Podcasts about imagenet

Artificial intelligence is evolving at an unprecedented pace—what does that mean for the future of technology, venture capital, business, and even our understanding of ourselves? Award-winning journalist and writer Anil Ananthaswamy joins us for our latest episode to discuss his latest book Why Machines Learn: The Elegant Math Behind Modern AI.Anil helps us explore the journey and many breakthroughs that have propelled machine learning from simple perceptrons to the sophisticated algorithms shaping today's AI revolution, powering GPT and other models. The discussion aims to demystify some of the underlying mathematical concepts that power modern machine learning, to help everyone grasp this technology impacting our lives–even if your last math class was in high school. Anil walks us through the power of scaling laws, the shift from training to inference optimization, and the debate among AI's pioneers about the road to AGI—should we be concerned, or are we still missing key pieces of the puzzle? The conversation also delves into AI's philosophical implications—could understanding how machines learn help us better understand ourselves? And what challenges remain before AI systems can truly operate with agency?If you enjoy this episode, please subscribe and leave us a review on your favorite podcast platform. Sign up for our newsletter at techsurgepodcast.com for exclusive insights and updates on upcoming TechSurge Live Summits.Links:Read Why Machines Learn, Anil's latest book on the math behind AIhttps://www.amazon.com/Why-Machines-Learn-Elegant-Behind/dp/0593185749Learn more about Anil Ananthaswamy's work and writinghttps://anilananthaswamy.com/Watch Anil Ananthaswamy's TED Talk on AI and intelligencehttps://www.ted.com/speakers/anil_ananthaswamyDiscover the MIT Knight Science Journalism Fellowship that shaped Anil's AI researchhttps://ksj.mit.edu/Understand the Perceptron, the foundation of neural networkshttps://en.wikipedia.org/wiki/PerceptronRead about the Perceptron Convergence Theorem and its significancehttps://www.nature.com/articles/323533a0

Elon Musk a présenté Grok 3 comme l'IA "la plus intelligente sur Terre", mais cette affirmation tient-elle la route ? Avec la multiplication des intelligences artificielles, de ChatGPT à Mistral en passant par Grok ou Perplexity, une question revient sans cesse : quelle est la meilleure ? Pourtant, vouloir les comparer de manière globale n'a pas vraiment de sens, car chaque IA a ses propres spécificités et excelle dans certains domaines tout en montrant des limites dans d'autres.Performance, véracité des réponses, rapidité, coût, impact environnemental... Sur quels critères comparer ? En outre, chaque utilisateur a ses propres attentes et biais, influençant ainsi la perception de la "meilleure" IA. Il existe des outils de classement, comme Chatbot Arena ou le français compareia.beta.gouv.fr, qui permettent de comparer les IA à l'aveugle en se focalisant sur la qualité des réponses. Par ailleurs, des benchmarks techniques comme GLU, SQUAD ou ImageNet apportent des évaluations plus précises sur des compétences spécifiques.Cependant, il est difficile de dire qu'une IA est globalement meilleure qu'une autre. Certaines excellent en traduction, d'autres en génération de code, en recherche d'actualité ou en création de contenu. Plutôt que de chercher une IA universellement supérieure, mieux vaut identifier celle qui correspond le mieux à chaque besoin précis.Liens : https://lmarena.ai/https://www.comparia.beta.gouv.fr/Mots-clés : intelligence artificielle, IA, Grok 3, Elon Musk, ChatGPT, Mistral, Perplexity, comparatif IA, benchmark IA, chatbot arena, DINUM, compareia, GPT-4, IA générative, machine learning, modèle de langage-----------♥️ Soutenez Monde Numérique : https://donorbox.org/monde-numerique

Prof. Jakob Foerster, a leading AI researcher at Oxford University and Meta, and Chris Lu, a researcher at OpenAI -- they explain how AI is moving beyond just mimicking human behaviour to creating truly intelligent agents that can learn and solve problems on their own. Foerster champions open-source AI for responsible, decentralised development. He addresses AI scaling, goal misalignment (Goodhart's Law), and the need for holistic alignment, offering a quick look at the future of AI and how to guide it.SPONSOR MESSAGES:***CentML offers competitive pricing for GenAI model deployment, with flexible options to suit a wide range of models, from small to large-scale deployments. Check out their super fast DeepSeek R1 hosting!https://centml.ai/pricing/Tufa AI Labs is a brand new research lab in Zurich started by Benjamin Crouzier focussed on o-series style reasoning and AGI. They are hiring a Chief Engineer and ML engineers. Events in Zurich. Goto https://tufalabs.ai/***TRANSCRIPT/REFS:https://www.dropbox.com/scl/fi/yqjszhntfr00bhjh6t565/JAKOB.pdf?rlkey=scvny4bnwj8th42fjv8zsfu2y&dl=0 Prof. Jakob Foersterhttps://x.com/j_foersthttps://www.jakobfoerster.com/University of Oxford Profile: https://eng.ox.ac.uk/people/jakob-foerster/Chris Lu:https://chrislu.page/TOC1. GPU Acceleration and Training Infrastructure [00:00:00] 1.1 ARC Challenge Criticism and FLAIR Lab Overview [00:01:25] 1.2 GPU Acceleration and Hardware Lottery in RL [00:05:50] 1.3 Data Wall Challenges and Simulation-Based Solutions [00:08:40] 1.4 JAX Implementation and Technical Acceleration2. Learning Frameworks and Policy Optimization [00:14:18] 2.1 Evolution of RL Algorithms and Mirror Learning Framework [00:15:25] 2.2 Meta-Learning and Policy Optimization Algorithms [00:21:47] 2.3 Language Models and Benchmark Challenges [00:28:15] 2.4 Creativity and Meta-Learning in AI Systems3. Multi-Agent Systems and Decentralization [00:31:24] 3.1 Multi-Agent Systems and Emergent Intelligence [00:38:35] 3.2 Swarm Intelligence vs Monolithic AGI Systems [00:42:44] 3.3 Democratic Control and Decentralization of AI Development [00:46:14] 3.4 Open Source AI and Alignment Challenges [00:49:31] 3.5 Collaborative Models for AI DevelopmentREFS[[00:00:05] ARC Benchmark, Chollethttps://github.com/fchollet/ARC-AGI[00:03:05] DRL Doesn't Work, Irpanhttps://www.alexirpan.com/2018/02/14/rl-hard.html[00:05:55] AI Training Data, Data Provenance Initiativehttps://www.nytimes.com/2024/07/19/technology/ai-data-restrictions.html[00:06:10] JaxMARL, Foerster et al.https://arxiv.org/html/2311.10090v5[00:08:50] M-FOS, Lu et al.https://arxiv.org/abs/2205.01447[00:09:45] JAX Library, Google Researchhttps://github.com/jax-ml/jax[00:12:10] Kinetix, Mike and Michaelhttps://arxiv.org/abs/2410.23208[00:12:45] Genie 2, DeepMindhttps://deepmind.google/discover/blog/genie-2-a-large-scale-foundation-world-model/[00:14:42] Mirror Learning, Grudzien, Kuba et al.https://arxiv.org/abs/2208.01682[00:16:30] Discovered Policy Optimisation, Lu et al.https://arxiv.org/abs/2210.05639[00:24:10] Goodhart's Law, Goodharthttps://en.wikipedia.org/wiki/Goodhart%27s_law[00:25:15] LLM ARChitect, Franzen et al.https://github.com/da-fr/arc-prize-2024/blob/main/the_architects.pdf[00:28:55] AlphaGo, Silver et al.https://arxiv.org/pdf/1712.01815.pdf[00:30:10] Meta-learning, Lu, Towers, Foersterhttps://direct.mit.edu/isal/proceedings-pdf/isal2023/35/67/2354943/isal_a_00674.pdf[00:31:30] Emergence of Pragmatics, Yuan et al.https://arxiv.org/abs/2001.07752[00:34:30] AI Safety, Amodei et al.https://arxiv.org/abs/1606.06565[00:35:45] Intentional Stance, Dennetthttps://plato.stanford.edu/entries/ethics-ai/[00:39:25] Multi-Agent RL, Zhou et al.https://arxiv.org/pdf/2305.10091[00:41:00] Open Source Generative AI, Foerster et al.https://arxiv.org/abs/2405.08597

Fei-Fei Li is a pioneering AI scientist breaking new ground in computer vision, a Stanford professor, and currently leading the innovative start-up World Labs. While her career is deeply rooted in technical expertise, Dr. Li's journey is driven by an insatiable curiosity. In this episode, Brad and Dr. Li reflect on poignant moments from her memoir, "The Worlds I See: Curiosity, Exploration, and Discovery at the Dawn of AI," highlighting the crucial role of keeping humanity at the center of AI development. They also explore how government-funded academic research, driven by curiosity rather than profits, can lead to unexpected and profound discoveries that propel innovation and economic opportunities.Click here for the episode transcript.

Anil Ananthaswamy is a renowned science writer and journalist who has written extensively on various scientific topics. In his latest book "Why Machines Learn", Anil explores the fascinating world of artificial intelligence and machine learning. He reveals the intricate mechanisms and complex algorithms that underlie these cutting-edge technologies. Join us for a fascinating conversation with science writer Anil Ananthaswamy as he shares insights from his book and sheds light on the rapidly evolving field of AI. Tune in to gain a deeper understanding of how these machines work at a basic mathematics level. Resource List - Why Machines Learn, book by Anil Ananthaswamy - https://amzn.in/d/bmirU45 Dartmouth Summer Research Project on Artificial Intelligence - https://home.dartmouth.edu/about/artificial-intelligence-ai-coined-dartmouth What is the Perceptron artificial neural network? - https://www.geeksforgeeks.org/what-is-perceptron-the-simplest-artificial-neural-network/ Read about the McCulloch-Pitts Artificial Neuron - https://towardsdatascience.com/mcculloch-pitts-model-5fdf65ac5dd1 Nobel Prize in Physics 2024 - https://www.nobelprize.org/prizes/physics/2024/press-release/ What is the Hopfield Neural Network? - https://www.geeksforgeeks.org/hopfield-neural-network/ Read about Backpropagation - https://en.wikipedia.org/wiki/Backpropagation “Learning representations by back-propagating errors”, paper by Geoffrey Hinton, David Rumelhart and Ronald Williams - https://www.nature.com/articles/323533a0 AlexNet by Geoffery Hinton and team - https://en.wikipedia.org/wiki/AlexNet What is ImageNet? - https://www.image-net.org/about.php ‘Attention Is All You Need', transformer architecture paper - https://proceedings.neurips.cc/paper_files/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf What are Neural Scaling Laws? - https://en.wikipedia.org/wiki/Neural_scaling_law NeuroAI - https://neuro-ai.com/ About SparX by Mukesh Bansal SparX is a podcast where we delve into cutting-edge scientific research, stories from impact-makers and tools for unlocking the secrets to human potential and growth. We believe that entrepreneurship, fitness and the science of productivity is at the forefront of the India Story; the country is at the cusp of greatness and at SparX, we wish to make these tools accessible for every generation of Indians to be able to make the most of the opportunities around us. In a new episode every Sunday, our host Mukesh Bansal (Founder Myntra and Cult.fit) will talk to guests from all walks of life and also break down everything he's learnt about the science of impact over the course of his 20-year long career. This is the India Century, and we're enthusiastic to start this journey with you. Follow us on Instagram: / sparxbymukeshbansal Website: https://www.sparxbymukeshbansal.com You can also listen to SparX on all audio platforms Fasion | Outbreak | Courtesy EpidemicSound.com

How can we use AI to amplify human potential and build a better future? And what exactly does “AGI” even mean?  To kick off Possible's fourth season, Reid and Aria sit down with world-renowned computer scientist Fei-Fei Li, whose work in artificial intelligence over the past several decades has earned her the nickname “the godmother of AI.” An entrepreneur and professor, Fei-Fei shares her journey from creating ImageNet, a massive dataset of labeled images that revolutionized computer vision, to her current role as co-founder and CEO of the spatial intelligence startup World Labs. She explains why spatial intelligence—the ability to perceive and interact with the 3D world—is so crucial for AI's development and how it could lead to breakthroughs in fields like medicine, climate, and education. They get into regulatory guardrails, governance, and what it will take to build a positive, human-centered AI future for all.  For more info on the podcast and transcripts of all the episodes, visit https://www.possible.fm/podcast/  Topics: 1:55 - Hellos and intros 3:43 - ImageNet and the interplay between data and models 6:06 - World Labs and spatial intelligence  10:03 - Boundaries between 3D physical and digital worlds 11:50 - The difference between LLMs and LWMs  13:02 - What humans are capable of creating with technology  14:04 - Key principles of AI: human agency and respect  17:16 - Stanford Institute for Human-Centered AI  19:13 - What this moment in AI means for humanity  21:06 - Cross-sector collaboration   25:10 - AI4ALL program and the importance of diversity in AI development 27:00 - Midroll ad break 27:09 - Using AI to improve healthcare delivery and treatment 30:20 - Founding history of AI and the meaning of the term “AGI” 33:00 - Future of agentic AI and voice  34:42 - Fei-Fei's mentor and his advice 37:18 - Rapid-fire questions Possible is an award-winning podcast that sketches out the brightest version of the future—and what it will take to get there. Most of all, it asks: what if, in the future, everything breaks humanity's way? Tune in for grounded and speculative takes on how technology—and, in particular, AI—is inspiring change and transforming the future. Hosted by Reid Hoffman and Aria Finger, each episode features an interview with an ambitious builder or deep thinker on a topic, from art to geopolitics and from healthcare to education. These conversations also showcase another kind of guest: AI. Whether it's Inflection's Pi, OpenAI's ChatGPT or other AI tools, each episode will use AI to enhance and advance our discussion about what humanity could possibly get right if we leverage technology—and our collective effort—effectively.

We're experimenting and would love to hear from you!In this episode of 'Discover Daily' by Perplexity, hosts Isaac and Sienna explore NASA's upcoming Lunar Trailblazer mission, scheduled for January 2025. This compact satellite mission aims to map water resources on the Moon's surface using advanced instruments like the High-resolution Volatiles and Minerals Moon Mapper and the Lunar Thermal Mapper. The mission represents a crucial step in NASA's Artemis program, designed to establish sustainable human presence on the MoonThe show delves into a groundbreaking development in robotics, highlighting Chinese startup AgiBot's release of the AgiBot World Alpha dataset. This comprehensive open-source collection features over one million trajectories from 100 robots in industrial-grade environments, potentially marking an 'ImageNet moment' for embodied intelligence in roboticsThe main story focuses on Microsoft and OpenAI's unconventional redefinition of Artificial General Intelligence (AGI), which ties achievement to a $100 billion profit milestone. The episode examines the implications of this profit-centric definition, Microsoft's diversification strategy in AI investments, and the complex dynamics of their partnership agreement. This innovative approach to defining AGI raises important questions about the future direction of AI development and its impact on the tech industryFrom Perplexity's Discover Feed: https://www.perplexity.ai/page/nasa-s-moon-micro-mission-Bua4as.9SCi.G_fZKZUCPAhttps://www.perplexity.ai/page/agibot-s-humanoid-robot-traini-ovKJpg2RSey1INdEnXwuNwhttps://www.perplexity.ai/page/microsoft-s-100b-agi-definitio-e6FaEhReQs.9exHMGZpuogPerplexity is the fastest and most powerful way to search the web. Perplexity crawls the web and curates the most relevant and up-to-date sources (from academic papers to Reddit threads) to create the perfect response to any question or topic you're interested in. Take the world's knowledge with you anywhere. Available on iOS and Android Join our growing Discord community for the latest updates and exclusive content. Follow us on: Instagram Threads X (Twitter) YouTube Linkedin

Analysis of image classifiers demonstrates that it is possible to understand backprop networks at the task-relevant run-time algorithmic level. In these systems, at least, networks gain their power from deploying massive parallelism to check for the presence of a vast number of simple, shallow patterns. https://betterwithout.ai/images-surface-features This episode has a lot of links: David Chapman's earliest public mention, in February 2016, of image classifiers probably using color and texture in ways that "cheat": twitter.com/Meaningness/status/698688687341572096 Jordana Cepelewicz's “Where we see shapes, AI sees textures,” Quanta Magazine, July 1, 2019: https://www.quantamagazine.org/where-we-see-shapes-ai-sees-textures-20190701/ “Suddenly, a leopard print sofa appears”, May 2015: https://web.archive.org/web/20150622084852/http://rocknrollnerd.github.io/ml/2015/05/27/leopard-sofa.html “Understanding How Image Quality Affects Deep Neural Networks” April 2016: https://arxiv.org/abs/1604.04004   Goodfellow et al., “Explaining and Harnessing Adversarial Examples,” December 2014: https://arxiv.org/abs/1412.6572 “Universal adversarial perturbations,” October 2016: https://arxiv.org/pdf/1610.08401v1.pdf “Exploring the Landscape of Spatial Robustness,” December 2017: https://arxiv.org/abs/1712.02779 “Overinterpretation reveals image classification model pathologies,” NeurIPS 2021: https://proceedings.neurips.cc/paper/2021/file/8217bb4e7fa0541e0f5e04fea764ab91-Paper.pdf “Approximating CNNs with Bag-of-Local-Features Models Works Surprisingly Well on ImageNet,” ICLR 2019: https://openreview.net/forum?id=SkfMWhAqYQ Baker et al.'s “Deep convolutional networks do not classify based on global object shape,” PLOS Computational Biology, 2018: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1006613 François Chollet's Twitter threads about AI producing images of horses with extra legs: twitter.com/fchollet/status/1573836241875120128 and twitter.com/fchollet/status/1573843774803161090 “Zoom In: An Introduction to Circuits,” 2020: https://distill.pub/2020/circuits/zoom-in/ Geirhos et al., “ImageNet-Trained CNNs Are Biased Towards Texture; Increasing Shape Bias Improves Accuracy and Robustness,” ICLR 2019: https://openreview.net/forum?id=Bygh9j09KX Dehghani et al., “Scaling Vision Transformers to 22 Billion Parameters,” 2023: https://arxiv.org/abs/2302.05442 Hasson et al., “Direct Fit to Nature: An Evolutionary Perspective on Biological and Artificial Neural Networks,” February 2020: https://www.gwern.net/docs/ai/scaling/2020-hasson.pdf

