30:30ENLex Fridman· Jul 31, 2021Podcast · 2 speakersThe Fallacy of the Successful First StepGuest Ishan Misra · Host Lex FridmanBerkeley's Jitendra Malik explains why computer vision still looks easy, why fully autonomous driving is not near, and what children know that today's networks do not.
1:56:17ENLex Fridman· Apr 19, 2021Podcast · 2 speakersIf You Reran Earth a Million Times, Would Anything Like Us Appear?Guest Risto Miikkulainen · Host Lex FridmanRisto Miikkulainen on neuroevolution, hyena raids, basil that never sleeps, and why variation-plus-selection still finds solutions backpropagation cannot see.
1:49:56ENLex Fridman· Feb 7, 2021Podcast · 2 speakersA 21-Year-Old at a Party Called GameStop. He Was Right.Guest Richard Craib · Host Lex FridmanNumerai founder Richard Craib talks with Lex Fridman about WallStreetBets, crowdsourced machine learning, and why finance is a machine for pulling the future into the present.Load moreShowing 24 of 88