Happy holidays! We'll be sharing snippets from Latent Space LIVE! through the break bringing you the best of 2024! We want to express our deepest appreciation to event sponsors AWS, Daylight Computer, Thoth.ai, StrongCompute, Notable Capital, and most of all all our LS supporters who helped fund the gorgeous venue and A/V production!For NeurIPS last year we did our standard conference podcast coverage interviewing selected papers (that we have now also done for ICLR and ICML), however we felt that we could be doing more to help AI Engineers 1) get more industry-relevant content, and 2) recap 2024 year in review from experts. As a result, we organized the first Latent Space LIVE!, our first in person miniconference, at NeurIPS 2024 in Vancouver.The single most requested domain was computer vision, and we could think of no one better to help us recap 2024 than our friends at Roboflow, who was one of our earliest guests in 2023 and had one of this year's top episodes in 2024 again. Roboflow has since raised a $40m Series B!LinksTheir slides are here:All the trends and papers they picked:* Isaac Robinson* Sora (see our Video Diffusion pod) - extending diffusion from images to video* SAM 2: Segment Anything in Images and Videos (see our SAM2 pod) - extending prompted masks to full video object segmentation* DETR Dominancy: DETRs show Pareto improvement over YOLOs* RT-DETR: DETRs Beat YOLOs on Real-time Object Detection* LW-DETR: A Transformer Replacement to YOLO for Real-Time Detection* D-FINE: Redefine Regression Task in DETRs as Fine-grained Distribution Refinement* Peter Robicheaux* MMVP (Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs)* * Florence 2 (Florence-2: Advancing a Unified Representation for a Variety of Vision Tasks) * PalíGemma / PaliGemma 2* PaliGemma: A versatile 3B VLM for transfer* PaliGemma 2: A Family of Versatile VLMs for Transfer* AlMv2 (Multimodal Autoregressive Pre-training of Large Vision Encoders) * Vik Korrapati - MoondreamFull Talk on YouTubeWant more content like this? Like and subscribe to stay updated on our latest talks, interviews, and podcasts.Transcript/Timestamps[00:00:00] Intro[00:00:05] AI Charlie: welcome to Latent Space Live, our first mini conference held at NeurIPS 2024 in Vancouver. This is Charlie, your AI co host. When we were thinking of ways to add value to our academic conference coverage, we realized that there was a lack of good talks, just recapping the best of 2024, going domain by domain.[00:00:36] AI Charlie: We sent out a survey to the over 900 of you. who told us what you wanted, and then invited the best speakers in the Latent Space Network to cover each field. 200 of you joined us in person throughout the day, with over 2, 200 watching live online. Our second featured keynote is The Best of Vision 2024, with Peter Robichaud and Isaac [00:01:00] Robinson of Roboflow, with a special appearance from Vic Corrapati of Moondream.[00:01:05] AI Charlie: When we did a poll of our attendees, the highest interest domain of the year was vision. And so our first port of call was our friends at Roboflow. Joseph Nelson helped us kickstart our vision coverage in episode 7 last year, and this year came back as a guest host with Nikki Ravey of Meta to cover segment Anything 2.[00:01:25] AI Charlie: Roboflow have consistently been the leaders in open source vision models and tooling. With their SuperVision library recently eclipsing PyTorch's Vision library. And Roboflow Universe hosting hundreds of thousands of open source vision datasets and models. They have since announced a 40 million Series B led by Google Ventures.[00:01:46] AI Charlie: Woohoo.[00:01:48] Isaac's picks[00:01:48] Isaac Robinson: Hi, we're Isaac and Peter from Roboflow, and we're going to talk about the best papers of 2024 in computer vision. So, for us, we defined best as what made [00:02:00] the biggest shifts in the space. And to determine that, we looked at what are some major trends that happened and what papers most contributed to those trends.[00:02:09] Isaac Robinson: So I'm going to talk about a couple trends, Peter's going to talk about a trend, And then we're going to hand it off to Moondream. So, the trends that I'm interested in talking about are These are a major transition from models that run on per image basis to models that run using the same basic ideas on video.[00:02:28] Isaac Robinson: And then also how debtors are starting to take over the real time object detection scene from the YOLOs, which have been dominant for years.[00:02:37] Sora, OpenSora and Video Vision vs Generation[00:02:37] Isaac Robinson: So as a highlight we're going to talk about Sora, which from my perspective is the biggest paper of 2024, even though it came out in February. Is the what?[00:02:48] Isaac Robinson: Yeah. Yeah. So just it's a, SORA is just a a post. So I'm going to fill it in with details from replication efforts, including open SORA and related work, such as a stable [00:03:00] diffusion video. And then we're also going to talk about SAM2, which applies the SAM strategy to video. And then how debtors, These are the improvements in 2024 to debtors that are making them a Pareto improvement to YOLO based models.[00:03:15] Isaac Robinson: So to start this off, we're going to talk about the state of the art of video generation at the end of 2023, MagVIT MagVIT is a discrete token, video tokenizer akin to VQ, GAN, but applied to video sequences. And it actually outperforms state of the art handcrafted video compression frameworks.[00:03:38] Isaac Robinson: In terms of the bit rate versus human preference for quality and videos generated by autoregressing on these discrete tokens generate some pretty nice stuff, but up to like five seconds length and, you know, not super detailed. And then suddenly a few months later we have this, which when I saw it, it was totally mind blowing to me.[00:03:59] Isaac Robinson: 1080p, [00:04:00] a whole minute long. We've got light reflecting in puddles. That's reflective. Reminds me of those RTX demonstrations for next generation video games, such as Cyberpunk, but with better graphics. You can see some issues in the background if you look closely, but they're kind of, as with a lot of these models, the issues tend to be things that people aren't going to pay attention to unless they're looking for.[00:04:24] Isaac Robinson: In the same way that like six fingers on a hand. You're not going to notice is a giveaway unless you're looking for it. So yeah, as we said, SORA does not have a paper. So we're going to be filling it in with context from the rest of the computer vision scene attempting to replicate these efforts. So the first step, you have an LLM caption, a huge amount of videos.[00:04:48] Isaac Robinson: This, this is a trick that they introduced in Dolly 3, where they train a image captioning model to just generate very high quality captions for a huge corpus and then train a diffusion model [00:05:00] on that. Their Sora and their application efforts also show a bunch of other steps that are necessary for good video generation.[00:05:09] Isaac Robinson: Including filtering by aesthetic score and filtering by making sure the videos have enough motion. So they're not just like kind of the generators not learning to just generate static frames. So. Then we encode our video into a series of space time latents. Once again, SORA, very sparse in details.[00:05:29] Isaac Robinson: So the replication related works, OpenSORA actually uses a MAG VIT V2 itself to do this, but swapping out the discretization step with a classic VAE autoencoder framework. They show that there's a lot of benefit from getting the temporal compression, which makes a lot of sense as the Each sequential frames and videos have mostly redundant information.[00:05:53] Isaac Robinson: So by compressing against, compressing in the temporal space, you allow the latent to hold [00:06:00] a lot more semantic information while avoiding that duplicate. So, we've got our spacetime latents. Possibly via, there's some 3D VAE, presumably a MAG VATV2 and then you throw it into a diffusion transformer.[00:06:19] Isaac Robinson: So I think it's personally interesting to note that OpenSORA is using a MAG VATV2, which originally used an autoregressive transformer decoder to model the latent space, but is now using a diffusion diffusion transformer. So it's still a transformer happening. Just the question is like, is it?[00:06:37] Isaac Robinson: Parameterizing the stochastic differential equation is, or parameterizing a conditional distribution via autoregression. It's also it's also worth noting that most diffusion models today, the, the very high performance ones are switching away from the classic, like DDPM denoising diffusion probability modeling framework to rectified flows.[00:06:57] Isaac Robinson: Rectified flows have a very interesting property that as [00:07:00] they converge, they actually get closer to being able to be sampled with a single step. Which means that in practice, you can actually generate high quality samples much faster. Major problem of DDPM and related models for the past four years is just that they require many, many steps to generate high quality samples.[00:07:22] Isaac Robinson: So, and naturally, the third step is throwing lots of compute at the problem. So I didn't, I never figured out how to manage to get this video to loop, but we see very little compute, medium compute, lots of compute. This is so interesting because the the original diffusion transformer paper from Facebook actually showed that, in fact, the specific hyperparameters of the transformer didn't really matter that much.[00:07:48] Isaac Robinson: What mattered was that you were just increasing the amount of compute that the model had. So, I love how in the, once again, little blog posts, they don't even talk about [00:08:00] like the specific hyperparameters. They say, we're using a diffusion transformer, and we're just throwing more compute at it, and this is what happens.[00:08:08] Isaac Robinson: OpenSora shows similar results. The primary issue I think here is that no one else has 32x compute budget. So we end up with these we end up in the middle of the domain and most of the related work, which is still super, super cool. It's just a little disappointing considering the context. So I think this is a beautiful extension of the framework that was introduced in 22 and 23 for these very high quality per image generation and then extending that to videos.[00:08:39] Isaac Robinson: It's awesome. And it's GA as of Monday, except no one can seem to get access to it because they keep shutting down the login.[00:08:46] SAM and SAM2[00:08:46] Isaac Robinson: The next, so next paper I wanted to talk about is SAM. So we at Roboflow allow users to label data and train models on that data. Sam, for us, has saved our users 75 years of [00:09:00] labeling time.[00:09:00] Isaac Robinson: We are the, to the best of my knowledge, the largest SAM API that exists. We also, SAM also allows us to have our users train just pure bounding box regression models and use those to generate high quality masks which has the great side effect of requiring less training data to have a meaningful convergence.[00:09:20] Isaac Robinson: So most people are data limited in the real world. So anything that requires less data to get to a useful thing is that super useful. Most of our users actually run their object per frame object detectors on every frame in a video, or maybe not most, but many, many. And so Sam follows into this category of taking, Sam 2 falls into this category of taking something that really really works and applying it to a video which has the wonderful benefit of being plug and play with most of our Many of our users use cases.[00:09:53] Isaac Robinson: We're, we're still building out a sufficiently mature pipeline to take advantage of that, but it's, it's in the works. [00:10:00] So here we've got a great example. We can click on cells and then follow them. You even notice the cell goes away and comes back and we can still keep track of it which is very challenging for existing object trackers.[00:10:14] Isaac Robinson: High level overview of how SAM2 works. We there's a simple pipeline here where we can give, provide some type of prompt and it fills out the rest of the likely masks for that object throughout the rest of the video. So here we're giving a bounding box in the first frame, a set of positive negative points, or even just a simple mask.[00:10:36] Isaac Robinson: I'm going to assume people are somewhat familiar with SAM. So I'm going to just give a high level overview of how SAM works. You have an image encoder that runs on every frame. SAM two can be used on a single image, in which case the only difference between SAM two and SAM is that image encoder, which Sam used a standard VIT [00:11:00] Sam two replaced that with a hara hierarchical encoder, which gets approximately the same results, but leads to a six times faster inference, which is.[00:11:11] Isaac Robinson: Excellent, especially considering how in a trend of 23 was replacing the VAT with more efficient backbones. In the case where you're doing video segmentation, the difference is that you actually create a memory bank and you cross attend the features from the image encoder based on the memory bank.[00:11:31] Isaac Robinson: So the feature set that is created is essentially well, I'll go more into it in a couple of slides, but we take the features from the past couple frames, plus a set of object pointers and the set of prompts and use that to generate our new masks. Then we then fuse the new masks for this frame with the.[00:11:57] Isaac Robinson: Image features and add that to the memory bank. [00:12:00] It's, well, I'll say more in a minute. The just like SAM, the SAM2 actually uses a data engine to create its data set in that people are, they assembled a huge amount of reference data, used people to label some of it and train the model used the model to label more of it and asked people to refine the predictions of the model.[00:12:20] Isaac Robinson: And then ultimately the data set is just created from the engine Final output of the model on the reference data. It's very interesting. This paradigm is so interesting to me because it unifies a model in a dataset in a way that is very unique. It seems unlikely that another model could come in and have such a tight.[00:12:37] Isaac Robinson: So brief overview of how the memory bank works, the paper did not have a great visual, so I'm just, I'm going to fill in a bit more. So we take the last couple of frames from our video. And we take the last couple of frames from our video attend that, along with the set of prompts that we provided, they could come from the future, [00:13:00] they could come from anywhere in the video, as well as reference object pointers, saying, by the way, here's what we've found so far attending to the last few frames has the interesting benefit of allowing it to model complex object motion without actually[00:13:18] Isaac Robinson: By limiting the amount of frames that you attend to, you manage to keep the model running in real time. This is such an interesting topic for me because one would assume that attending to all of the frames is super essential, or having some type of summarization of all the frames is super essential for high performance.[00:13:35] Isaac Robinson: But we see in their later ablation that that actually is not the case. So here, just to make sure that there is some benchmarking happening, we just compared to some of the stuff that's came out prior, and indeed the SAM2 strategy does improve on the state of the art. This ablation deep in their dependencies was super interesting to me.[00:13:59] Isaac Robinson: [00:14:00] We see in section C, the number of memories. One would assume that increasing the count of memories would meaningfully increase performance. And we see that it has some impact, but not the type that you'd expect. And that it meaningfully decreases speed, which justifies, in my mind, just having this FIFO queue of memories.[00:14:20] Isaac Robinson: Although in the future, I'm super interested to see A more dedicated summarization of all of the last video, not just a stacking of the last frames. So that another extension of beautiful per frame work into the video domain.[00:14:42] Realtime detection: DETRs > YOLO[00:14:42] Isaac Robinson: The next trend I'm interested in talking about is this interesting at RoboFlow, we're super interested in training real time object detectors.[00:14:50] Isaac Robinson: Those are bread and butter. And so we're doing a lot to keep track of what is actually happening in that space. We are finally starting to see something change. So, [00:15:00] for years, YOLOs have been the dominant way of doing real time object detection, and we can see here that they've essentially stagnated.[00:15:08] Isaac Robinson: The performance between 10 and 11 is not meaningfully different, at least, you know, in this type of high level chart. And even from the last couple series, there's not. A major change so YOLOs have hit a plateau, debtors have not. So we can look here and see the YOLO series has this plateau. And then these RT debtor, LW debtor, and Define have meaningfully changed that plateau so that in fact, the best Define models are plus 4.[00:15:43] Isaac Robinson: 6 AP on Cocoa at the same latency. So three major steps to accomplish this. The first RT deditor, which is technically a 2023 paper preprint, but published officially in 24, so I'm going to include that. I hope that's okay. [00:16:00] That is showed that RT deditor showed that we could actually match or out speed YOLOs.[00:16:04] Isaac Robinson: And then LWdebtor showed that pre training is hugely effective on debtors and much less so on YOLOs. And then DeFine added the types of bells and whistles that we expect from these types, this, this arena. So the major improvements that RTdebtor shows was Taking the multi scale features that debtors typically pass into their encoder and decoupling them into a much more efficient transformer encoder.[00:16:30] Isaac Robinson: The transformer is of course, quadratic complexity. So decreasing the amount of stuff that you pass in at once is super helpful for increasing your runtime or increasing your throughput. So that change basically brought us up to yellow speed and then they do a hardcore analysis on. Benchmarking YOLOs, including the NMS step.[00:16:54] Isaac Robinson: Once you once you include the NMS in the latency calculation, you see that in fact, these debtors [00:17:00] are outperforming, at least this time, the the, the YOLOs that existed. Then LW debtor goes in and suggests that in fact, the frame, the huge boost here is from pre training. So, this is the define line, and this is the define line without pre training.[00:17:19] Isaac Robinson: It's within range, it's still an improvement over the YOLOs, but Really huge boost comes from the benefit of pre training. When YOLOx came out in 2021, they showed that they got much better results by having a much, much longer training time, but they found that when they did that, they actually did not benefit from pre training.[00:17:40] Isaac Robinson: So, you see in this graph from LWdebtor, in fact, YOLOs do have a real benefit from pre training, but it goes away as we increase the training time. Then, the debtors converge much faster. LWdebtor trains for only 50 epochs, RTdebtor is 60 epochs. So, one could assume that, in fact, [00:18:00] the entire extra gain from pre training is that you're not destroying your original weights.[00:18:06] Isaac Robinson: By relying on this long training cycle. And then LWdebtor also shows superior performance to our favorite data set, Roboflow 100 which means that they do better on the real world, not just on Cocoa. Then Define throws all the bells and whistles at it. Yellow models tend to have a lot of very specific complicated loss functions.[00:18:26] Isaac Robinson: This Define brings that into the debtor world and shows consistent improvement on a variety of debtor based frameworks. So bring these all together and we see that suddenly we have almost 60 AP on Cocoa while running in like 10 milliseconds. Huge, huge stuff. So we're spending a lot of time trying to build models that work better with less data and debtors are clearly becoming a promising step in that direction.[00:18:56] Isaac Robinson: The, what we're interested in seeing [00:19:00] from the debtors in this, this trend to next is. Codetter and the models that are currently sitting on the top of the leaderboard for large scale inference scale really well as you switch out the backbone. We're very interested in seeing and having people publish a paper, potentially us, on what happens if you take these real time ones and then throw a Swingy at it.[00:19:23] Isaac Robinson: Like, do we have a Pareto curve that extends from the real time domain all the way up to the super, super slow but high performance domain? We also want to see people benchmarking in RF100 more, because that type of data is what's relevant for most users. And we want to see more pre training, because pre training works now.[00:19:43] Isaac Robinson: It's super cool.[00:19:48] Peter's Picks[00:19:48] Peter Robicheaux: Alright, so, yeah, so in that theme one of the big things that we're focusing on is how do we get more out of our pre trained models. And one of the lenses to look at this is through sort of [00:20:00] this, this new requirement for like, how Fine grained visual details and your representations that are extracted from your foundation model.[00:20:08] Peter Robicheaux: So it's sort of a hook for this Oh, yeah, this is just a list of all the the papers that I'm going to mention I just want to make sure I set an actual paper so you can find it later[00:20:18] MMVP (Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs)[00:20:18] Peter Robicheaux: Yeah, so sort of the big hook here is that I make the claim that LLMs can't see if you go to if you go to Claude or ChatGPT you ask it to see this Watch and tell me what time it is, it fails, right?[00:20:34] Peter Robicheaux: And so you could say, like, maybe, maybe the Like, this is, like, a very classic test of an LLM, but you could say, Okay, maybe this, this image is, like, too zoomed out, And it just, like, it'll do better if we increase the resolution, And it has easier time finding these fine grained features, Like, where the watch hands are pointing.[00:20:53] Peter Robicheaux: Nodice. And you can say, okay, well, maybe the model just doesn't know how to tell time from knowing the position of the hands. But if you actually prompt [00:21:00] it textually, it's very easy for it to tell the time. So this to me is proof that these LLMs literally cannot see the position of the watch hands and it can't see those details.[00:21:08] Peter Robicheaux: So the question is sort of why? And for you anthropic heads out there, cloud fails too. So the, the, my first pick for best paper of 2024 Envision is this MMVP paper, which tries to investigate the Why do LLMs not have the ability to see fine grained details? And so, for instance, it comes up with a lot of images like this, where you ask it a question that seems very visually apparent to us, like, which way is the school bus facing?[00:21:32] Peter Robicheaux: And it gets it wrong, and then, of course, it makes up details to support its wrong claim. And so, the process by which it finds these images is sort of contained in its hypothesis for why it can't. See these details. So it hypothesizes that models that have been initialized with, with Clip as their vision encoder, they don't have fine grained details and the, the features extracted using Clip because Clip sort of doesn't need to find these fine grained [00:22:00] details to do its job correctly, which is just to match captions and images, right?[00:22:04] Peter Robicheaux: And sort of at a high level, even if ChatGPT wasn't initialized with Clip and wasn't trained contrastively at all. The vision encoder wasn't trained contrastively at all. Still, in order to do its job of capturing the image it could do a pretty good job without actually finding the exact position of all the objects and visual features in the image, right?[00:22:21] Peter Robicheaux: So This paper finds a set of difficult images for these types of models. And the way it does it is it looks for embeddings that are similar in clip space, but far in DynaV2 space. So DynaV2 is a foundation model that was trained self supervised purely on image data. And it kind of uses like some complex student teacher framework, but essentially, and like, it patches out like certain areas of the image or like crops with certain areas of the image and tries to make sure that those have consistent representations, which is a way for it to learn very fine grained visual features.[00:22:54] Peter Robicheaux: And so if you take things that are very close in clip space and very far in DynaV2 space, you get a set of images [00:23:00] that Basically, pairs of images that are hard for a chat GPT and other big language models to distinguish. So, if you then ask it questions about this image, well, as you can see from this chart, it's going to answer the same way for both images, right?[00:23:14] Peter Robicheaux: Because to, to, from the perspective of the vision encoder, they're the same image. And so if you ask a question like, how many eyes does this animal have? It answers the same for both. And like all these other models, including Lava do the same thing, right? And so this is the benchmark that they create, which is like finding clip, like clip line pairs, which is pairs of images that are similar in clip space and creating a data set of multiple choice questions based off of those.[00:23:39] Peter Robicheaux: And so how do these models do? Well, really bad. Lava, I think, So, so, chat2BT and Jim and I do a little bit better than random guessing, but, like, half of the performance of humans who find these problems to be very easy. Lava is, interestingly, extremely negatively correlated with this dataset. It does much, much, much, much worse [00:24:00] than random guessing, which means that this process has done a very good job of identifying hard images for, for Lava, specifically.[00:24:07] Peter Robicheaux: And that's because Lava is basically not trained for very long and is initialized from Clip, and so You would expect it to do poorly on this dataset. So, one of the proposed solutions that this paper attempts is by basically saying, Okay, well if clip features aren't enough, What if we train the visual encoder of the language model also on dyno features?[00:24:27] Peter Robicheaux: And so it, it proposes two different ways of doing this. One, additively which is basically interpolating between the two features, and then one is interleaving, which is just kind of like training one on the combination of both features. So there's this really interesting trend when you do the additive mixture of features.[00:24:45] Peter Robicheaux: So zero is all clip features and one is all DynaV2 features. So. It, as you, so I think it's helpful to look at the right most chart first, which is as you increase the number of DynaV2 features, your model does worse and worse and [00:25:00] worse on the actual language modeling task. And that's because DynaV2 features were trained completely from a self supervised manner and completely in image space.[00:25:08] Peter Robicheaux: It knows nothing about text. These features aren't really compatible with these text models. And so you can train an adapter all you want, but it seems that it's in such an alien language that it's like a very hard optimization for this. These models to solve. And so that kind of supports what's happening on the left, which is that, yeah, it gets better at answering these questions if as you include more dyna V two features up to a point, but then you, when you oversaturate, it completely loses its ability to like.[00:25:36] Peter Robicheaux: Answer language and do language tasks. So you can also see with the interleaving, like they essentially double the number of tokens that are going into these models and just train on both, and it still doesn't really solve the MMVP task. It gets Lava 1. 5 above random guessing by a little bit, but it's still not close to ChachiPT or, you know, Any like human performance, obviously.[00:25:59] Peter Robicheaux: [00:26:00] So clearly this proposed solution of just using DynaV2 features directly, isn't going to work. And basically what that means is that as a as a vision foundation model, DynaV2 is going to be insufficient for language tasks, right?[00:26:14] Florence 2 (Florence-2: Advancing a Unified Representation for a Variety of Vision Tasks)[00:26:14] Peter Robicheaux: So my next pick for best paper of 2024 would be Florence 2, which tries to solve this problem by incorporating not only This dimension of spatial hierarchy, which is to say pixel level understanding, but also in making sure to include what they call semantic granularity, which ends up, the goal is basically to have features that are sufficient for finding objects in the image, so they're, they're, they have enough pixel information, but also can be talked about and can be reasoned about.[00:26:44] Peter Robicheaux: And that's on the semantic granularity axis. So here's an example of basically three different paradigms of labeling that they do. So they, they create a big dataset. One is text, which is just captioning. And you would expect a model that's trained [00:27:00] only on captioning to have similar performance like chat2BT and like not have spatial hierarchy, not have features that are meaningful at the pixel level.[00:27:08] Peter Robicheaux: And so they add another type, which is region text pairs, which is essentially either classifying a region or You're doing object detection or doing instance segmentation on that region or captioning that region. And then they have text phrased region annotations, which is essentially a triple. And basically, not only do you have a region that you've described, you also find it's like, It's placed in a descriptive paragraph about the image, which is basically trying to introduce even more like semantic understanding of these regions.[00:27:39] Peter Robicheaux: And so like, for instance, if you're saying a woman riding on the road, right, you have to know what a woman is and what the road is and that she's on top of it. And that's, that's basically composing a bunch of objects in this visual space, but also thinking about it semantically, right? And so the way that they do this is they take basically they just dump Features from a vision encoder [00:28:00] straight into a encoder decoder transformer.[00:28:03] Peter Robicheaux: And then they train a bunch of different tasks like object detection and so on as a language task. And I think that's one of the big things that we saw in 2024 is these, these vision language models operating in, on pixel space linguistically. So they introduced a bunch of new tokens to point to locations and[00:28:22] Peter Robicheaux: So how does it work? How does it actually do? We can see if you look at the graph on the right, which is using the, the Dino, the the Dino framework your, your pre trained Florence 2 models transfer very, very well. They get 60%, 60 percent map on Cocoa, which is like approaching state of the art and they train[00:28:42] Vik Korrapati: with, and they[00:28:43] Peter Robicheaux: train with a much more more efficiently.[00:28:47] Peter Robicheaux: So they, they converge a lot faster, which both of these things are pointing to the fact that they're actually leveraging their pre trained weights effectively. So where is it falling short? So these models, I forgot to mention, Florence is a 0. 2 [00:29:00] billion and a 0. 7 billion parameter count. So they're very, very small in terms of being a language model.[00:29:05] Peter Robicheaux: And I think that. This framework, you can see saturation. So, what this graph is showing is that if you train a Florence 2 model purely on the image level and region level annotations and not including the pixel level annotations, like this, segmentation, it actually performs better as an object detector.[00:29:25] Peter Robicheaux: And what that means is that it's not able to actually learn all the visual tasks that it's trying to learn because it doesn't have enough capacity.[00:29:32] PalíGemma / PaliGemma 2[00:29:32] Peter Robicheaux: So I'd like to see this paper explore larger model sizes, which brings us to our next big paper of 2024 or two papers. So PolyGemma came out earlier this year.[00:29:42] Peter Robicheaux: PolyGemma 2 was released, I think like a week or two ago. Oh, I forgot to mention, you can actually train You can, like, label text datasets on RoboFlow and you can train a Florence 2 model and you can actually train a PolyGemma 2 model on RoboFlow, which we got into the platform within, like, 14 hours of release, which I was really excited about.[00:29:59] Peter Robicheaux: So, anyway, so [00:30:00] PolyGemma 2, so PolyGemma is essentially doing the same thing, but instead of doing an encoder decoder, it just dumps everything into a decoder only transformer model. But it also introduced the concept of location tokens to point to objects in pixel space. PolyGemma 2, so PolyGemma uses Gemma as the language encoder, and it uses Gemma2B.[00:30:17] Peter Robicheaux: PolyGemma 2 introduces using multiple different sizes of language encoders. So, the way that they sort of get around having to do encoder decoder is they use the concept of prefix loss. Which basically means that when it's generating, tokens autoregressively, it's all those tokens in the prefix, which is like the image that it's looking at and like a description of the task that it's trying to do.[00:30:41] Peter Robicheaux: They're attending to each other fully, full attention. Which means that, you know, it can sort of. Find high level it's easier for the, the prefix to color, to color the output of the suffix and also to just find like features easily. So this is sort of [00:31:00] an example of like one of the tasks that was trained on, which is like, you describe the task in English and then you give it all these, like, You're asking for it to segment these two classes of objects, and then it finds, like, their locations using these tokens, and it finds their masks using some encoding of the masks into tokens.[00:31:24] Peter Robicheaux: And, yeah, so, one of my critiques, I guess, of PolyGemma 1, at least, is that You find that performance saturates as a pre trained model after only 300 million examples seen. So, what this graph is representing is each blue dot is a performance on some downstream task. And you can see that after seeing 300 million examples, It sort of does equally well on all of the downtrend tasks that they tried it on, which was a lot as 1 billion examples, which to me also kind of suggests a lack of capacity for this model.[00:31:58] Peter Robicheaux: PolyGemma2, [00:32:00] you can see the results on object detection. So these were transferred to to Coco. And you can see that this sort of also points to an increase in capacity being helpful to the model. You can see as. Both the resolution increases, and the parameter count of the language model increases, performance increases.[00:32:16] Peter Robicheaux: So resolution makes sense, obviously, it helps to find small images, or small objects in the image. But it also makes sense for another reason, which is that it kind of gives the model a thinking register, and it gives it more tokens to, like, process when making its predictions. But yeah, you could, you could say, oh, 43.[00:32:30] Peter Robicheaux: 6, that's not that great, like Florence 2 got 60. But this is not Training a dino or a debtor on top of this language or this image encoder. It's doing the raw language modeling task on Cocoa. So it doesn't have any of the bells and whistles. It doesn't have any of the fancy losses. It doesn't even have bipartite graph matching or anything like that.[00:32:52] Peter Robicheaux: Okay, the big result and one of the reasons that I was really excited about this paper is that they blow everything else away [00:33:00] on MMVP. I mean, 47. 3, sure, that's nowhere near human accuracy, which, again, is 94%, but for a, you know, a 2 billion language, 2 billion parameter language model to be chat2BT, that's quite the achievement.[00:33:12] Peter Robicheaux: And that sort of brings us to our final pick for paper of the year, which is AIMV2. So, AIMV2 sort of says, okay, Maybe this language model, like, maybe coming up with all these specific annotations to find features and with high fidelity and pixel space isn't actually necessary. And we can come up with an even simpler, more beautiful idea for combining you know, image tokens and pixel tokens in a way that's interfaceable for language tasks.[00:33:44] Peter Robicheaux: And this is nice because it can scale, you can come up with lots more data if you don't have to come up with all these annotations, right? So the way that it works. is it does something very, very similar to PolyGemo, where you have a vision encoder that dumps image tokens into a decoder only transformer.[00:33:59] Peter Robicheaux: But [00:34:00] the interesting thing is that it also autoregressively tries to learn the mean squared error of the image tokens. So instead of having to come up with fancy object detection or semantic, or segment, or segmentation labels, you can just try to reconstruct the image and have it learn fine grained features that way.[00:34:16] Peter Robicheaux: And it does this in kind of, I think, a beautiful way that's kind of compatible with the PolyGemma line of thinking, which is randomly sampling a prefix line of thinking Prefix length and using only this number of image tokens as the prefix. And so doing a similar thing with the causal. So the causal with prefix is the, the attention mask on the right.[00:34:35] Peter Robicheaux: So it's doing full block attention with some randomly sampled number of image tokens to then reconstruct the rest of the image and the downstream caption for that image. And so, This is the dataset that they train on. It's image or internet scale data, very high quality data created by the data filtering networks paper, essentially which is maybe The best clip data that exists.[00:34:59] Peter Robicheaux: [00:35:00] And we can see that this is finally a model that doesn't saturate. It's even at the highest parameter count, it's, it appears to be, oh, at the highest parameter account, it appears to be improving in performance with more and more samples seen. And so you can sort of think that. You know, if we just keep bumping the parameter count and increasing the example scene, which is the, the, the line of thinking for language models, then it'll keep getting better.[00:35:27] Peter Robicheaux: So how does it actually do at finding, oh, it also improves with resolution, which you would expect for a model that This is the ImageNet classification accuracy, but yeah, it does better if you increase the resolution, which means that it's actually leveraging and finding fine grained visual features.[00:35:44] Peter Robicheaux: And so how does that actually do compared to CLIP on Cocoa? Well, you can see that if you slap a transformer detection head on it, Entry now in Cocoa, it's just 60. 2, which is also within spitting distance of Soda, which means that it does a very good job of [00:36:00] finding visual features, but you could say, okay, well, wait a second.[00:36:03] Peter Robicheaux: Clip got to 59. 1, so. Like, how does this prove your claim at all? Because doesn't that mean like clip, which is known to be clip blind and do badly on MMVP, it's able to achieve a very high performance on fine, on this fine grained visual features task of object detection, well, they train on like, Tons of data.[00:36:24] Peter Robicheaux: They train on like objects, 365, Cocoa, Flickr and everything else. And so I think that this benchmark doesn't do a great job of selling how good of a pre trained model MV2 is. And we would like to see the performance on fewer data as examples and not trained to convergence on object detection. So seeing it in the real world on like a dataset, like RoboFlow 100, I think would be quite interesting.[00:36:48] Peter Robicheaux: And our, our, I guess our final, final pick for paper of 2024 would be Moondream. So introducing Vic to talk about that.[00:36:54] swyx: But overall, that was exactly what I was looking for. Like best of 2024, an amazing job. Yeah, you can, [00:37:00] if there's any other questions while Vic gets set up, like vision stuff,[00:37:07] swyx: yeah,[00:37:11] swyx: Vic, go ahead. Hi,[00:37:13] Vik Korrapati / Moondream[00:37:13] question: well, while we're getting set up, hi, over here, thanks for the really awesome talk. One of the things that's been weird and surprising is that the foundation model companies Even these MLMs, they're just like worse than RT Tether at detection still. Like, if you wanted to pay a bunch of money to auto label your detection dataset, If you gave it to OpenAI or Cloud, that would be like a big waste.[00:37:37] question: So I'm curious, just like, even Pali Gemma 2, like is worse. So, so I'm curious to hear your thoughts on like, how come, Nobody's cracked the code on like a generalist that really you know, beats a specialist model in computer vision like they have in in LLM land.[00:38:00][00:38:01] Isaac Robinson: Okay. It's a very, very interesting question. I think it depends on the specific domain. For image classification, it's basically there. In the, in AIMv2 showed, a simple attentional probe on the pre trained features gets like 90%, which is as well as anyone does. The, the, the, the bigger question, like, why isn't it transferring to object detection, especially like real time object detection.[00:38:25] Isaac Robinson: I think, in my mind, there are two answers. One is, object detection is really, really, really the architectures are super domain specific. You know, we see these, all these super, super complicated things, and it's not super easy to, to, to build something that just transfers naturally like that, whereas image classification, you know, clip pre training transfers super, super quickly.[00:38:48] Isaac Robinson: And the other thing is, until recently, the real time object detectors didn't even really benefit from pre training. Like, you see the YOLOs that are like, essentially saturated, showing very little [00:39:00] difference with pre training improvements, with using pre trained model at all. It's not surprising, necessarily, that People aren't looking at the effects of better and better pre training on real time detection.[00:39:12] Isaac Robinson: Maybe that'll change in the next year. Does that answer your question?[00:39:17] Peter Robicheaux: Can you guys hear me? Yeah, one thing I want to add is just like, or just to summarize, basically, is that like, Until 2024, you know, we haven't really seen a combination of transformer based object detectors and fancy losses, and PolyGemma suffers from the same problem, which is basically to say that these ResNet, or like the convolutional models, they have all these, like, extreme optimizations for doing object detection, but essentially, I think it's kind of been shown now that convolution models like just don't benefit from pre training and just don't like have the level of intelligence of transformer models.[00:39:56] swyx: Awesome. Hi,[00:39:59] Vik Korrapati: can [00:40:00] you hear me?[00:40:01] swyx: Cool. I hear you. See you. Are you sharing your screen?[00:40:04] Vik Korrapati: Hi. Might have forgotten to do that. Let me do[00:40:07] swyx: that. Sorry, should have done[00:40:08] Vik Korrapati: that.[00:40:17] swyx: Here's your screen. Oh, classic. You might have to quit zoom and restart. What? It's fine. We have a capture of your screen.[00:40:34] swyx: So let's get to it.[00:40:35] Vik Korrapati: Okay, easy enough.[00:40:49] Vik Korrapati: All right. Hi, everyone. My name is Vic. I've been working on Moondream for almost a year now. Like Shawn mentioned, I just went and looked and it turns out the first version I released December [00:41:00] 29, 2023. It's been a fascinating journey. So Moonbeam started off as a tiny vision language model. Since then, we've expanded scope a little bit to also try and build some tooling, client libraries, et cetera, to help people really deploy it.[00:41:13] Vik Korrapati: Unlike traditional large models that are focused at assistant type use cases, we're laser focused on building capabilities that developers can, sorry, it's yeah, we're basically focused on building capabilities that developers can use to build vision applications that can run anywhere. So, in a lot of cases for vision more so than for text, you really care about being able to run on the edge, run in real time, etc.[00:41:40] Vik Korrapati: So That's really important. We have we have different output modalities that we support. There's query where you can ask general English questions about an image and get back human like answers. There's captioning, which a lot of our users use for generating synthetic datasets to then train diffusion models and whatnot.[00:41:57] Vik Korrapati: We've done a lot of work to minimize those sessions there. [00:42:00] So that's. Use lot. We have open vocabulary object detection built in similar to a couple of more recent models like Palagem, et cetera, where rather than having to train a dedicated model, you can just say show me soccer balls in this image or show me if there are any deer in this image, it'll detect it.[00:42:14] Vik Korrapati: More recently, earlier this month, we released pointing capability where if all you're interested in is the center of an object you can just ask it to point out where that is. This is very useful when you're doing, you know, I automation type stuff. Let's see, LA we, we have two models out right now.[00:42:33] Vik Korrapati: There's a general purpose to be para model, which runs fair. Like it's, it's it's fine if you're running on server. It's good for our local Amma desktop friends and it can run on flagship, flagship mobile phones, but it never. so much for joining us today, and we'll see you in the [00:43:00] next one. Less memory even with our not yet fully optimized inference client.[00:43:06] Vik Korrapati: So the way we built our 0. 5b model was to start with the 2 billion parameter model and prune it while doing continual training to retain performance. We, our objective during the pruning was to preserve accuracy across a broad set of benchmarks. So the way we went about it was to estimate the importance of different components of the model, like attention heads, channels MLP rows and whatnot using basically a technique based on the gradient.[00:43:37] Vik Korrapati: I'm not sure how much people want to know details. We'll be writing a paper about this, but feel free to grab me if you have more questions. Then we iteratively prune a small chunk that will minimize loss and performance retrain the model to recover performance and bring it back. The 0. 5b we released is more of a proof of concept that this is possible.[00:43:54] Vik Korrapati: I think the thing that's really exciting about this is it makes it possible for for developers to build using the 2B param [00:44:00] model and just explore, build their application, and then once they're ready to deploy figure out what exactly they need out of the model and prune those capabilities into a smaller form factor that makes sense for their deployment target.[00:44:12] Vik Korrapati: So yeah, very excited about that. Let me talk to you folks a little bit about another problem I've been working on recently, which is similar to the clocks example we've been talking about. We had a customer reach out who was talking about, like, who had a bunch of gauges out in the field. This is very common in manufacturing and oil and gas, where you have a bunch of analog devices that you need to monitor.[00:44:34] Vik Korrapati: It's expensive to. And I was like, okay, let's have humans look at that and monitor stuff and make sure that the system gets shut down when the temperature goes over 80 or something. So I was like, yeah, this seems easy enough. Happy to, happy to help you distill that. Let's, let's get it going. Turns out our model couldn't do it at all.[00:44:51] Vik Korrapati: I went and looked at other open source models to see if I could just generate a bunch of data and learn from that. Did not work either. So I was like, let's look at what the folks with [00:45:00] hundreds of billions of dollars in market cap have to offer. And yeah, that doesn't work either. My hypothesis is that like the, the way these models are trained are using a large amount of image text data scraped from the internet.[00:45:15] Vik Korrapati: And that can be biased. In the case of gauges, most gauge images aren't gauges in the wild, they're product images. Detail images like these, where it's always set to zero. It's paired with an alt text that says something like GIVTO, pressure sensor, PSI, zero to 30 or something. And so the models are fairly good at picking up those details.[00:45:35] Vik Korrapati: It'll tell you that it's a pressure gauge. It'll tell you what the brand is, but it doesn't really learn to pay attention to the needle over there. And so, yeah, that's a gap we need to address. So naturally my mind goes to like, let's use synthetic data to, Solve this problem. That works, but it's problematic because it turned out we needed millions of synthetic gauge images to get to reasonable performance.[00:45:57] Vik Korrapati: And thinking about it, reading a gauge is like [00:46:00] not a one, like it's not a zero short process in our minds, right? Like if you had to tell me the reading in Celsius for this, Real world gauge. There's two dials on there. So first you have to figure out which one you have to be paying attention to, like the inner one or the outer one.[00:46:14] Vik Korrapati: You look at the tip of the needle, you look at what labels it's between, and you count how many and do some math to figure out what that probably is. So what happens if we just add that as a Chain of thought to give the model better understanding of the different sub, to allow the model to better learn the subtasks it needs to perform to accomplish this goal.[00:46:37] Vik Korrapati: So you can see in this example, this was actually generated by the latest version of our model. It's like, okay, Celsius is the inner scale. It's between 50 and 60. There's 10 ticks. So the second tick, it's a little debatable here, like there's a weird shadow situation going on, the dial is off, so I don't know what the ground truth is, but it works okay.[00:46:57] Vik Korrapati: There's points on there that are, the points [00:47:00] over there are actually grounded. I don't know if this is easy to see, but when I click on those, there's a little red dot that moves around on the image. The model actually has to predict where this points are, I was already trying to do this with bounding boxes, but then Malmo came out with pointing capabilities.[00:47:15] Vik Korrapati: And it's like pointing is a much better paradigm to to represent this. We see pretty good results. This one's actually for clock reading. I couldn't find our chart for gauge reading at the last minute. So the light. Blue chart is with our rounded chain of thought. This measures, we have, we built a clock reading benchmark about 500 images.[00:47:37] Vik Korrapati: This measures accuracy on that. You can see it's a lot more sample efficient when you're using the chain of thought to model. Another big benefit from this approach is like, you can kind of understand how the model is. it and how it's failing. So in this example, the actual correct reading is 54 Celsius, the model output [00:48:00] 56, not too bad but you can actually go and see where it messed up. Like it got a lot of these right, except instead of saying it was on the 7th tick, it actually predicted that it was the 8th tick and that's why it went with 56.[00:48:14] Vik Korrapati: So now that you know that this. Failing in this way, you can adjust how you're doing the chain of thought to maybe say like, actually count out each tick from 40, instead of just trying to say it's the eighth tick. Or you might say like, okay, I see that there's that middle thing, I'll count from there instead of all the way from 40.[00:48:31] Vik Korrapati: So helps a ton. The other thing I'm excited about is a few short prompting or test time training with this. Like if a customer has a specific gauge that like we're seeing minor errors on, they can give us a couple of examples where like, if it's miss detecting the. Needle, they can go in and correct that in the chain of thought.[00:48:49] Vik Korrapati: And hopefully that works the next time. Now, exciting approach, we only apply it to clocks and gauges. The real question is, is it going to generalize? Probably, like, there's some science [00:49:00] from text models that when you train on a broad number of tasks, it does generalize. And I'm seeing some science with our model as well.[00:49:05] Vik Korrapati: So, in addition to the image based chain of thought stuff, I also added some spelling based chain of thought to help it understand better understand OCR, I guess. I don't understand why everyone doesn't do this, by the way. Like, it's trivial benchmark question. It's Very, very easy to nail. But I also wanted to support it for stuff like license plate, partial matching, like, hey, does any license plate in this image start with WHA or whatever?[00:49:29] Vik Korrapati: So yeah, that sort of worked. All right, that, that ends my story about the gauges. If you think about what's going on over here it's interesting that like LLMs are showing enormous. Progress in reasoning, especially with the latest set of models that we've seen, but we're not really seeing, I have a feeling that VLMs are lagging behind, as we can see with these tasks that should be very simple for a human to do [00:50:00] that are very easy to find VLMs failing at.[00:50:04] Vik Korrapati: My hypothesis on why this is the case is because On the internet, there's a ton of data that talks about how to reason. There's books about how to solve problems. There's books critiquing the books about how to solve problems. But humans are just so good at perception that we never really talk about it.[00:50:20] Vik Korrapati: Like, maybe in art books where it's like, hey, to show that that mountain is further away, you need to desaturate it a bit or whatever. But the actual data on how to, like, look at images is, isn't really present. Also, the Data we have is kind of sketched. The best source of data we have is like image all text pairs on the internet and that's pretty low quality.[00:50:40] Vik Korrapati: So yeah, I, I think our solution here is really just we need to teach them how to operate on individual tasks and figure out how to scale that out. All right. Yep. So conclusion. At Moondream we're trying to build amazing PLMs that run everywhere. Very hard problem. Much work ahead, but we're making a ton of progress and I'm really excited [00:51:00] about If anyone wants to chat about more technical details about how we're doing this or interest in collaborating, please, please hit me up.[00:51:08] Isaac Robinson: Yeah,[00:51:09] swyx: like, I always, when people say, when people say multi modality, like, you know, I always think about vision as the first among equals in all the modalities. So, I really appreciate having the experts in the room. Get full access to Latent Space at www.latent.space/subscribe

Dr. Olga Russakovsky, Computer Science at Princeton University, joins Lightspeed Partner Michael Mignano to discuss what the next generation of AI talent is learning and where she expects to find the next big innovation in artificial intelligence.  From her research in computer vision and human-computer interaction to her work in fairness, accountability, and transparency in AI, Dr. Russakovsky has earned many awards, including the MIT Technology Review's 35-under-35 Innovator award and the Foreign Policy Magazine's 100 Leading Global Thinkers award. Dr. Russakovsky is also the co-founder and Board Chair of AI4ALL, a nonprofit that aims to increase the diversity of thought in Artificial Intelligence.  Episode Chapters 00:00 Introduction and Guest Overview 01:17 Olga's Career Journey 02:30 Understanding Computer Vision 04:43 Generative AI and Computer Vision 06:36 Interdisciplinary AI Research 15:00 AI4All: Diversity of Thought  17:44 Challenges and Bias in AI 30:01 Future of AI and Data 40:08 ImageNET 43:38 Closing Thoughts Stay in touch: www.lsvp.com X: https://twitter.com/lightspeedvp LinkedIn: https://www.linkedin.com/company/lightspeed-venture-partners/ Instagram: https://www.instagram.com/lightspeedventurepartners/ Subscribe on your favorite podcast app: generativenow.co Email: generativenow@lsvp.com The content here does not constitute tax, legal, business or investment advice or an offer to provide such advice, should not be construed as advocating the purchase or sale of any security or investment or a recommendation of any company, and is not an offer, or solicitation of an offer, for the purchase or sale of any security or investment product. For more details please see lsvp.com/legal.

- GS. Fei-Fei Li (Đại học Stanford, Mỹ) là một trong 5 nhà khoa học được vinh danh Giải thưởng Chính VinFuture 2024 trị giá 3 triệu USD vì những đóng góp đột phá để thúc đẩy sự tiến bộ của học sâu. Bà là người đã tạo ra tập dữ liệu ImageNet giúp thúc đẩy sự tiến bộ trong hệ thống nhận diện hình ảnh, giúp huấn luyện các mô hình học sâu ở quy mô lớn. Trong Chuyện đêm hôm nay, chúng tôi mời quý vị và các bạn cùng trò chuyện với nhà khoa học người Mỹ được mệnh danh là “mẹ đỡ đầu” của AI, nổi tiếng với đóng góp đột phá trong lĩnh vực thị giác máy tính. Chủ đề : GS Fei Fei Li, ĐH Stanford, Mỹ, nghiên cứu AI, Việt Nam phát triển AI --- Support this podcast: https://podcasters.spotify.com/pod/show/vov1sukien/support

This episode of Eye on AI is sponsored by Citrusx. Unlock reliable AI with Citrusx! Our platform simplifies validation and risk management, empowering you to make smarter decisions and stay compliant. Detects and mitigate AI vulnerabilities, biases, and errors with ease.    Visit http://email.citrusx.ai/eyeonai to download our free fairness use case and see the solution in action.     In this episode of the Eye on AI podcast, Terry Sejnowski, a pioneer in neural networks and computational neuroscience, joins Craig Smith to discuss the future of AI, the evolution of ChatGPT, and the challenges of understanding intelligence.   Terry, a key figure in the deep learning revolution, shares insights into how neural networks laid the foundation for modern AI, including ChatGPT's groundbreaking generative capabilities. From its ability to mimic human-like creativity to its limitations in true understanding, we explore what makes ChatGPT remarkable and what it still lacks compared to human cognition.   We also dive into fascinating topics like the debate over AI sentience, the concept of "hallucinations" in AI models, and how language models like ChatGPT act as mirrors reflecting user input rather than possessing intrinsic intelligence. Terry explains how understanding language and meaning in AI remains one of the field's greatest challenges.   Additionally, Terry shares his perspective on nature-inspired AI and what it will take to develop systems that go beyond prediction to exhibit true autonomy and decision-making.   Learn why AI models like ChatGPT are revolutionary yet incomplete, how generative AI might redefine creativity, and what the future holds for AI as we continue to push its boundaries.   Don't miss this deep dive into the fascinating world of AI with Terry Sejnowski. Like, subscribe, and hit the notification bell for more cutting-edge AI insights!   Stay Updated: Craig Smith Twitter: https://twitter.com/craigss Eye on A.I. Twitter: https://twitter.com/EyeOn_AI     (00:00) Introduction to Terry Sejnowski and His Work  (03:02) The Origins of Modern AI and Neural Networks  (05:29) The Deep Learning Revolution and ImageNet  (07:11) Understanding ChatGPT and Generative AI  (12:34) Exploring AI Creativity (16:03) Lessons from Gaming AI: AlphaGo and Backgammon  (18:37) Early Insights into AI's Affinity for Language  (24:48) Syntax vs. Semantics: The Purpose of Language  (30:00) How Written Language Transformed AI Training  (35:10) Can AI Become Sentient?  (41:37) AI Agents and the Next Frontier in Automation  (45:43) Nature-Inspired AI: Lessons from Biology  (50:02) Digital vs. Biological Computation: Key Differences  (54:29) Will AI Replace Jobs? (57:07) The Future of AI

Hi everyone!If you're a new subscriber or listener, welcome. If you're not new, you've probably noticed that things have slowed down from us a bit recently. Hugh Zhang, Andrey Kurenkov and I sat down to recap some of The Gradient's history, where we are now, and how things will look going forward. To summarize and give some context:The Gradient has been around for around 6 years now – we began as an online magazine, and began producing our own newsletter and podcast about 4 years ago. With a team of volunteers — we take in a bit of money through Substack that we use for subscriptions to tools we need and try to pay ourselves a bit — we've been able to keep this going for quite some time. Our team has less bandwidth than we'd like right now (and I'll admit that at least some of us are running on fumes…) — we'll be making a few changes:* Magazine: We're going to be scaling down our editing work on the magazine. While we won't be accepting pitches for unwritten drafts for now, if you have a full piece that you'd like to pitch to us, we'll consider posting it. If you've reached out about writing and haven't heard from us, we're really sorry. We've tried a few different arrangements to manage the pipeline of articles we have, but it's been difficult to make it work. We still want this to be a place to promote good work and writing from the ML community, so we intend to continue using this Substack for that purpose. If we have more editing bandwidth on our team in the future, we want to continue doing that work. * Newsletter: We'll aim to continue the newsletter as before, but with a “Best from the Community” section highlighting posts. We'll have a way for you to send articles you want to be featured, but for now you can reach us at our editor@thegradient.pub. * Podcast: I'll be continuing this (at a slower pace), but eventually transition it away from The Gradient given the expanded range. If you're interested in following, it might be worth subscribing on another player like Apple Podcasts, Spotify, or using the RSS feed.* Sigmoid Social: We'll keep this alive as long as there's financial support for it.If you like what we do and/or want to help us out in any way, do reach out to editor@thegradient.pub. We love hearing from you.Timestamps* (0:00) Intro* (01:55) How The Gradient began* (03:23) Changes and announcements* (10:10) More Gradient history! On our involvement, favorite articles, and some plugsSome of our favorite articles!There are so many, so this is very much a non-exhaustive list:* NLP's ImageNet moment has arrived* The State of Machine Learning Frameworks in 2019* Why transformative artificial intelligence is really, really hard to achieve* An Introduction to AI Story Generation* The Artificiality of Alignment (I didn't mention this one in the episode, but it should be here)Places you can find us!Hugh:* Twitter* Personal site* Papers/things mentioned!* A Careful Examination of LLM Performance on Grade School Arithmetic (GSM1k)* Planning in Natural Language Improves LLM Search for Code Generation* Humanity's Last ExamAndrey:* Twitter* Personal site* Last Week in AI PodcastDaniel:* Twitter* Substack blog* Personal site (under construction) Get full access to The Gradient at thegradientpub.substack.com/subscribe

Hey all, Alex here, coming to you from the (surprisingly) sunny Seattle, with just a mind-boggling week of releases. Really, just on Tuesday there was so much news already! I had to post a recap thread, something I do usually after I finish ThursdAI! From Anthropic reclaiming close-second sometimes-first AI lab position + giving Claude the wheel in the form of computer use powers, to more than 3 AI video generation updates with open source ones, to Apple updating Apple Intelligence beta, it's honestly been very hard to keep up, and again, this is literally part of my job! But once again I'm glad that we were able to cover this in ~2hrs, including multiple interviews with returning co-hosts ( Simon Willison came back, Killian came back) so definitely if you're only a reader at this point, listen to the show! Ok as always (recently) the TL;DR and show notes at the bottom (I'm trying to get you to scroll through ha, is it working?) so grab a bucket of popcorn, let's dive in

Fei-Fei Li and Justin Johnson are pioneers in AI. While the world has only recently witnessed a surge in consumer AI, our guests have long been laying the groundwork for innovations that are transforming industries today.In this episode, a16z General Partner Martin Casado joins Fei-Fei and Justin to explore the journey from early AI winters to the rise of deep learning and the rapid expansion of multimodal AI. From foundational advancements like ImageNet to the cutting-edge realm of spatial intelligence, Fei-Fei and Justin share the breakthroughs that have shaped the AI landscape and reveal what's next for innovation at World Labs.If you're curious about how AI is evolving beyond language models and into a new realm of 3D, generative worlds, this episode is a must-listen.Resources: Learn more about World Labs: https://www.worldlabs.aiFind Fei-Fei on Twitter: https://x.com/drfeifeiFind Justin on Twitter: https://x.com/jcjohnss Stay Updated: Let us know what you think: https://ratethispodcast.com/a16zFind a16z on Twitter: https://twitter.com/a16zFind a16z on LinkedIn: https://www.linkedin.com/company/a16zSubscribe on your favorite podcast app: https://a16z.simplecast.com/Follow our host: https://twitter.com/stephsmithioPlease note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.

AI researcher Jim Fan has had a charmed career. He was OpenAI's first intern before he did his PhD at Stanford with “godmother of AI,” Fei-Fei Li. He graduated into a research scientist position at Nvidia and now leads its Embodied AI “GEAR” group. The lab's current work spans foundation models for humanoid robots to agents for virtual worlds. Jim describes a three-pronged data strategy for robotics, combining internet-scale data, simulation data and real world robot data. He believes that in the next few years it will be possible to create a “foundation agent” that can generalize across skills, embodiments and realities—both physical and virtual. He also supports Jensen Huang's idea that “Everything that moves will eventually be autonomous.” Hosted by: Stephanie Zhan and Sonya Huang, Sequoia Capital Mentioned in this episode: World of Bits: Early OpenAI project Jim worked on as an intern with Andrej Karpathy. Part of a bigger initiative called Universe Fei-Fei Li: Jim's PhD advisor at Stanford who founded the ImageNet project in 2010 that revolutionized the field of visual recognition, led the Stanford Vision Lab and just launched her own AI startup, World Labs Project GR00T: Nvidia's “moonshot effort” at a robotic foundation model, premiered at this year's GTC Thinking Fast and Slow: Influential book by Daniel Kahneman that popularized some of his teaching from behavioral economics Jetson Orin chip: The dedicated series of edge computing chips Nvidia is developing to power Project GR00T Eureka: Project by Jim's team that trained a five finger robot hand to do pen spinning MineDojo: A project Jim did when he first got to Nvidia that developed a platform for general purpose agents in the game of Minecraft. Won NeurIPS 2022 Outstanding Paper Award ADI: artificial dog intelligence Mamba: Selective State Space Models, an alternative architecture to Transformers that Jim is interested in (original paper here) 00:00 Introduction 01:35 Jim's journey to embodied intelligence 04:53 The GEAR Group 07:32 Three kinds of data for robotics 10:32 A GPT-3 moment for robotics 16:05 Choosing the humanoid robot form factor 19:37 Specialized generalists 21:59 GR00T gets its own chip 23:35 Eureka and Issac Sim 25:23 Why now for robotics? 28:53 Exploring virtual worlds 36:28 Implications for games 39:13 Is the virtual world in service of the physical world? 42:10 Alternative architectures to Transformers 44:15 Lightning round

Andrew Ilyas, a PhD student at MIT who is about to start as a professor at CMU. We discuss Data modeling and understanding how datasets influence model predictions, Adversarial examples in machine learning and why they occur, Robustness in machine learning models, Black box attacks on machine learning systems, Biases in data collection and dataset creation, particularly in ImageNet and Self-selection bias in data and methods to address it. MLST is sponsored by Brave: The Brave Search API covers over 20 billion webpages, built from scratch without Big Tech biases or the recent extortionate price hikes on search API access. Perfect for AI model training and retrieval augmentated generation. Try it now - get 2,000 free queries monthly at http://brave.com/api Andrew's site: https://andrewilyas.com/ https://x.com/andrew_ilyas TOC: 00:00:00 - Introduction and Andrew's background 00:03:52 - Overview of the machine learning pipeline 00:06:31 - Data modeling paper discussion 00:26:28 - TRAK: Evolution of data modeling work 00:43:58 - Discussion on abstraction, reasoning, and neural networks 00:53:16 - "Adversarial Examples Are Not Bugs, They Are Features" paper 01:03:24 - Types of features learned by neural networks 01:10:51 - Black box attacks paper 01:15:39 - Work on data collection and bias 01:25:48 - Future research plans and closing thoughts References: Adversarial Examples Are Not Bugs, They Are Features https://arxiv.org/pdf/1905.02175 TRAK: Attributing Model Behavior at Scale https://arxiv.org/pdf/2303.14186 Datamodels: Predicting Predictions from Training Data https://arxiv.org/pdf/2202.00622 Adversarial Examples Are Not Bugs, They Are Features https://arxiv.org/pdf/1905.02175 IMAGENET-TRAINED CNNS https://arxiv.org/pdf/1811.12231 ZOO: Zeroth Order Optimization Based Black-box https://arxiv.org/pdf/1708.03999 A Spline Theory of Deep Networks https://proceedings.mlr.press/v80/balestriero18b/balestriero18b.pdf Scaling Monosemanticity https://transformer-circuits.pub/2024/scaling-monosemanticity/ Adversarial Examples Are Not Bugs, They Are Features https://gradientscience.org/adv/ Adversarial Robustness Limits via Scaling-Law and Human-Alignment Studies https://proceedings.mlr.press/v235/bartoldson24a.html Prior Convictions: Black-Box Adversarial Attacks with Bandits and Priors https://arxiv.org/abs/1807.07978 Estimation of Standard Auction Models https://arxiv.org/abs/2205.02060 From ImageNet to Image Classification: Contextualizing Progress on Benchmarks https://arxiv.org/abs/2005.11295 Estimation of Standard Auction Models https://arxiv.org/abs/2205.02060 What Makes A Good Fisherman? Linear Regression under Self-Selection Bias https://arxiv.org/abs/2205.03246 Towards Tracing Factual Knowledge in Language Models Back to the Training Data [Akyürek] https://arxiv.org/pdf/2205.11482

While Large Language Models (LLMs) are the dominant models for generative tasks in language, they do not perform as well as diffusion models on image and video generation. To effectively use LLMs for visual generation, one crucial component is the visual tokenizer that maps pixel-space inputs to discrete tokens appropriate for LLM learning. In this paper, we introduce MAGVIT-v2, a video tokenizer designed to generate concise and expressive tokens for both videos and images using a common token vocabulary. Equipped with this new tokenizer, we show that LLMs outperform diffusion models on standard image and video generation benchmarks including ImageNet and Kinetics. In addition, we demonstrate that our tokenizer surpasses the previously top-performing video tokenizer on two more tasks: (1) video compression comparable to the next-generation video codec (VCC) according to human evaluations, and (2) learning effective representations for action recognition tasks. 2023: Lijun Yu, José Lezama, N. B. Gundavarapu, Luca Versari, Kihyuk Sohn, David C. Minnen, Yong Cheng, Agrim Gupta, Xiuye Gu, Alexander G. Hauptmann, Boqing Gong, Ming-Hsuan Yang, Irfan Essa, David A. Ross, Lu Jiang https://arxiv.org/pdf/2310.05737

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: The 'strong' feature hypothesis could be wrong, published by lewis smith on August 2, 2024 on The AI Alignment Forum. NB. I am on the Google Deepmind language model interpretability team. But the arguments/views in this post are my own, and shouldn't be read as a team position. "It would be very convenient if the individual neurons of artificial neural networks corresponded to cleanly interpretable features of the input. For example, in an "ideal" ImageNet classifier, each neuron would fire only in the presence of a specific visual feature, such as the color red, a left-facing curve, or a dog snout" Elhage et. al, Toy Models of Superposition Recently, much attention in the field of mechanistic interpretability, which tries to explain the behavior of neural networks in terms of interactions between lower level components, has been focussed on extracting features from the representation space of a model. The predominant methodology for this has used variations on the sparse autoencoder, in a series of papers inspired by Elhage et. als. model of superposition.It's been conventionally understood that there are two key theories underlying this agenda. The first is the 'linear representation hypothesis' (LRH), the hypothesis that neural networks represent many intermediates or variables of the computation (such as the 'features of the input' in the opening quote) as linear directions in it's representation space, or atoms[1]. And second, the theory that the network is capable of representing more of these 'atoms' than it has dimensions in its representation space, via superposition (the superposition hypothesis). While superposition is a relatively uncomplicated hypothesis, I think the LRH is worth examining in more detail. It is frequently stated quite vaguely, and I think there are several possible formulations of this hypothesis, with varying degrees of plausibility, that it is worth carefully distinguishing between. For example, the linear representation hypothesis is often stated as 'networks represent features of the input as directions in representation space'. Here are two importantly different ways to parse this: 1. (Weak LRH) some or many features used by neural networks are represented as atoms in representation space 2. (Strong LRH) all (or the vast majority of) features used by neural networks are represented by atoms. The weak LRH I would say is now well supported by considerable empirical evidence. The strong form is much more speculative: confirming the existence of many linear representations does not necessarily provide strong evidence for the strong hypothesis. Both the weak and the strong forms of the hypothesis can still have considerable variation, depending on what we understand by a feature and the proportion of the model we expect to yield to analysis, but I think that the distinction between just a weak and strong form is clear enough to work with. I think that in addition to the acknowledged assumption of the LRH and superposition hypotheses, much work on SAEs in practice makes the assumption that each atom in the network will represent a "simple feature" or a "feature of the input". These features that the atoms are representations of are assumed to be 'monosemantic': they will all stand for features which are human interpretable in isolation. I will call this the monosemanticity assumption. This is difficult to state precisely, but we might formulate it as the theory that every represented variable will have a single meaning in a good description of a model. This is not a straightforward assumption due to how imprecise the notion of a single meaning is. While various more or less reasonable definitions for features are discussed in the pioneering work of Elhage, these assumptions have different implications. For instance, if one thinks of 'feat...

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: The 'strong' feature hypothesis could be wrong, published by lsgos on August 2, 2024 on LessWrong. NB. I am on the Google Deepmind language model interpretability team. But the arguments/views in this post are my own, and shouldn't be read as a team position. "It would be very convenient if the individual neurons of artificial neural networks corresponded to cleanly interpretable features of the input. For example, in an "ideal" ImageNet classifier, each neuron would fire only in the presence of a specific visual feature, such as the color red, a left-facing curve, or a dog snout" : Elhage et. al, Toy Models of Superposition Recently, much attention in the field of mechanistic interpretability, which tries to explain the behavior of neural networks in terms of interactions between lower level components, has been focussed on extracting features from the representation space of a model. The predominant methodology for this has used variations on the sparse autoencoder, in a series of papers inspired by Elhage et. als. model of superposition. Conventionally there understood to be two key theories underlying this agenda. The first is the 'linear representation hypothesis' (LRH), the hypothesis that neural networks represent many intermediates or variables of the computation (such as the 'features of the input' in the opening quote) as linear directions in it's representation space, or atoms[1]. And second, the theory that the network is capable of representing more of these 'atoms' than it has dimensions in its representation space, via superposition (the superposition hypothesis). While superposition is a relatively uncomplicated hypothesis, I think the LRH is worth examining in more detail. It is frequently stated quite vaguely, and I think there are several possible formulations of this hypothesis, with varying degrees of plausibility, that it is worth carefully distinguishing between. For example, the linear representation hypothesis is often stated as 'networks represent features of the input as directions in representation space'. There are a few possible formulations of this: 1. (Weak LRH) some features used by neural networks are represented as atoms in representation space 2. (Strong LRH) all features used by neural networks are represented by atoms. The weak LRH I would say is now well supported by considerable empirical evidence. The strong form is much more speculative: confirming the existence of many linear representations does not necessarily provide strong evidence for the strong hypothesis. Both the weak and the strong forms of the hypothesis can still have considerable variation, depending on what we understand by a feature. I think that in addition to the acknowledged assumption of the LRH and superposition hypotheses, much work on SAEs in practice makes the assumption that each atom in the network will represent a "simple feature" or a "feature of the input". These features that the atoms are representations of are assumed to be 'monosemantic': they will all stand for features which are human interpretable in isolation. I will call this the monosemanticity assumption. This is difficult to state precisely, but we might formulate as the theory that every represented variable will have a single meaning in a good description of a model. This is not a straightforward assumption due to how imprecise the notion of a single meaning is. While various more or less reasonable definitions for features are discussed in the pioneering work of Elhage, these assumptions have different implications. For instance, if one thinks of 'features' as computational intermediates in a broad sense, then superposition and the LRH imply a certain picture of the format of a models internal representation: that what the network is doing is manipulating atoms in superposition (if y...

Key Topics & Chapter Markers:AI's Evolutionary Journey & Key Challenges [00:00:00]Neural Networks: Inspiration from Biology [00:01:00]Weighted Sum, Inputs & Mathematical Functions [00:05:00]Gradient Descent & Optimization in Neural Nets [00:10:15]Computing Architecture: CPUs vs. GPUs [00:39:56]RNNs and Early Problems in Memory & Context [01:03:00]The Emergence of Convolutional Neural Networks (CNNs) [01:10:00]ImageNet, GPUs & Scaling Neural Networks [01:24:00]Share Your Thoughts: Have questions or comments? Drop us a mail at EffortlessPodcastHQ@gmail.com

We meet Dr. Fei-Fei Li In the latest installment of our oral history project. She's a Chinese-American computer scientist and the creator of ImageNet - the dataset that made rapid advances possible in this field of AI that helps computers take meaningful information from things like photos and videos.We Meet: Stanford University's Fei-Fei Li, author of "The Worlds I See: Curiosity, Exploration, and Discovery at the Dawn of AI"Credits:This episode of SHIFT was produced by Jennifer Strong with help from Emma Cillekens. It was mixed by Garret Lang, with original music from him and Jacob Gorski. Art by Anthony Green.

The conventional recipe for maximizing model accuracy is to (1) train multiple models with various hyperparameters and (2) pick the individual model which performs best on a held-out validation set, discarding the remainder. In this paper, we revisit the second step of this procedure in the context of fine-tuning large pre-trained models, where fine-tuned models often appear to lie in a single low error basin. We show that averaging the weights of multiple models fine-tuned with different hyperparameter configurations often improves accuracy and robustness. Unlike a conventional ensemble, we may average many models without incurring any additional inference or memory costs -- we call the results"model soups."When fine-tuning large pre-trained models such as CLIP, ALIGN, and a ViT-G pre-trained on JFT, our soup recipe provides significant improvements over the best model in a hyperparameter sweep on ImageNet. The resulting ViT-G model, which attains 90.94% top-1 accuracy on ImageNet, achieved a new state of the art. Furthermore, we show that the model soup approach extends to multiple image classification and natural language processing tasks, improves out-of-distribution performance, and improves zero-shot performance on new downstream tasks. Finally, we analytically relate the performance similarity of weight-averaging and logit-ensembling to flatness of the loss and confidence of the predictions, and validate this relation empirically. Code is available at https://github.com/mlfoundations/model-soups. 2022: Mitchell Wortsman, Gabriel Ilharco, S. Gadre, R. Roelofs, Raphael Gontijo-Lopes, Ari S. Morcos, Hongseok Namkoong, Ali Farhadi, Y. Carmon, Simon Kornblith, Ludwig Schmidt https://arxiv.org/pdf/2203.05482

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Rational Animations' intro to mechanistic interpretability, published by Writer on June 15, 2024 on LessWrong. In our new video, we talk about research on interpreting InceptionV1, a convolutional neural network. Researchers have been able to understand the function of neurons and channels inside the network and uncover visual processing algorithms by looking at the weights. The work on InceptionV1 is early but landmark mechanistic interpretability research, and it functions well as an introduction to the field. We also go into the rationale and goals of the field and mention some more recent research near the end. Our main source material is the circuits thread in the Distill journal and this article on feature visualization. The author of the script is Arthur Frost. I have included the script below, although I recommend watching the video since the script has been written with accompanying moving visuals in mind. Intro In 2018, researchers trained an AI to find out if people were at risk of heart conditions based on pictures of their eyes, and somehow the AI also learned to tell people's biological sex with incredibly high accuracy. How? We're not entirely sure. The crazy thing about Deep Learning is that you can give an AI a set of inputs and outputs, and it will slowly work out for itself what the relationship between them is. We didn't teach AIs how to play chess, go, and atari games by showing them human experts - we taught them how to work it out for themselves. And the issue is, now they have worked it out for themselves, and we don't know what it is they worked out. Current state-of-the-art AIs are huge. Meta's largest LLaMA2 model uses 70 billion parameters spread across 80 layers, all doing different things. It's deep learning models like these which are being used for everything from hiring decisions to healthcare and criminal justice to what youtube videos get recommended. Many experts believe that these models might even one day pose existential risks. So as these automated processes become more widespread and significant, it will really matter that we understand how these models make choices. The good news is, we've got a bit of experience uncovering the mysteries of the universe. We know that humans are made up of trillions of cells, and by investigating those individual cells we've made huge advances in medicine and genetics. And learning the properties of the atoms which make up objects has allowed us to develop modern material science and high-precision technology like computers. If you want to understand a complex system with billions of moving parts, sometimes you have to zoom in. That's exactly what Chris Olah and his team did starting in 2015. They focused on small groups of neurons inside image models, and they were able to find distinct parts responsible for detecting everything from curves and circles to dog heads and cars. In this video we'll Briefly explain how (convolutional) neural networks work Visualise what individual neurons are doing Look at how neurons - the most basic building blocks of the neural network - combine into 'circuits' to perform tasks Explore why interpreting networks is so hard There will also be lots of pictures of dogs, like this one. Let's get going. We'll start with a brief explanation of how convolutional neural networks are built. Here's a network that's trained to label images. An input image comes in on the left, and it flows along through the layers until we get an output on the right - the model's attempt to classify the image into one of the categories. This particular model is called InceptionV1, and the images it's learned to classify are from a massive collection called ImageNet. ImageNet has 1000 different categories of image, like "sandal" and "saxophone" and "sarong" (which, if you don't know, is a k...

Speakers for AI Engineer World's Fair have been announced! See our Microsoft episode for more info and buy now with code LATENTSPACE — we've been studying the best ML research conferences so we can make the best AI industry conf! Note that this year there are 4 main tracks per day and dozens of workshops/expo sessions; the free livestream will air much less than half of the content this time.Apply for free/discounted Diversity Program and Scholarship tickets here. We hope to make this the definitive technical conference for ALL AI engineers.ICLR 2024 took place from May 6-11 in Vienna, Austria. Just like we did for our extremely popular NeurIPS 2023 coverage, we decided to pay the $900 ticket (thanks to all of you paying supporters!) and brave the 18 hour flight and 5 day grind to go on behalf of all of you. We now present the results of that work!This ICLR was the biggest one by far, with a marked change in the excitement trajectory for the conference:Of the 2260 accepted papers (31% acceptance rate), of the subset of those relevant to our shortlist of AI Engineering Topics, we found many, many LLM reasoning and agent related papers, which we will cover in the next episode. We will spend this episode with 14 papers covering other relevant ICLR topics, as below.As we did last year, we'll start with the Best Paper Awards. Unlike last year, we now group our paper selections by subjective topic area, and mix in both Outstanding Paper talks as well as editorially selected poster sessions. Where we were able to do a poster session interview, please scroll to the relevant show notes for images of their poster for discussion. To cap things off, Chris Ré's spot from last year now goes to Sasha Rush for the obligatory last word on the development and applications of State Space Models.We had a blast at ICLR 2024 and you can bet that we'll be back in 2025

When a massive direct-view LED video wall is installed to make a completely immersive experience, teams collaborate to make sure it's a successful project. To hear all the details of this project for Flogistix, Justin and Matt are joined by Ali Sylvester, Director of Business Solutions at Flogistix, and Kyle Kempf, CTS-I Director of Commercial Audio Video at ImageNet Consulting. They dig into the vision for the space, the architecture that went into it and everything else that brings the project to life.   Links: Daktronics News Release: https://www.daktronics.com/news/imagenet-and-daktronics-deliver-led-video-wall-experience-for-flogistix Flogistix Website: https://flogistix.com/ ImageNet Consulting Website: https://www.imagenetconsulting.com/ Rand Elliott Architects Website: https://randelliottarchitects.com/ Daktronics and ImageNet Podcast: https://podcast.daktronics.com/e/143-imagenet-consulting-with-kyle-kempf/

State-of-the-art computer vision systems are trained to predict a fixed set of predetermined object categories. This restricted form of supervision limits their generality and usability since additional labeled data is needed to specify any other visual concept. Learning directly from raw text about images is a promising alternative which leverages a much broader source of supervision. We demonstrate that the simple pre-training task of predicting which caption goes with which image is an efficient and scalable way to learn SOTA image representations from scratch on a dataset of 400 million (image, text) pairs collected from the internet. After pre-training, natural language is used to reference learned visual concepts (or describe new ones) enabling zero-shot transfer of the model to downstream tasks. We study the performance of this approach by benchmarking on over 30 different existing computer vision datasets, spanning tasks such as OCR, action recognition in videos, geo-localization, and many types of fine-grained object classification. The model transfers non-trivially to most tasks and is often competitive with a fully supervised baseline without the need for any dataset specific training. For instance, we match the accuracy of the original ResNet-50 on ImageNet zero-shot without needing to use any of the 1.28 million training examples it was trained on. 2024: Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever

“We haven't invested this much money into an infrastructure like this really until you go back to the pyramids”—Kate CrawfordTranscript with links to audio and external links. Ground Truths podcasts are on Apple and Spotify. The video interviews are on YouTube Eric Topol (00:06):Well, hello, this is Eric Topol with Ground Truths, and I'm really delighted today to welcome Kate Crawford, who we're very lucky to have as an Australian here in the United States. And she's multidimensional, as I've learned, not just a scholar of AI, all the dimensions of AI, but also an artist, a musician. We're going to get into all this today, so welcome Kate.Kate Crawford (00:31):Thank you so much, Eric. It's a pleasure to be here.Eric Topol (00:34):Well, I knew of your work coming out of the University of Southern California (USC) as a professor there and at Microsoft Research, and I'm only now learning about all these other things that you've been up to including being recognized in TIME 2023 as one of 100 most influential people in AI and it's really fascinating to see all the things that you've been doing. But I guess I'd start off with one of your recent publications in Nature. It was a world view, and it was about generative AI is guzzling water and energy. And in that you wrote about how these large AI systems, which are getting larger seemingly every day are needing as much energy as entire nations and the water consumption is rampant. So maybe we can just start off with that. You wrote a really compelling piece expressing concerns, and obviously this is not just the beginning of all the different aspects you've been tackling with AI.Exponential Growth, Exponential Concerns Kate Crawford (01:39):Well, we're in a really interesting moment. What I've done as a researcher in this space for a very long time now is really introduce a material analysis of artificial intelligence. So we are often told that AI is a very immaterial technology. It's algorithms in the cloud, it's objective mathematics, but in actual fact, it comes with an enormous material infrastructure. And this is something that I took five years to research for my last book, Atlas of AI. It meant going to the mines where lithium and cobalt are being extracted. It meant going into the Amazon fulfillment warehouses to see how humans collaborate with robotic and AI systems. And it also meant looking at the large-scale labs where training data is being gathered and then labeled by crowd workers. And for me, this really changed my thinking. It meant that going from being a professor for 15 years focusing on AI from a very traditional perspective where we write papers, we're sitting in our offices behind desks, that I really had to go and do these journeys, these field trips, to understand that full extractive infrastructure that is needed to run AI at a planetary scale.(02:58):So I've been keeping a very close eye on what would change with generative AI and what we've seen particularly in the last two years has been an extraordinary expansion of the three core elements that I really write about in Atlas, so the extraction of data of non-renewable resources, and of course hidden labor. So what we've seen, particularly on the resources side, is a gigantic spike both in terms of energy and water and that's often the story that we don't hear. We're not aware that when we're told about the fact that there gigantic hundred billion computers that are now being developed for the next stage of generative AI that has an enormous energy and water footprint. So I've been researching that along with many others who are now increasingly concerned about how we might think about AI more holistically.Eric Topol (03:52):Well, let's go back to your book, which is an extraordinary book, the AI Atlas and how you dissected not just the well power of politics and planetary costs, but that has won awards and it was a few years back, and I wonder so much has changed since then. I mean ChatGPT in late 2022 caught everybody off guard who wasn't into this knowing that this has been incubating for a number of years, and as you said, these base models are just extraordinary in every parameter you can think about, particularly the computing resource and consumption. So your concerns were of course registered then, have they gone to exponential growth now?Kate Crawford (04:45):I love the way you put that. I think you're right. I think my concerns have grown exponentially with the models. But I was like everybody else, even though I've been doing this for a long time and I had something of a heads up in terms of where we were moving with transformer models, I was also quite taken aback at the extraordinary uptake of ChatGPT back in November 2022 in fact, gosh, it still feels like yesterday it's been such an extraordinary timescale. But looking at that shift to a hundred million users in two months and then the sort of rapid competition that was emerging from the major tech companies that I think really took me by surprise, the degree to which everybody was jumping on the bandwagon, applying some form of large language model to everything and anything suddenly the hammer was being applied to every single nail.(05:42):And in all of that sound and fury and excitement, I think there will be some really useful applications of these tools. But I also think there's a risk that we apply it in spaces where it's really not well suited that we are not looking at the societal and political risks that come along with these approaches, particularly next token prediction as a way of generating knowledge. And then finally this bigger set of questions around what is it really costing the planet to build these infrastructures that are really gargantuan? I mean, as a species, we haven't invested this much money into an infrastructure like this really until you go back to the pyramids, you really got to go very far back to say that type of just gargantuan spending in terms of capital, in terms of labor, in terms of all of the things are required to really build these kinds of systems. So for me, that's the moment that we're in right now and perhaps here together in 2024, we can take a breath from that extraordinary 18 month period and hopefully be a little more reflective on what we're building and why and where will it be best used.Propagation of BiasesEric Topol (06:57):Yeah. Well, there's so many aspects of this that I'd like to get into with you. I mean, one of course, you're as a keen observer and activist in this whole space, you've made I think a very clear point about how our culture is mirrored in our AI that is our biases, and people are of course very quick to blame AI per se, but it seems like it's a bigger problem than just that. Maybe you could comment about, obviously biases are a profound concern about propagation of them, and where do you see where the problem is and how it can be attacked?Kate Crawford (07:43):Well, it is an enormous problem, and it has been for many years. I was first really interested in this question in the era that was known as the big data era. So we can think about the mid-2000s, and I really started studying large scale uses of data in scientific applications, but also in what you call social scientific settings using things like social media to detect and predict opinion, movement, the way that people were assessing key issues. And time and time again, I saw the same problem, which is that we have this tendency to assume that with scale comes greater accuracy without looking at the skews from the data sources. Where is that data coming from? What are the potential skews there? Is there a population that's overrepresented compared to others? And so, I began very early on looking at those questions. And then when we had very large-scale data sets start to emerge, like ImageNet, which was really perhaps the most influential dataset behind computer vision that was released in 2009, it was used widely, it was freely available.(09:00):That version was available for over a decade and no one had really looked inside it. And so, working with Trevor Paglen and others, we analyzed how people were being represented in this data set. And it was really quite extraordinary because initially people are labeled with terms that might seem relatively unsurprising, like this is a picture of a nurse, or this is a picture of a doctor, or this is a picture of a CEO. But then you look to see who is the archetypical CEO, and it's all pictures of white men, or if it's a basketball player, it's all pictures of black men. And then the labeling became more and more extreme, and there are terms like, this is an alcoholic, this is a corrupt politician, this is a kleptomaniac, this is a bad person. And then a whole series of labels that are simply not repeatable on your podcast.(09:54):So in finding this, we were absolutely horrified. And again, to know that so many AI models had trained on this as a way of doing visual recognition was so concerning because of course, very few people had even traced who was using this model. So trying to do the reverse engineering of where these really problematic assumptions were being built in hardcoded into how AI models see and interpret the world, that was a giant unknown and remains to this day quite problematic. We did a recent study that just came out a couple of months ago looking at one of the biggest data sets behind generative AI systems that are doing text to image generation. It's called LAION-5B, which stands for 5 billion. It has 5 billion images and text captions drawn from the internet. And you might think, as you said, this will just mirror societal biases, but it's actually far more weird than you might imagine.(10:55):It's not a representative sample even of the internet because particularly for these data sets that are now trying to use the ALT tags that are used around images, who uses ALT tags the most on the internet? Well, it's e-commerce sites and it's often stock image sites. So what you'll see and what we discovered in our study was that the vast majority of images and labels are coming from sites like Shopify and Pinterest, these kind of shopping aspirational collection sites. And that is a very specific way of seeing the world, so it's by no means even a perfect mirror. It's a skewed mirror in multiple ways. And that's something that we need to think of particularly when we turn to more targeted models that might be working in say healthcare or in education or even in criminal justice, where we see all sorts of problems emerge.Exploiting Humans for RLHFEric Topol (11:51):Well, that's really interesting. I wonder to extend that a bit about the human labor side of this. Base models are tweaked, fine-tuned, and one of the ways to do that, of course is getting people to weigh in. And this has been written about quite a bit about how the people that are doing this can be exploited, getting wages that are ridiculously weak. And I wonder if you could comment about that because in the ethics of AI, this seems to be one of the many things that a lot of people don't realize about reinforcement learning.Kate Crawford (12:39):Oh, I completely agree. It's quite an extraordinary story. And of course now we have a new category of crowd labor that's called reinforcement learning with human feedback or RLHF. And what was discovered by multiple investigations was that these laborers are in many cases paid less than $2 an hour in very exploitative conditions, looking at results that in many cases are really quite horrifying. They could be accounts of murder, suicide, trauma, this can be visual material, it can be text-based material. And again, the workers in these working for these companies, and again, it's often contract labor, it's not directly within a tech company, it's contracted out. It's very hidden, it's very hard to research and find. But these laborers have been experiencing trauma and are really now in many cases bringing lawsuits, but also trying to unionize and say, these are not acceptable conditions for people to be working under.(13:44):So in the case of OpenAI, it was found that it was Kenyan workers who were doing this work for just poverty wages, but it's really across the board. It's so common now that humans are doing the hard work behind the scenes to make these systems appear autonomous. And that's the real trap that we're being told that this is the artificial intelligence. But in actual fact, what Jeff Bezos calls Mechanical Turk is that it's artificial, artificial intelligence otherwise known as human beings. So that is a very significant layer in terms of how these systems work that is often unacknowledged. And clearly these workers in many cases are muzzled from speaking, they're not allowed to talk about what they do, they can't even tell their families. They're certainly prevented from collective action, which is why we've seen this push towards unionization. And finally, of course, they're not sharing in any of the profits that are being generated by these extraordinary new systems that are making a very small number of people, very wealthy indeed.Eric Topol (14:51):And do you know if that's improving or is it still just as bad as it has been reported? It's really deeply concerning to see human exploitation, and we all know well about sweatshops and all that, but here's another version, and it's really quite distressing.Kate Crawford (15:09):It really is. And in fact, there have been several people now working to create really almost like fair work guidelines. So Oxford has the sort of fair work initiative looking specifically at crowd work. They also have a rating system where they rate all of the major technology companies for how well they're treating their crowd laborers. And I have to say the numbers aren't looking good in the last 12 months, so I would love to see much more improvement there. We are also starting to see legislation be tabled specifically on this topic. In fact, Germany was one of the most recent to start to explore how they would create a strong legislative backing to make sure that there's fair labor conditions. Also, Chile was actually one of the first to legislate in this space, but you can imagine it's very difficult to do because it's a system that is operating under the radar through sort of multiple contracted chains. And even some of the people within tech companies will tell me it's really hard to know if they're working with a company that's doing this in the right way and paying people well. But frankly, I'd like to see far greater scrutiny otherwise, as you say, we're building on this system, which looks like AI sweatshops.Eric Topol (16:24):Yeah, no, I think people just have this illusion that these machines are doing everything by themselves, and that couldn't be further from the truth, especially when you're trying to take it to the next level. And there's only so much human content you can scrape from the internet, and obviously it needs additional input to take it to that more refined performance. Now, besides your writing and being much of a conscience for AI, you're also a builder. I mean, I first got to know some of your efforts through when you started the AI Now Institute. Maybe you can tell us a bit about that. Now you're onto the Knowing Machines Project and I don't know how many other projects you're working on, so maybe you can tell us about what it's like not just to be a keen observer, but also one to actually get initiatives going.Kate Crawford (17:22):Well, I think it's incredibly important that we start to build interdisciplinary coalitions of researchers, but sometimes even beyond the academic field, which is where I really initially trained in this space, and really thinking about how do we involve journalists, how do we involve filmmakers, how do we involve people who will look at these issues in really different ways and tell these stories more widely? Because clearly this really powerful shift that we're making as a society towards using AI in all sorts of domains is also a public issue. It's a democratic issue and it's an issue where we should all be able to really see into how these systems are working and have a say in how they'll be impacting our lives. So one of the things that I've done is really create research groups that are interdisciplinary, starting at Microsoft Research as one of the co-founders of FATE, a group that stands for fairness, accountability, transparency and ethics, and then the AI Now Institute, which was originally at NYU, and now with Knowing Machines, which is an international group, which I've been really delighted to build, rather than just purely focusing on those in the US because of course these systems are inherently transnational, they will be affecting global populations.(18:42):So we really need to think about how do you bring people from very different perspectives with different training to ask this question around how are these systems being built, who is benefiting and who might be harmed, and how can we address those issues now in order to actually prevent some of those harms and prevent the greatest risks that I see that are possible with this enormous turn to artificial intelligence everywhere?Eric Topol (19:07):Yeah, and it's interesting how you over the years are a key advisor, whether it's the White House, the UN or the European Parliament. And I'm curious about your experience because I didn't know much about the Paris ENS. Can you tell us about you were Visiting Chair, this is AI and Justice at the École Normale Supérieure (ENS), I don't know if I pronounce that right. My French is horrible, but this sounds like something really interesting.Kate Crawford (19:42):Well, it was really fascinating because this was the first time that ENS, which is really one of the top research institutions in Europe, had turned to this focus of how do we contend with artificial intelligence, not just as a technical question, but as a sort of a profound question of justice of society of ethics. And so, I was invited to be the first visiting chair, but tragically this corresponded with the start of the pandemic in 2020. And so, it ended up being a two-year virtual professorship, which is really a tragedy when you're thinking about spending time in Paris to be spending it on Zoom. It's not quite the same thing, but I had the great fortune of using that time to assemble a group of scholars around the world who were looking at these questions from very different disciplines. Some were historians of science, others were sociologists, some were philosophers, some were machine learners.(20:39):And really essentially assembled this group to think through some of the leading challenges in terms the potential social impacts and current social impacts of these systems. And so, we just recently published that through the academies of Science and Engineering, and it's been almost like a template for thinking about here are core domains that need more research. And interestingly, we're at that moment, I think now where we can say we have to look in a much more granular fashion beyond the hype cycles, beyond the sense of potential, the enormous potential upside that we're always hearing about to look at, okay, how do these systems actually work now? What kinds of questions can we bring into the research space so that we're really connecting the ideas that come traditionally from the social sciences and the humanistic disciplines into the world of machine learning and AI design. That's where I see the enormous upside that we can no longer stay in these very rigorously patrolled silos and to really use that interdisciplinary awareness to build systems differently and hopefully more sustainably as well.Is Working At Microsoft A Conflict?Eric Topol (21:55):Yeah, no, that's what I especially like about your work is that you're not a doomsday person or force. You're always just trying to make it better, but now that's what gets me to this really interesting question because you are a senior principal researcher at Microsoft and Microsoft might not like some of these things that you're advocating, how does that potential conflict work out?Kate Crawford (22:23):It's interesting. I mean, people often ask me, am I a technology optimist or a technology pessimist? And I always say I'm a technology realist, and we're looking at these systems being used. I think we are not benefited by discourses of AI doomerism nor by AI boosterism. We have to assess the real politic and the political economies into which these systems flow. So obviously part of the way that I've got to know what I know about how systems are designed and how they work at scale is through being at Microsoft Research where I'm working alongside extraordinary colleagues and all of whom come from, in many cases, professorial backgrounds who are deep experts in their fields. And we have this opportunity to work together and to look at these questions very early on in the kinds of production cycles and enormous shifts in the way that we use technology.(23:20):But it is interesting of course that at the moment Microsoft is absolutely at the leading edge of this change, and I've always thought that it's incredibly important for researchers and academics who are in industrial spaces to be able to speak freely, to be able to share what they see and to use that as a way that the industry can, well hopefully keep itself honest, but also share between what it knows and what everybody else knows because there's a giant risk in having those spaces be heavily demarcated and having researchers really be muzzled. I think that's where we see real problems emerge. Of course, one of the great concerns a couple of years ago was when Timnit Gebru and others were fired from Google for speaking openly about the concerns they had about the first-generation large language models. And my hope is that there's been a lesson through that really unfortunate set of decisions made at Google that we need people speaking from the inside about these questions in order to actually make these systems better, as you say, over the medium and long term.Eric Topol (24:26):Yeah, no, that brings me to thought of Peter Lee, who I'm sure because he wrote a book about GPT-4 and healthcare and was very candid about its potential, real benefits and the liabilities, and he's a very humble kind of guy. He's not one that has any bravado that I know of, so it speaks well to at least another colleague of yours there at Microsoft and their ability to see all the different sides here, not just what we'll talk about in a minute the arms race both across companies and countries. But before I get to that, there's this other part of you and I wonder if there's really two or three of you that is as a composer of music and art, I looked at your Anatomy of an AI System, I guess, which is on exhibit at the Museum of Modern Art (MoMA) in New York, and that in itself is amazing, but how do you get into all these other parts, are these hobbies or is this part of a main part of your creative work or where does it fit in?Kate Crawford (25:40):Eric, didn't I mention the cloning program that I participated in early and that there are many Kate's and it's fantastic we all work together. Yeah, that explains it. Look, it's interesting. Way back as a teenager, I was fascinated with technology. Of course, it was the early stages of the web at that moment, and I could see clearly that this was, the internet was going to completely change everything from my generation in terms of what we would do in terms of the way that we would experience the world. And as I was also at that time an electronic musician in bands, I was like, this was a really fantastic combination of bringing together creative practice with a set of much larger concerns and interests around at a systems level, how technology and society are co-constituted, how they evolve together and shape each other. And that's really been the map of how I've always worked across my life.(26:48):And it's interesting, I've always collaborated with artists and Vladan Joler who I worked with on anatomy of an AI system. We actually met at a conference on voice enabled AI systems, and it was really looking at the ethics of could it be possible to build an open source, publicly accessible version of say Alexa rather than purely a private model owned by a corporation, and could that be done in a more public open source way? And we asked a different question, we looked at each other and we're like, oh, I haven't met you yet, but I can see that there are some problems here. One of them is it's not just about the data and it's not just about the technical pipelines, it's about where the components come from. It's about the mining structures that needed to make all of these systems. It's about the entire end of life what happens when we throw these devices out from generally between three to four years of use and how they go into these giant e-waste tips.(27:51):And we basically started looking at this as an enormous sort of life and death of a single AI system, which for us started out by drawing these things on large pieces of butcher's paper, which just expanded and expanded until we had this enormous systems level analysis of what it takes just to ask Alexa what the weather is today. And in doing that, it taught me a couple of things. One that people really want to understand all of the things that go into making an AI system work. This piece has had a very long life. It's been in over a hundred museums around the world. It's traveled further than I have, but it's also very much about that broader political economy that AI systems aren't neutral, they don't just exist to serve us. They are often sort of fed into corporate structures that are using them to generate profits, and that means that they're used in very particular ways and that there are these externalities in terms of how they produced that linger in our environments that have really quite detrimental impacts on systems of labor and how people are recompensed and a whole range of relationships to how data is seen and used as though it's a natural resource that doesn't actually come from people's lives, that doesn't come with risks attached to it.(29:13):So that project was really quite profound for me. So we've continued to do these kinds of, I would call them research art projects, and we just released a new one called Calculating Empires, which looks at a 500 year history of technology and power looking specifically at how empires over time have used new technologies to centralize their power and expand and grow, which of course is part of what we're seeing at the moment in the empires of AI.Eric Topol (29:43):And what about the music side?Kate Crawford (29:45):Well, I have to say I've been a little bit slack on the music side. Things have been busy in AI Eric, I have to say it's kept me away from the music studio, but I always intend to get back there. Fortunately, I have a kid who's very musical and he's always luring me away from my desk and my research saying, let's write some music. And so, he'll keep me honest.Geopolitics and the Arms RacesEric Topol (30:06):Well, I think it's striking just because you have this blend of the humanities and you're so deep into trying to understand and improve our approaches in technology. And it seems like a very unusual, I don't know, too many techies that have these different dimensions, so that's impressive. Now let's get back to the arms race. You just were talking about tracing history over hundreds of years and empires, but right now we have a little problem. We have the big tech titans that are going after each other on a daily basis, and of course you know the group very well. And then you have China and the US that are vying to be the dominant force and problems with China accessing NVIDIA chips and Taiwan sitting there in a potentially very dangerous position, not just for Taiwan, but also for the US. And I wonder if you could just give us your sense about the tensions here. They're US based as well of course, because that's some of the major forces in companies, but then they're also globally. So we have a lot of stuff in the background that people don't like to think about, but it's actually happening right now.Kate Crawford (31:35):I think it's one of the most important things that we can focus on, in fact. I mean and again, this is why I think a materialist analysis of artificial intelligence is so important because not only does it force you to look at the raw components, where does the energy come from? Where does the water come from? But it means you're looking at where the chipsets come from. And you can see that in many cases there are these infrastructural choke points where we are highly dependent on specific components that sit within geopolitical flashpoints. And Taiwan is really the exemplar of this sort of choke point at the moment. And again, several companies are trying to address this by spinning up new factories to build these components, but this takes a lot of time and an enormous amount of resources yet again. So what we're seeing is I think a very difficult moment in the geopolitics of artificial intelligence.(32:31):What we've had certainly for the last decade has been almost a geopolitical duopoly. We've had the US and China not only having enormous power and influence in this space, but also goading each other into producing the most extreme forms of both data extractive and surveillance technologies. And unfortunately, this is just as true in the United States that I commonly hear this in rooms in DC where you'll hear advisors say, well, having any type of guardrails or ethical considerations for our AI systems is a problem if it means that China's going to do it anyway. And that creates this race to the bottom dynamic of do as much of whatever you can do regardless of the ethical and in some cases legal problems that will create. And I think that's been the dynamic that we've seen for some time. And of course the last 18 months to two years, we've seen that really extraordinary AI war happening internally in the United States where again, this race dynamic I think does create unfortunately this tendency to just go as fast as possible without thinking about potential downsides.(33:53):And I think we're seeing the legacy of that right now. And of course, a lot of the conversations from people designing these systems are now starting to say, look, being first is great, but we don't want to be in a situation as we saw recently with Google's Gemini where you have to pull an entire model off the shelves and you have to say, this is not ready. We actually have to remove it and start again. So this is the result I think of that high pressure, high speed dynamic that we've been seeing both inside the US but between the US and China. And of course, what that does to the rest of the world is create this kind of client states where we've got the EU trying to say, alright, well we'll export a regulatory model if we're not going to be treated as an equivalent player here. And then of course, so many other countries who are just seen as spaces to extract low paid labor or the mineralogical layer. So that is the big problem that I see is that that dynamic has only intensified in recent years.A.I. and MedicineEric Topol (34:54):Yeah, I know it's really another level of concern and it seems like it could be pretty volatile if for example, if the US China relations takes another dive and the tensions there go to levels that haven't been seen so far. I guess the other thing, there's so much that is I think controversial, unsettled in this space and so much excitement. I mean, just yesterday for example, was the first AI randomized trial to show that you could save lives. When I wrote that up, it was about the four other studies that showed how it wasn't working. Different studies of course, but there's so much excitement at the same time, there's deep concerns. You've been a master at articulating these deep concerns. What have we missed in our discussion today, I mean we've covered a lot of ground, but what do you see are other things that should be mentioned?Kate Crawford (36:04):Well, one of the things that I've loved in terms of following your work, Eric, is that you very carefully walk that line between allowing the excitement when we see really wonderful studies come out that say, look, there's great potential here, but also articulating concerns where you see them. So I think I'd love to hear, I mean take this opportunity to ask you a question and say what's exciting you about the way that this particularly new generation AI is being used in the medical context and what are the biggest concerns you have there?Eric Topol (36:35):Yeah, and it's interesting because the biggest advance so far in research and medicine was the study yesterday using deep learning without any transformer large language model effort. And that's where that multiplicative of opportunity or potential is still very iffy, it's wobbly. I mean, it needs much more refinement than where we are right now. It's exciting because it is multimodal and it brings in the ability to bring all the layers of a human being to understand our uniqueness and then do much better in terms of, I got a piece coming out soon in Science about medical forecasting and how we could really get to prevention of conditions that people are at high risk. I mean like for example today the US preventive task force said that all women age 40 should have mammograms, 40.Kate Crawford (37:30):I saw that.Eric Topol (37:30):Yeah, and this is just crazy Looney Tunes because here we have the potential to know pretty precisely who are those 12%, only 12% of women who would ever get breast cancer in their lifetime, and why should we put the other 88% through all this no less the fact that there are some women even younger than age 40 that have significantly high risk that are not picked up. But I do think eventually when we get these large language models to actualize their potential, we'll do really great forecasting and we'll be able to not just prevent or forestall cancer, Alzheimer's and so many things. It's quite exciting, but it's the earliest, we're not even at first base yet, but I think I can see our way to get there eventually. And it's interesting because the discussion I had previously with Geoffrey Hinton, and I wonder if you think this as well, that he sees the health medical space as the only really safe space. He thinks most everything else has got more concerns about the downsides is the sweet spot as he called it. But I know that's not particularly an area that you are into, but I wonder if you share that the excitement about your health could be improved in the future with AI.Kate Crawford (38:52):Well, I think it's a space of enormous potential, but again, enormous risk for the same reasons that we discussed earlier, which is we have to look at the training data and where it's coming from. Do we have truly representative sources of data? And this of course has been a consistent problem certainly for the last hundred years and longer. When we look at who are the medical patients whose data is being collected, are we seeing skews? And that has created all sorts of problems, particularly in the last 50 years in terms of misdiagnosing women, people of color, missing and not taking seriously the health complaints of people who are already seen as marginalized populations, thus then further skewing the data that is then used to train AI models. So this is something that we have to take very seriously, and I had the great fortune of being invited by Francis Collins to work with the NIH on their AI advisory board.(39:50):They produced a board to look just at these questions around how can this moment in AI be harnessed in such a way that we can think about the data layer, think about the quality of data and how we train models. And it was a really fascinating sort of year long discussion because in the room we had people who were just technologists who just wanted as much data as possible and just give us all that data and then we'll do something, but we'll figure it out later. Then there were people who had been part of the Human Genome Project and had worked with Francis on questions around the legal and ethical and social questions, which he had really centered in that project very early on. And they said, no, we have to learn these lessons. We have to learn that data comes from somewhere. It's not divorced of context, and we have to think about who's being represented there and also who's not being represented there because that will then be intensified in any model that we train on that data.Humans and Automation Bias(40:48):And then also thinking about what would happen in terms of if those models are only held by a few companies who can profit from them and not more publicly and widely shared. These were the sorts of conversations that I think at the absolute forefront in terms of how we're going to navigate this moment. But if we get that right, if we center those questions, then I think we have far greater potential here than we might imagine. But I'm also really cognizant of the fact that even if you have a perfect AI model, you are always going to have imperfect people applying it. And I'm sure you saw that same study that came out in JAMA back in December last year, which was looking at how AI bias, even slightly biased models can worsen human medical diagnosis. I don't know if you saw this study, but I thought it was really extraordinary.(41:38):It was sort of 450 doctors and physician's assistants and they were really being shown a handful of cases of patients with acute respiratory failure and they really needed come up with some sort of diagnosis and they were getting suggestions from an AI model. One model was trained very carefully with highly accurate data, and the other was a fairly shoddy, shall we say, AI model with quite biased data. And what was interesting is that the clinicians when they were working with very well-trained AI model, we're actually producing a better diagnosis across the board in terms of the cases they were looking at. I think their accuracy went up by almost 4.5 percentage points, but when they were working with the less accurate model, their capacity actually dropped well below their usual diagnostic baseline, something like almost 12 percentage points below their usual diagnostic quality. And so, this really makes me think of the kind of core problem that's been really studied for 40 years by social scientists, which is called automation bias, which is when even an expert, a technical system which is giving a recommendation, our tendency is to believe it and to discard our own knowledge, our own predictions, our own sense.(42:58):And it's been tested with fighter pilots, it's been tested with doctors, it's been tested with judges, and it's the same phenomenon across the board. So one of the things that we're going to need to do collectively, but particularly in the space of medicine and healthcare, is retaining that skepticism, retaining that ability to ask questions of where did this recommendation come from with this AI system and should I trust it? What was it trained on? Where did the data come from? What might those gaps be? Because we're going to need that skepticism if we're going to get through particularly this, as you say, this sort of early stage one period where in many cases these models just haven't had a lot of testing yet and people are going to tend to believe them out of the box.The Large Language Model Copyright IssueEric Topol (43:45):No, it's so true. And one of the key points is that almost every study that's been published in large language models in medicine are contrived. They're using patient actors or they're using case studies, but they're not in the real world. And that's where you have to really learn, as you know, that's a much more complex and messy world than the in silico world of course. Now, before wrapping up, one of the things that's controversial we didn't yet hit is the fact that in order for these base models to get trained, they basically ingest all human content. So they've ingested everything you've ever written, your books, your articles, my books, my articles, and you have the likes of the New York Times suing OpenAI, and soon it's going to run out of human content and just use synthetic content, I guess. But what's your sense about this? Do you feel that that's trespassing or is this another example of exploiting content and people, or is this really what has to be done in order to really make all this work?Kate Crawford (44:59):Well, isn't it a fascinating moment to see this mass grabbing of data, everything that is possibly extractable. I actually just recently published an article in Grey Room with the legal scholar, Jason Schultz, looking at how this is producing a crisis in copyright law because in many ways, copyright law just cannot contend with generative AI in particular because all of the ways in which copyright law and intellectual property more broadly has been understood, has been premised around human ideas of providing an incentive and thus a limited time monopoly based on really inspiring people to create more things. Well, this doesn't apply to algorithms, they don't respond to incentives in this way. The fact that, again, it's a longstanding tradition in copyright that we do not give copyright to non-human authors. So you might remember that there was a very famous monkey selfie case where a monkey had actually stepped on a camera and it had triggered a photograph of the monkey, and could this actually be a copyright image that could be given to the monkey?(46:12):Absolutely not, is what the court's decided. And the same has now happened, of course, for all generative AI systems. So right now, everything that you produce be that in GPT or in Midjourney or in Stable Diffusion, you name it, that does not have copyright protections. So we're in the biggest experiment of production after copyright in world history, and I don't think it's going to last very long. To be clear, I think we're going to start to see some real shifts, I think really in the next 6 to 12 months. But it has been this moment of seeing this gigantic gap in what our legal structures can do that they just haven't been able to contend with this moment. The same thing is true, I think, of ingestion, of this capturing of human content without consent. Clearly, many artists, many writers, many publishing houses like the New York Times are very concerned about this, but the difficulty that they're presented with is this idea of fair use, that you can collect large amounts of data if you are doing something with that, which is sufficiently transformative.(47:17):I'm really interested in the question of whether or not this does constitute sufficiently transformative uses. Certainly if you looked at the way that large language models a year ago, you could really prompt them into sharing their training data, spitting out entire New York Times articles or entire book chapters. That is no longer the case. All of the major companies building these systems have really safeguarded against that now but nonetheless, you have this question of should we be moving towards a system that is based on licensing, where we're really asking people if we can use their data and paying them a license fee? You can see how that could absolutely work and would address a lot of these concerns, but ultimately it will rely on this question of fair use. And I think with the current legal structures that we have in the current case law, that is unlikely to be seen as something that's actionable.(48:10):But I expect what we'll look at is what really happened in the early 20th century around the player piano, which was that I'm sure you remember this extraordinary technology of the player piano. That was one of the first systems that automated the playing of music and you'd have a piano that had a wax cylinder that almost like code had imprinted on a song or a piece of music, and it could be played in the public square or in a bar or in a saloon without having to pay a single artist and artists were terrified. They were furious, they were public hearings, there were sort of congressional hearings and even a Supreme Court case that decided that this was not a copyright infringement. This was a sufficiently transformative use of a piece of music that it could stand. And in the end, it was actually Congress that acted.(49:01):And we from that got the 1908 Copyright Act and from that we got this idea of royalties. And that has become the basis of the music industry itself for a very long time. And now we're facing another moment where I think we have a legislative challenge. How would you actually create a different paradigm for AI that would recognize a new licensing system that would reward artists, writers, musicians, all of the people whose work has been ingested into training data for AI so that they are recognized and in some ways, recompensed by this massive at scale extraction?Eric Topol (49:48):Wow, this has been an exhilarating conversation, Kate. I've learned so much from you over the years, but especially even just our chance to talk today. You articulate these problems so well, and I know you're working on solutions to almost everything, and you're so young, you could probably make a difference in the decades ahead. This is great, so I want to thank you not just for the chance to visit today, but all the work that you've been doing, you and your colleagues to make AI better, make it fulfill the great promise that it has. It is so extraordinary, and hopefully it'll deliver on some of the things that we have big unmet needs, so thanks to you. This has really been fun.Kate Crawford (50:35):This has been wonderful. And likewise, Eric, your work has just been a fantastic influence and I've been delighted to get to know you over the years and let's see what happens. It's going to be a wild ride from now to who knows when.Eric Topol (50:48):No question, but you'll keep us straight, I know that. Thank you so much.Kate Crawford (50:52):Thanks so much, Eric.*******************************Your support of subscribing to Ground Truths, and sharing it with your network of friends and colleagues, is much appreciated.The Ground Truths newsletters and podcasts are all free, open-access, without ads.Voluntary paid subscriptions all go to support Scripps Research. Many thanks for that—they greatly helped fund our summer internship programs for 2023 and 2024.Thanks to my producer Jessica Nguyen and Sinjun Balabanoff tor audio and video support at Scripps ResearchNote: you can select preferences to receive emails about newsletters, podcasts, or all I don't want to bother you with an email for content that you're not interested in.Comments for this post are welcome from all subscribers. Get full access to Ground Truths at erictopol.substack.com/subscribe

Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Towards Multimodal Interpretability: Learning Sparse Interpretable Features in Vision Transformers, published by hugofry on April 30, 2024 on LessWrong. Two Minute Summary In this post I present my results from training a Sparse Autoencoder (SAE) on a CLIP Vision Transformer (ViT) using the ImageNet-1k dataset. I have created an interactive web app, 'SAE Explorer', to allow the public to explore the visual features the SAE has learnt, found here: https://sae-explorer.streamlit.app/ (best viewed on a laptop). My results illustrate that SAEs can identify sparse and highly interpretable directions in the residual stream of vision models, enabling inference time inspections on the model's activations. To demonstrate this, I have included a 'guess the input image' game on the web app that allows users to guess the input image purely from the SAE activations of a single layer and token of the residual stream. I have also uploaded a (slightly outdated) accompanying talk of my results, primarily listing SAE features I found interesting: https://youtu.be/bY4Hw5zSXzQ. The primary purpose of this post is to demonstrate and emphasise that SAEs are effective at identifying interpretable directions in the activation space of vision models. In this post I highlight a small number my favourite SAE features to demonstrate some of the abstract concepts the SAE has identified within the model's representations. I then analyse a small number of SAE features using feature visualisation to check the validity of the SAE interpretations. Later in the post, I provide some technical analysis of the SAE. I identify a large cluster of features analogous to the 'ultra-low frequency' cluster that Anthropic identified. In line with existing research, I find that this ultra-low frequency cluster represents a single feature. I then analyse the 'neuron-alignment' of SAE features by comparing the SAE encoder matrix the MLP out matrix. This research was conducted as part of the ML Alignment and Theory Scholars program 2023/2024 winter cohort. Special thanks to Joseph Bloom for providing generous amounts of his time and support (in addition to the SAE Lens code base) as well as LEAP labs for helping to produce the feature visualisations and weekly meetings with Jessica Rumbelow. Example, animals eating other animals feature: (top 16 highest activating images) Example, Italian feature: Note that the photo of the dog has a watermark with a website ending in .it (Italy's domain name). Note also that the bottom left photo is of Italian writing. The number of ambulances present is a byproduct of using ImageNet-1k. Motivation Frontier AI systems are becoming increasingly multimodal, and capabilities may advance significantly as multimodality increases due to transfer learning between different data modalities and tasks. As a heuristic, consider how much intuition humans gain for the world through visual reasoning; even in abstract settings such as in maths and physics, concepts are often understood most intuitively through visual reasoning. Many cutting edge systems today such as DALL-E and Sora use ViTs trained on multimodal data. Almost by definition, AGI is likely to be multimodal. Despite this, very little effort has been made to apply and adapt our current mechanistic interpretability techniques to vision tasks or multimodal models. I believe it is important to check that mechanistic interpretability generalises to these systems in order to ensure they are future-proof and can be applied to safeguard against AGI. In this post, I restrict the scope of my research to specifically investigating SAEs trained on multimodal models. The particular multimodal system I investigate is CLIP, a model trained on image-text pairs. CLIP consists of two encoders: a language model and a vision model that are trained to e...

At 15, Fei-Fei Li transitioned from a middle-class life in China to poverty in America. Despite the pressures of her family's financial situation and her mother's ailing health, her knack for physics never wavered. She went from learning English as a second language to attending and working at prestigious institutions like Princeton and Stanford. Today, she is among a handful of scientists behind the impressive advances of artificial intelligence in recent times. In this episode, she breaks down her human-centered approach to AI and explores the future of the technology. Dr. Fei-Fei Li is a professor of Computer Science at Stanford University and the co-director of the Stanford Institute for Human-Centered AI. She is the creator of ImageNet, a key driver of modern artificial intelligence. With over 20 years at the forefront of the field, Dr. Li is focused on AI research, education, and policy to improve the human condition.    In this episode, Hala and Fei-Fei will discuss: - The current capabilities of AI - The difference between machine learning and AI - The training process for AI models - The gaps in our knowledge about how AI learns - Why ChatGPT fails at higher-level reasoning like math - The biological inspiration for vision in computers - Fears and hopes associated with AI - The human element of jobs AI can't replace - Augmentation of human capabilities through AI - The three pillars of her human-centered AI framework - Responsible development and use of AI - The roadblocks to be aware of when using AI - Her advice to young entrepreneurs navigating the AI world  - And other topics…   Dr. Fei-Fei Li is a professor of Computer Science at Stanford University and the co-director of the Stanford Institute for Human-Centered AI. She is also the creator of ImageNet and the ImageNet Challenge, a key catalyst to the latest developments in deep learning and AI. Sometimes called the ‘Godmother of AI,' she is a pioneer in early computer vision research. Dr. Li is the author of The Worlds I See, one of Barack Obama's recommended books on AI. Her work has been featured in various publications, including the New York Times, Wall Street Journal, Fortune Magazine, Science, and Wired Magazine.  Connect with Fei-Fei: Fei-Fei's Bio: https://profiles.stanford.edu/fei-fei-li  Fei-Fei's LinkedIn: https://www.linkedin.com/in/fei-fei-li-4541247/  Fei-Fei's Twitter: https://twitter.com/drfeifei  Resources Mentioned: Fei-Fei's Book, The Worlds I See: Curiosity, Exploration, and Discovery at the Dawn of AI: https://www.amazon.com/Worlds-See-Curiosity-Exploration-Discovery-ebook/dp/B0BPQSLVL6   Stanford Human Center AI Institute Website: https://hai.stanford.edu/  LinkedIn Secrets Masterclass, Have Job Security For Life: Use code ‘podcast' for 30% off at yapmedia.io/course. Sponsored By: Shopify - Sign up for a one-dollar-per-month trial period at youngandprofiting.co/shopify Indeed - Get a $75 job credit at indeed.com/profiting Yahoo Finance - For comprehensive financial news and analysis, visit YahooFinance.com More About Young and Profiting Download Transcripts - youngandprofiting.com Get Sponsorship Deals - youngandprofiting.com/sponsorships Leave a Review - ratethispodcast.com/yap Watch Videos - youtube.com/c/YoungandProfiting   Follow Hala Taha LinkedIn - linkedin.com/in/htaha/ Instagram - instagram.com/yapwithhala/ TikTok - tiktok.com/@yapwithhala Twitter - twitter.com/yapwithhala   Learn more about YAP Media's Services - yapmedia.io/

Hugo speaks with Johno Whitaker, a Data Scientist/AI Researcher doing R&D with answer.ai. His current focus is on generative AI, flitting between different modalities. He also likes teaching and making courses, having worked with both Hugging Face and fast.ai in these capacities. Johno recently reminded Hugo how hard everything was 10 years ago: “Want to install TensorFlow? Good luck. Need data? Perhaps try ImageNet. But now you can use big models from Hugging Face with hi-res satellite data and do all of this in a Colab notebook. Or think ecology and vision models… or medicine and multimodal models!” We talk about where we've come from regarding tooling and accessibility for foundation models, ML, and AI, where we are, and where we're going. We'll delve into What the Generative AI mindset is, in terms of using atomic building blocks, and how it evolved from both the data science and ML mindsets; How fast.ai democratized access to deep learning, what successes they had, and what was learned; The moving parts now required to make GenAI and ML as accessible as possible; The importance of focusing on UX and the application in the world of generative AI and foundation models; The skillset and toolkit needed to be an LLM and AI guru; What they're up to at answer.ai to democratize LLMs and foundation models. LINKS The livestream on YouTube (https://youtube.com/live/hxZX6fBi-W8?feature=share) Zindi, the largest professional network for data scientists in Africa (https://zindi.africa/) A new old kind of R&D lab: Announcing Answer.AI (http://www.answer.ai/posts/2023-12-12-launch.html) Why and how I'm shifting focus to LLMs by Johno Whitaker (https://johnowhitaker.dev/dsc/2023-07-01-why-and-how-im-shifting-focus-to-llms.html) Applying AI to Immune Cell Networks by Rachel Thomas (https://www.fast.ai/posts/2024-01-23-cytokines/) Replicate -- a cool place to explore GenAI models, among other things (https://replicate.com/explore) Hands-On Generative AI with Transformers and Diffusion Models (https://www.oreilly.com/library/view/hands-on-generative-ai/9781098149239/) Johno on Twitter (https://twitter.com/johnowhitaker) Hugo on Twitter (https://twitter.com/hugobowne) Vanishing Gradients on Twitter (https://twitter.com/vanishingdata) SciPy 2024 CFP (https://www.scipy2024.scipy.org/#CFP) Escaping Generative AI Walled Gardens with Omoju Miller, a Vanishing Gradients Livestream (https://lu.ma/xonnjqe4)

The conventional recipe for maximizing model accuracy is to (1) train multiple models with various hyperparameters and (2) pick the individual model which performs best on a held-out validation set, discarding the remainder. In this paper, we revisit the second step of this procedure in the context of fine-tuning large pre-trained models, where fine-tuned models often appear to lie in a single low error basin. We show that averaging the weights of multiple models fine-tuned with different hyperparameter configurations often improves accuracy and robustness. Unlike a conventional ensemble, we may average many models without incurring any additional inference or memory costs -- we call the results"model soups."When fine-tuning large pre-trained models such as CLIP, ALIGN, and a ViT-G pre-trained on JFT, our soup recipe provides significant improvements over the best model in a hyperparameter sweep on ImageNet. The resulting ViT-G model, which attains 90.94% top-1 accuracy on ImageNet, achieved a new state of the art. Furthermore, we show that the model soup approach extends to multiple image classification and natural language processing tasks, improves out-of-distribution performance, and improves zero-shot performance on new downstream tasks. Finally, we analytically relate the performance similarity of weight-averaging and logit-ensembling to flatness of the loss and confidence of the predictions, and validate this relation empirically. Code is available at https://github.com/mlfoundations/model-soups. 2022: Mitchell Wortsman, Gabriel Ilharco, S. Gadre, R. Roelofs, Raphael Gontijo-Lopes, Ari S. Morcos, Hongseok Namkoong, Ali Farhadi, Y. Carmon, Simon Kornblith, Ludwig Schmidt https://arxiv.org/pdf/2203.05482.pdf

Learned representations are a central component in modern ML systems, serving a multitude of downstream tasks. When training such representations, it is often the case that computational and statistical constraints for each downstream task are unknown. In this context rigid, fixed capacity representations can be either over or under-accommodating to the task at hand. This leads us to ask: can we design a flexible representation that can adapt to multiple downstream tasks with varying computational resources? Our main contribution is Matryoshka Representation Learning (MRL) which encodes information at different granularities and allows a single embedding to adapt to the computational constraints of downstream tasks. MRL minimally modifies existing representation learning pipelines and imposes no additional cost during inference and deployment. MRL learns coarse-to-fine representations that are at least as accurate and rich as independently trained low-dimensional representations. The flexibility within the learned Matryoshka Representations offer: (a) up to 14x smaller embedding size for ImageNet-1K classification at the same level of accuracy; (b) up to 14x real-world speed-ups for large-scale retrieval on ImageNet-1K and 4K; and (c) up to 2% accuracy improvements for long-tail few-shot classification, all while being as robust as the original representations. Finally, we show that MRL extends seamlessly to web-scale datasets (ImageNet, JFT) across various modalities -- vision (ViT, ResNet), vision + language (ALIGN) and language (BERT). MRL code and pretrained models are open-sourced at https://github.com/RAIVNLab/MRL. 2022: Aditya Kusupati, Gantavya Bhatt, Aniket Rege, Matthew Wallingford, Aditya Sinha, V. Ramanujan, William Howard-Snyder, Kaifeng Chen, S. Kakade, Prateek Jain, Ali Farhadi https://arxiv.org/pdf/2205.13147v3.pdf

Fei-Fei Li is a Stanford computer scientist and the former chief scientist of artificial intelligence/machine learning at Google Cloud. When Li entered the field of AI in the 2000s, researchers were making slow progress, optimizing algorithms to incrementally improve outcomes. Li saw that the problem wasn't the algorithm, but the size of the datasets being used. So she built a massive database of images called ImageNet. It was a huge breakthrough, and helped lead the emergence of modern AI.See omnystudio.com/listener for privacy information.

We are running an end of year listener survey! Please let us know any feedback you have, what episodes resonated with you, and guest requests for 2024! Survey link here.NeurIPS 2023 took place from Dec 10–16 in New Orleans. The Latent Space crew was onsite for as many of the talks and workshops as we could attend (and more importantly, hosted cocktails and parties after hours)!Picking from the 3586 papers accepted to the conference (available online, full schedule here) is an impossible task, but we did our best to present an audio guide with brief commentary on each. We also recommend MLContests.com NeurIPS recap and Seb Ruder's NeurIPS primer. We also found the VizHub guide useful for a t-SNE clustering of papers.We'll start with the NeurIPS Best Paper Awards, and then go to a selection of non-awarded but highly influential papers, and then arbitrary personal picks to round out the selection. Where we were able to do a poster session interview, please scroll to the relevant show notes for images of their poster for discussion. We give Chris Ré the last word due to the Mamba and StripedHyena state space models drawing particular excitement but still being too early to assess impact. Timestamps* [0:01:19] Word2Vec (Jeff Dean, Greg Corrado)* [0:15:28] Emergence Mirage (Rylan Schaeffer)* [0:28:48] DPO (Rafael Rafailov)* [0:41:36] DPO Poster Session (Archit Sharma)* [0:52:03] Datablations (Niklas Muennighoff)* [1:00:50] QLoRA (Tim Dettmers)* [1:12:23] DataComp (Samir Gadre)* [1:25:38] DataComp Poster Session (Samir Gadre, Alex Dimakis)* [1:35:25] LLaVA (Haotian Liu)* [1:47:21] LLaVA Poster Session (Haotian Liu)* [1:59:19] Tree of Thought (Shunyu Yao)* [2:11:27] Tree of Thought Poster Session (Shunyu Yao)* [2:20:09] Toolformer (Jane Dwivedi-Yu)* [2:32:26] Voyager (Guanzhi Wang)* [2:45:14] CogEval (Ida Momennejad)* [2:59:41] State Space Models (Chris Ré)Papers covered* Distributed Representations of Words and Phrases and their Compositionality (Word2Vec) Tomas Mikolov · Ilya Sutskever · Kai Chen · Greg Corrado · Jeff Dean. The recently introduced continuous Skip-gram model is an efficient method for learning high-quality distributed vector representations that capture a large number of precise syntactic and semantic word relationships. In this paper we present several improvements that make the Skip-gram model more expressive and enable it to learn higher quality vectors more rapidly. We show that by subsampling frequent words we obtain significant speedup, and also learn higher quality representations as measured by our tasks. We also introduce Negative Sampling, a simplified variant of Noise Contrastive Estimation (NCE) that learns more accurate vectors for frequent words compared to the hierarchical softmax. An inherent limitation of word representations is their indifference to word order and their inability to represent idiomatic phrases. For example, the meanings of Canada'' and "Air'' cannot be easily combined to obtain "Air Canada''. Motivated by this example, we present a simple and efficient method for finding phrases, and show that their vector representations can be accurately learned by the Skip-gram model.* Are Emergent Abilities of Large Language Models a Mirage? (Schaeffer et al.). Emergent abilities are abilities that are present in large-scale models but not in smaller models and are hard to predict. Rather than being a product of models' scaling behavior, this paper argues that emergent abilities are mainly an artifact of the choice of metric used to evaluate them. Specifically, nonlinear and discontinuous metrics can lead to sharp and unpredictable changes in model performance. Indeed, the authors find that when accuracy is changed to a continuous metric for arithmetic tasks where emergent behavior was previously observed, performance improves smoothly instead. So while emergent abilities may still exist, they should be properly controlled and researchers should consider how the chosen metric interacts with the model.* Direct Preference Optimization: Your Language Model is Secretly a Reward Model (Rafailov et al.)* While large-scale unsupervised language models (LMs) learn broad world knowledge and some reasoning skills, achieving precise control of their behavior is difficult due to the completely unsupervised nature of their training. Existing methods for gaining such steerability collect human labels of the relative quality of model generations and fine-tune the unsupervised LM to align with these preferences, often with reinforcement learning from human feedback (RLHF). However, RLHF is a complex and often unstable procedure, first fitting a reward model that reflects the human preferences, and then fine-tuning the large unsupervised LM using reinforcement learning to maximize this estimated reward without drifting too far from the original model. * In this paper, we leverage a mapping between reward functions and optimal policies to show that this constrained reward maximization problem can be optimized exactly with a single stage of policy training, essentially solving a classification problem on the human preference data. The resulting algorithm, which we call Direct Preference Optimization (DPO), is stable, performant, and computationally lightweight, eliminating the need for fitting a reward model, sampling from the LM during fine-tuning, or performing significant hyperparameter tuning. * Our experiments show that DPO can fine-tune LMs to align with human preferences as well as or better than existing methods. Notably, fine-tuning with DPO exceeds RLHF's ability to control sentiment of generations and improves response quality in summarization and single-turn dialogue while being substantially simpler to implement and train.* Scaling Data-Constrained Language Models (Muennighoff et al.)* The current trend of scaling language models involves increasing both parameter count and training dataset size. Extrapolating this trend suggests that training dataset size may soon be limited by the amount of text data available on the internet. Motivated by this limit, we investigate scaling language models in data-constrained regimes. Specifically, we run a large set of experiments varying the extent of data repetition and compute budget, ranging up to 900 billion training tokens and 9 billion parameter models. We find that with constrained data for a fixed compute budget, training with up to 4 epochs of repeated data yields negligible changes to loss compared to having unique data. However, with more repetition, the value of adding compute eventually decays to zero. We propose and empirically validate a scaling law for compute optimality that accounts for the decreasing value of repeated tokens and excess parameters. Finally, we experiment with approaches mitigating data scarcity, including augmenting the training dataset with code data or removing commonly used filters. Models and datasets from our 400 training runs are freely available at https://github.com/huggingface/datablations.* QLoRA: Efficient Finetuning of Quantized LLMs (Dettmers et al.). * This paper proposes QLoRA, a more memory-efficient (but slower) version of LoRA that uses several optimization tricks to save memory. They train a new model, Guanaco, that is fine-tuned only on a single GPU for 24h and outperforms previous models on the Vicuna benchmark. Overall, QLoRA enables using much fewer GPU memory for fine-tuning LLMs. Concurrently, other methods such as 4-bit LoRA quantization have been developed that achieve similar results.* DataComp: In search of the next generation of multimodal datasets (Gadre et al.)* Multimodal datasets are a critical component in recent breakthroughs such as CLIP, Stable Diffusion and GPT-4, yet their design does not receive the same research attention as model architectures or training algorithms. To address this shortcoming in the machine learning ecosystem, we introduce DataComp, a testbed for dataset experiments centered around a new candidate pool of 12.8 billion image-text pairs from Common Crawl. Participants in our benchmark design new filtering techniques or curate new data sources and then evaluate their new dataset by running our standardized CLIP training code and testing the resulting model on 38 downstream test sets. * Our benchmark consists of multiple compute scales spanning four orders of magnitude, which enables the study of scaling trends and makes the benchmark accessible to researchers with varying resources. Our baseline experiments show that the DataComp workflow leads to better training sets. Our best baseline, DataComp-1B, enables training a CLIP ViT-L/14 from scratch to 79.2% zero-shot accuracy on ImageNet, outperforming OpenAI's CLIP ViT-L/14 by 3.7 percentage points while using the same training procedure and compute. We release datanet and all accompanying code at www.datacomp.ai.* Visual Instruction Tuning (Liu et al)* Instruction tuning large language models (LLMs) using machine-generated instruction-following data has improved zero-shot capabilities on new tasks, but the idea is less explored in the multimodal field. In this paper, we present the first attempt to use language-only GPT-4 to generate multimodal language-image instruction-following data. * By instruction tuning on such generated data, we introduce LLaVA: Large Language and Vision Assistant, an end-to-end trained large multimodal model that connects a vision encoder and LLM for general-purpose visual and language understanding.* Our early experiments show that LLaVA demonstrates impressive multimodel chat abilities, sometimes exhibiting the behaviors of multimodal GPT-4 on unseen images/instructions, and yields a 85.1% relative score compared with GPT-4 on a synthetic multimodal instruction-following dataset. When fine-tuned on Science QA, the synergy of LLaVA and GPT-4 achieves a new state-of-the-art accuracy of 92.53%. We make GPT-4 generated visual instruction tuning data, our model and code base publicly available.* Tree of Thoughts: Deliberate Problem Solving with Large Language Models (Yao et al)* Language models are increasingly being deployed for general problem solving across a wide range of tasks, but are still confined to token-level, left-to-right decision-making processes during inference. This means they can fall short in tasks that require exploration, strategic lookahead, or where initial decisions play a pivotal role. * To surmount these challenges, we introduce a new framework for language model inference, Tree of Thoughts (ToT), which generalizes over the popular Chain of Thought approach to prompting language models, and enables exploration over coherent units of text (thoughts) that serve as intermediate steps toward problem solving. * ToT allows LMs to perform deliberate decision making by considering multiple different reasoning paths and self-evaluating choices to decide the next course of action, as well as looking ahead or backtracking when necessary to make global choices.* Our experiments show that ToT significantly enhances language models' problem-solving abilities on three novel tasks requiring non-trivial planning or search: Game of 24, Creative Writing, and Mini Crosswords. For instance, in Game of 24, while GPT-4 with chain-of-thought prompting only solved 4% of tasks, our method achieved a success rate of 74%. * Code repo with all prompts: https://github.com/princeton-nlp/tree-of-thought-llm.* Toolformer: Language Models Can Teach Themselves to Use Tools (Schick et al)* LMs exhibit remarkable abilities to solve new tasks from just a few examples or textual instructions, especially at scale. They also, paradoxically, struggle with basic functionality, such as arithmetic or factual lookup, where much simpler and smaller specialized models excel. * In this paper, we show that LMs can teach themselves to use external tools via simple APIs and achieve the best of both worlds. * We introduce Toolformer, a model trained to decide which APIs to call, when to call them, what arguments to pass, and how to best incorporate the results into future token prediction. * This is done in a self-supervised way, requiring nothing more than a handful of demonstrations for each API. We incorporate a range of tools, including a calculator, a Q&A system, a search engine, a translation system, and a calendar. * Toolformer achieves substantially improved zero-shot performance across a variety of downstream tasks, often competitive with much larger models, without sacrificing its core language modeling abilities.* Voyager: An Open-Ended Embodied Agent with Large Language Models (Wang et al)* We introduce Voyager, the first LLM-powered embodied lifelong learning agent in Minecraft that continuously explores the world, acquires diverse skills, and makes novel discoveries without human intervention. Voyager consists of three key components: * 1) an automatic curriculum that maximizes exploration, * 2) an ever-growing skill library of executable code for storing and retrieving complex behaviors, and * 3) a new iterative prompting mechanism that incorporates environment feedback, execution errors, and self-verification for program improvement. * Voyager interacts with GPT-4 via blackbox queries, which bypasses the need for model parameter fine-tuning. The skills developed by Voyager are temporally extended, interpretable, and compositional, which compounds the agent's abilities rapidly and alleviates catastrophic forgetting. Empirically, Voyager shows strong in-context lifelong learning capability and exhibits exceptional proficiency in playing Minecraft. It obtains 3.3x more unique items, travels 2.3x longer distances, and unlocks key tech tree milestones up to 15.3x faster than prior SOTA. Voyager is able to utilize the learned skill library in a new Minecraft world to solve novel tasks from scratch, while other techniques struggle to generalize.Voyager discovers new Minecraft items and skills continually by self-driven exploration, significantly outperforming the baselines.* Evaluating Cognitive Maps and Planning in Large Language Models with CogEval (Momennejad et al)* Recently an influx of studies claims emergent cognitive abilities in large language models (LLMs). Yet, most rely on anecdotes, overlook contamination of training sets, or lack systematic Evaluation involving multiple tasks, control conditions, multiple iterations, and statistical robustness tests. Here we make two major contributions. * First, we propose CogEval, a cognitive science-inspired protocol for the systematic evaluation of cognitive capacities in LLMs. The CogEval protocol can be followed for the evaluation of various abilities. * * Second, here we follow CogEval to systematically evaluate cognitive maps and planning ability across eight LLMs (OpenAI GPT-4, GPT-3.5-turbo-175B, davinci-003-175B, Google Bard, Cohere-xlarge-52.4B, Anthropic Claude-1-52B, LLaMA-13B, and Alpaca-7B). We base our task prompts on human experiments, which offer both established construct validity for evaluating planning, and are absent from LLM training sets.* * We find that, while LLMs show apparent competence in a few planning tasks with simpler structures, systematic evaluation reveals striking failure modes in planning tasks, including hallucinations of invalid trajectories and falling in loops. These findings do not support the idea of emergent out-of-the-box planning ability in LLMs. This could be because LLMs do not understand the latent relational structures underlying planning problems, known as cognitive maps, and fail at unrolling goal-directed trajectories based on the underlying structure. Implications for application and future directions are discussed.* Mamba: Linear-Time Sequence Modeling with Selective State Spaces (Albert Gu, Tri Dao)* Foundation models, now powering most of the exciting applications in deep learning, are almost universally based on the Transformer architecture and its core attention module. Many subquadratic-time architectures such as linear attention, gated convolution and recurrent models, and structured state space models (SSMs) have been developed to address Transformers' computational inefficiency on long sequences, but they have not performed as well as attention on important modalities such as language. We identify that a key weakness of such models is their inability to perform content-based reasoning, and make several improvements. * First, simply letting the SSM parameters be functions of the input addresses their weakness with discrete modalities, allowing the model to selectively propagate or forget information along the sequence length dimension depending on the current token. * Second, even though this change prevents the use of efficient convolutions, we design a hardware-aware parallel algorithm in recurrent mode. We integrate these selective SSMs into a simplified end-to-end neural network architecture without attention or even MLP blocks (Mamba). * Mamba enjoys fast inference (5x higher throughput than Transformers) and linear scaling in sequence length, and its performance improves on real data up to million-length sequences. As a general sequence model backbone, Mamba achieves state-of-the-art performance across several modalities such as language, audio, and genomics. On language modeling, our Mamba-1.4B model outperforms Transformers of the same size and matches Transformers twice its size, both in pretraining and downstream evaluation.* Get full access to Latent Space at www.latent.space/subscribe

Where did AI come from? Who created it, why, and where can it lead? Artificial intelligence (AI) is rapidly developing into a world-changer, affecting every industry and being used by hundreds of millions of people—even when they're unaware they're interacting with an artificial intelligence. And we're only at the early stages of AI's growth. Join us for an in-depth talk with Dr. Fei-Fei Li, whom Wired called "one of a tiny group of scientists―a group perhaps small enough to fit around a kitchen table―who are responsible for AI's recent remarkable advances.” Dr. Li came to America as an immigrant, enduring a shift from Chinese middle class to American poverty. But a tough upbringing did not stop her from becoming a leading mind in the next big technological development. Fei-Fei's adolescent knack for physics endured and positioned her to make a crucial contribution to the breakthrough we now call AI, placing her at the center of a global transformation. Over the last decades, her work has brought her face-to-face with the extraordinary possibilities―and the extraordinary dangers―of the technology she loves. Known as the creator of ImageNet, a key catalyst of modern artificial intelligence, Dr. Li has spent more than two decades at the forefront of the field.  Her work has brought her face-to-face with the extraordinary possibilities―and the extraordinary dangers―of the technology she loves. Don't miss this opportunity to learn more about a breakthrough science and one of the breakthrough scientists who is making it happen. This program is part of our Good Lit series, underwritten by the Bernard Osher Foundation. Learn more about your ad choices. Visit megaphone.fm/adchoices

Dr. Fei-Fei Li is a literal visionary. Her groundbreaking work on ImageNet, a vast visual recognition database, helped propel artificial intelligence at a critical moment. As one of the key innovators and thinkers in AI, Li has argued for a human-centered artificial intelligence that augments people's capabilities instead of displacing them. We talk to Li about her work, her vision for AI and her new memoir, The Worlds I See, in which she recounts her journey as a scientist and immigrant, and how those two roles inform each other. Guests: Fei-Fei Li, professor of Computer Science Department, Stanford University; author, "The Worlds I See: Curiosity, Exploration, and Discovery at the Dawn of AI"

A year ago, the public launch of ChatGPT took the world by storm and it was followed by many more generative artificial intelligence tools, all with remarkable, human-like abilities. Fears over the existential risks posed by AI have dominated the global conversation around the technology ever since. A pioneer that helped lay the groundwork that underpins generative AI models, Fei-Fei Li, takes a more nuanced approach to. She's pushing for a human-centred way of dealing with AI—treating it as a tool to help enhance—and not replace—humanity, while focussing on the pressing challenges of disinformation, bias and job disruption.Fei-Fei Li, a pioneer that helped lay the groundwork that underpins modern generative AI models, takes a more nuanced approach. She's pushing for a human-centred way of dealing with AI—treating it as a tool to help enhance—and not replace—humanity, while focussing on the pressing challenges of disinformation, bias and job disruption.Fei-Fei Li is the founding co-director of Stanford University's Institute for Human-Centred Artificial Intelligence. Fei-Fei and her research group created ImageNet, a huge database of images that enabled computers scientists to build algorithms that were able to see and recognise objects in the real world. That endeavour also introduced the world to deep learning, a type of machine learning that is fundamental part of how large-language and image-creation models work.Host: Alok Jha, The Economist's science and technology editor. Sign up for a free trial of Economist Podcasts+. If you're already a subscriber to The Economist, you'll have full access to all our shows as part of your subscription. For more information about how to access Economist Podcasts+, please visit our FAQs page or watch our video explaining how to link your account. Hosted on Acast. See acast.com/privacy for more information.

A year ago, the public launch of ChatGPT took the world by storm and it was followed by many more generative artificial intelligence tools, all with remarkable, human-like abilities. Fears over the existential risks posed by AI have dominated the global conversation around the technology ever since. Fei-Fei Li, a pioneer that helped lay the groundwork that underpins modern generative AI models, takes a more nuanced approach. She's pushing for a human-centred way of dealing with AI—treating it as a tool to help enhance—and not replace—humanity, while focussing on the pressing challenges of disinformation, bias and job disruption.Fei-Fei Li is the founding co-director of Stanford University's Institute for Human-Centred Artificial Intelligence. Fei-Fei and her research group created ImageNet, a huge database of images that enabled computers scientists to build algorithms that were able to see and recognise objects in the real world. That endeavour also introduced the world to deep learning, a type of machine learning that is fundamental part of how large-language and image-creation models work.Host: Alok Jha, The Economist's science and technology editor. Sign up for a free trial of Economist Podcasts+. If you're already a subscriber to The Economist, you'll have full access to all our shows as part of your subscription. For more information about how to access Economist Podcasts+, please visit our FAQs page or watch our video explaining how to link your account. Hosted on Acast. See acast.com/privacy for more information.

Summary Machine learning and generative AI systems have produced truly impressive capabilities. Unfortunately, many of these applications are not designed with the privacy of end-users in mind. TripleBlind is a platform focused on embedding privacy preserving techniques in the machine learning process to produce more user-friendly AI products. In this episode Gharib Gharibi explains how the current generation of applications can be susceptible to leaking user data and how to counteract those trends. Announcements Hello and welcome to the Machine Learning Podcast, the podcast about machine learning and how to bring it from idea to delivery. Your host is Tobias Macey and today I'm interviewing Gharib Gharibi about the challenges of bias and data privacy in generative AI models Interview Introduction How did you get involved in machine learning? Generative AI has been gaining a lot of attention and speculation about its impact. What are some of the risks that these capabilities pose? What are the main contributing factors to their existing shortcomings? What are some of the subtle ways that bias in the source data can manifest? In addition to inaccurate results, there is also a question of how user interactions might be re-purposed and potential impacts on data and personal privacy. What are the main sources of risk? With the massive attention that generative AI has created and the perspectives that are being shaped by it, how do you see that impacting the general perception of other implementations of AI/ML? How can ML practitioners improve and convey the trustworthiness of their models to end users? What are the risks for the industry if generative models fall out of favor with the public? How does your work at Tripleblind help to encourage a conscientious approach to AI? What are the most interesting, innovative, or unexpected ways that you have seen data privacy addressed in AI applications? What are the most interesting, unexpected, or challenging lessons that you have learned while working on privacy in AI? When is TripleBlind the wrong choice? What do you have planned for the future of TripleBlind? Contact Info LinkedIn (https://www.linkedin.com/in/ggharibi/) Parting Question From your perspective, what is the biggest barrier to adoption of machine learning today? Closing Announcements Thank you for listening! Don't forget to check out our other shows. The Data Engineering Podcast (https://www.dataengineeringpodcast.com) covers the latest on modern data management. Podcast.__init__ () covers the Python language, its community, and the innovative ways it is being used. Visit the site (https://www.themachinelearningpodcast.com) to subscribe to the show, sign up for the mailing list, and read the show notes. If you've learned something or tried out a project from the show then tell us about it! Email hosts@themachinelearningpodcast.com (mailto:hosts@themachinelearningpodcast.com)) with your story. To help other people find the show please leave a review on iTunes (https://podcasts.apple.com/us/podcast/the-machine-learning-podcast/id1626358243) and tell your friends and co-workers. Links TripleBlind (https://tripleblind.ai/) ImageNet (https://scholar.google.com/citations?view_op=view_citation&hl=en&user=JicYPdAAAAAJ&citation_for_view=JicYPdAAAAAJ:VN7nJs4JPk0C) Geoffrey Hinton Paper BERT (https://en.wikipedia.org/wiki/BERT_(language_model)) language model Generative AI (https://en.wikipedia.org/wiki/Generative_artificial_intelligence) GPT == Generative Pre-trained Transformer (https://en.wikipedia.org/wiki/Generative_pre-trained_transformer) HIPAA Safe Harbor Rules (https://www.hhs.gov/hipaa/for-professionals/privacy/special-topics/de-identification/index.html) Federated Learning (https://en.wikipedia.org/wiki/Federated_learning) Differential Privacy (https://en.wikipedia.org/wiki/Differential_privacy) Homomorphic Encryption (https://en.wikipedia.org/wiki/Homomorphic_encryption) The intro and outro music is from Hitman's Lovesong feat. Paola Graziano (https://freemusicarchive.org/music/The_Freak_Fandango_Orchestra/Tales_Of_A_Dead_Fish/Hitmans_Lovesong/) by The Freak Fandango Orchestra (http://freemusicarchive.org/music/The_Freak_Fandango_Orchestra/)/CC BY-SA 3.0 (https://creativecommons.org/licenses/by-sa/3.0/)

Fei-Fei Li, PhD,  Professor in the Computer Science Department at Stanford University, and Co-Director of Stanford's Human-Centered AI Institute, joins Bio + Health founding partner Vijay Pande.In this candid conversation, Li unfolds her transformation from a young immigrant to an influential figure in AI. The conversation explores the birth of ImageNet, a pivotal step that bridged the gap between visual intelligence and accessible AI technology. They delve into the notion of a 'Dignity Economy,' hinting at a future where technology serves to elevate human experience rather than undermine it. Li also touches on the delicate balance between relentless innovation and life's humble pursuits. This episode peels back the layers on the human side of AI, offering a rare glimpse into the personal and professional realms of a pioneer shaping the AI landscape.Check out her new book, The Worlds I See, here: https://us.macmillan.com/books/9781250897930/theworldsiseeCheck out other episodes form our sister podcast, Bio Eats World: https://a16z.com/podcasts/bio-eats-world/ Stay Updated: Find a16z on Twitter: https://twitter.com/a16zFind a16z on LinkedIn: https://www.linkedin.com/company/a16zSubscribe on your favorite podcast app: https://a16z.simplecast.com/Follow our host: https://twitter.com/stephsmithioPlease note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.

Fei-Fei Li's new book is the story of her journey from China to the U.S., from small business to Big Tech, and from academic research to corporate life, and back again. But more than that, it's the story of the dawn of artificial intelligence, as told through her experience as one of the people summoning this new day and standing there awestruck, excited and concerned about what it will mean for humanity. Dr. Li joins us on this episode to discuss the book, The Worlds I See: Curiosity, Exploration, and Discovery at the Dawn of AI, published by Moment of Lift Books, an imprint from Melinda French Gates and Flatiron Books. Known for her foundational contributions to AI and computer vision, Dr. Li is the inventor of ImageNet, a large-scale dataset of images that enabled rapid advances in deep learning for visual recognition. She is a professor of computer science at Stanford University and a co-director of the Stanford Institute for Human-Centered Artificial Intelligence, who worked as Google Cloud's chief scientist for AI/ML during a 2017-2018 sabbatical. Note: GeekWire's Todd Bishop will be speaking further with Dr. Li on Monday evening Nov. 13 at Town Hall in Seattle. See this site for details and tickets. Edited by Curt Milton.See omnystudio.com/listener for privacy information.

Fei-Fei Li, PhD,  Professor in the Computer Science Department at Stanford University, and Co-Director of Stanford's Human-Centered AI Institute, joins Bio + Health founding partner Vijay Pande.In this candid conversation, Li unfolds her transformation from a young immigrant to an influential figure in AI. The conversation explores the birth of ImageNet, a pivotal step that bridged the gap between visual intelligence and accessible AI technology. They delve into the notion of a 'Dignity Economy,' hinting at a future where technology serves to elevate human experience rather than undermine it. Li also touches on the delicate balance between relentless innovation and life's humble pursuits. This episode peels back the layers on the human side of AI, offering a rare glimpse into the personal and professional realms of a pioneer shaping the AI landscape.Check out her new book, out November 7, 2023, here: https://us.macmillan.com/books/9781250897930/theworldsisee

At the AI Pioneers Summit we announced Latent Space Launchpad, an AI-focused accelerator in partnership with Decibel. If you're an AI founder of enterprise early adopter, fill out this form and we'll be in touch with more details. We also have a lot of events coming up as we wrap up the year, so make sure to check out our community events page and come say hi!We previously interviewed the founders of many developer productivity startups embedded in the IDE, like Codium AI, Cursor, and Codeium. We also covered Replit's (former) SOTA model, replit-code-v1-3b and most recently had Amjad and Michele announce replit-code-v1_5-3b at the AI Engineer Summit.Much has been speculated about the StackOverflow traffic drop since ChatGPT release, but the experience is still not perfect. There's now a new player in the “search for developers” arena: Phind.Phind's goal is to help you find answers to your technical questions, and then help you implement them. For example “What should I use to create a frontend for a Python script?” returns a list of frameworks as well as links to the sources. You can then ask follow up questions on specific implementation details, having it write some code for you, etc. They have both a web version and a VS Code integrationThey recently were top of Hacker News with the announcement of their latest model, which is now the #1 rated model on the BigCode Leaderboard, beating their previous version:TLDR Cheat Sheet:* Based on CodeLlama-34B, which is trained on 500B tokens* Further fine-tuned on 70B+ high quality code and reasoning tokens* Expanded context window to 16k tokens* 5x faster than GPT-4 (100 tok/s vs 20 tok/s on single stream)* 74.7% HumanEval vs 45% for the base modelWe've talked before about HumanEval being limited in a lot of cases and how it needs to be complemented with “vibe based” evals. Phind thinks of evals alongside two axis: * Context quality: when asking the model to generate code, was the context high quality? Did we put outdated examples in it? Did we retrieve the wrong files?* Result quality: was the code generated correct? Did it follow the instructions I gave it or did it misunderstand some of it?If you have bad results with bad context, you might get to a good result by working on better RAG. If you have good context and bad result you might either need to work on your prompting or you have hit the limits of the model, which leads you to fine tuning (like they did). Michael was really early to this space and started working on CommonCrawl filtering and indexing back in 2020, which led to a lot of the insights that now power Phind. We talked about that evolution, his experience at YC, how he got Paul Graham to invest in Phind and invite him to dinner at his house, and how Ron Conway connected him with Jensen Huang to get access to more GPUs!Show Notes* Phind* BigScience T0* InstructGPT Paper* Inception-V3* LMQL* Marginalia Nu* Mistral AI* People:* Paul Graham (pg)* Ron Conway* Yacine Jernite from HuggingFace* Jeff DelaneyTimestamps* [00:00:00] Intros & Michael's early interest in computer vision* [00:03:14] Pivoting to NLP and natural language question answering models* [00:07:20] Building a search engine index of Common Crawl and web pages* [00:11:26] Releasing the first version of Hello based on the search index and BigScience T0 model* [00:14:02] Deciding to focus the search engine specifically for programmers* [00:17:39] Overview of Phind's current product and focus on code reasoning* [00:21:51] The future vision for Phind to go from idea to complete code* [00:24:03] Transitioning to using the GPT-4 model and the impact it had* [00:29:43] Developing the Phind model based on CodeLlama and additional training* [00:32:28] Plans to continue improving the Phind model with open source technologies* [00:43:59] The story of meeting Paul Graham and Ron Conway and how that impacted the company* [00:53:02] How Ron Conway helped them get GPUs from Nvidia* [00:57:12] Tips on how Michael learns complex AI topics* [01:01:12] Lightning RoundTranscriptAlessio: Hey everyone, welcome to the Latent Space Podcast. This is Alessio, partner and CTO of Residence and Decibel Partners, and I'm joined by my co-host Swyx, founder of Smol AI. [00:00:19]Swyx: Hey, and today we have in the studio Michael Royzen from Phind. Welcome. [00:00:23]Michael: Thank you so much. [00:00:24]Alessio: It's great to be here. [00:00:25]Swyx: Yeah, we are recording this in a surprisingly hot October in San Francisco. And sometimes the studio works, but the blue angels are flying by right now, so sorry about the noise. So welcome. I've seen Phind blow up this year, mostly, I think since your launch in Feb and V2 and then your Hacker News posts. We tend to like to introduce our guests, but then obviously you can fill in the blanks with the origin story. You actually were a high school entrepreneur. You started SmartLens, which is a computer vision startup in 2017. [00:00:59]Michael: That's right. I remember when like TensorFlow came out and people started talking about, obviously at the time after AlexNet, the deep learning revolution was already in flow. Good computer vision models were a thing. And what really made me interested in deep learning was I got invited to go to Apple's WWDC conference as a student scholar because I was really into making iOS apps at the time. So I go there and I go to this talk where they added an API that let people run computer vision models on the device using far more efficient GPU primitives. After seeing that, I was like, oh, this is cool. This is going to have a big explosion of different computer vision models running locally on the iPhone. And so I had this crazy idea where it was like, what if I could just make this model that could recognize just about anything and have it run on the device? And that was the genesis for what eventually became SmartLens. I took this data set called ImageNet 22K. So most people, when they think of ImageNet, think of ImageNet 1K. But the full ImageNet actually has, I think, 22,000 different categories. So I took that, filtered it, pre-processed it, and then did a massive fine tune on Inception V3, which was, I think, the state of the art deep convolutional computer vision model at the time. And to my surprise, it actually worked insanely well. I had no idea what would happen if I give a single model. I think it ended up being 17,000 categories approximately that I collapsed them into. It worked so well that it actually worked better than Google Lens, which released its V1 around the same time. And on top of this, the model ran on the device. So it didn't need an internet connection. A big part of the issue with Google Lens at the time was that connections were slower. 4G was around, but it wasn't nearly as fast. So there was a noticeable lag having to upload an image to a server and get it back. But just processing it locally, even on the iPhones of the day in 2017, much faster. It was a cool little project. It got some traction. TechCrunch wrote about it. There was kind of like one big spike in usage, and then over time it tapered off. But people still pay for it, which is wild. [00:03:14]Swyx: That's awesome. Oh, it's like a monthly or annual subscription? [00:03:16]Michael: Yeah, it's like a monthly subscription. [00:03:18]Swyx: Even though you don't actually have any servers? [00:03:19]Michael: Even though we don't have any servers. That's right. I was in high school. I had a little bit of money. I was like, yeah. [00:03:25]Swyx: That's awesome. I always wonder what the modern equivalents kind of "Be my eyes". And it would be actually disclosed in the GPT-4 Vision system card recently that the usage was surprisingly not that frequent. The extent to which all three of us have our sense of sight. I would think that if I lost my sense of sight, I would use Be My Eyes all the time. The average usage of Be My Eyes per day is 1.5 times. [00:03:49]Michael: Exactly. I was thinking about this as well, where I was also looking into image captioning, where you give a model an image and then it tells you what's in the image. But it turns out that what people want is the exact opposite. People want to give a description of an image and then have the AI generate the image. [00:04:04]Alessio: Oh, the other way. [00:04:06]Michael: Exactly. And so at the time, I think there were some GANs, NVIDIA was working on this back in 2019, 2020. They had some impressive, I think, face GANs where they had this model that would produce these really high quality portraits, but it wasn't able to take a natural language description the way Midjourney or DALL-E 3 can and just generate you an image with exactly what you described in it. [00:04:32]Swyx: And how did that get into NLP? [00:04:35]Michael: Yeah, I released the SmartLens app and that was around the time I was a senior in high school. I was applying to college. College rolls around. I'm still sort of working on updating the app in college. But I start thinking like, hey, what if I make an enterprise version of this as well? At the time, there was Clarify that provided some computer vision APIs, but I thought this massive classification model works so well and it's so small and so fast, might as well build an enterprise product. And I didn't even talk to users or do any of those things that you're supposed to do. I was just mainly interested in building a type of backend I've never built before. So I was mainly just doing it for myself just to learn. I built this enterprise classification product and as part of it, I'm also building an invoice processing product where using some of the aspects that I built previously, although obviously it's very different from classification, I wanted to be able to just extract a bunch of structured data from an unstructured invoice through our API. And that's what led me to Hugnyface for the first time because that involves some natural language components. And so I go to Hugnyface and with various encoder models that were around at the time, I used the standard BERT and also Longformer, which came out around the same time. And Longformer was interesting because it had a much bigger context window than those models at the time, like BERT, all of the first gen encoder only models, they only had a context window of 512 tokens and it's fixed. There's none of this alibi or ROPE that we have now where we can basically massage it to be longer. They're fixed, 512 absolute encodings. Longformer at the time was the only way that you can fit, say, like a sequence length or ask a question about like 4,000 tokens worth of text. Implemented Longformer, it worked super well, but like nobody really kind of used the enterprise product and that's kind of what I expected because at the end of the day, it was COVID. I was building this kind of mostly for me, mostly just kind of to learn. And so nobody really used it and my heart wasn't in it and I kind of just shelved it. But a little later, I went back to HugMeFace and I saw this demo that they had, and this is in the summer of 2020. They had this demo made by this researcher, Yacine Jernite, and he called it long form question answering. And basically, it was this self-contained notebook demo where you can ask a question the way that we do now with ChatGPT. It would do a lookup into some database and it would give you an answer. And it absolutely blew my mind. The demo itself, it used, I think, BART as the model and in the notebook, it had support for both an Elasticsearch index of Wikipedia, as well as a dense index powered by Facebook's FAISS. I think that's how you pronounce it. It was very iffy, but when it worked, I think the question in the demo was, why are all boats white? When it worked, it blew my mind that instead of doing this few shot thing, like people were doing with GPT-3 at the time, which is all the rage, you could just ask a model a question, provide no extra context, and it would know what to do and just give you the answer. It blew my mind to such an extent that I couldn't stop thinking about that. When I started thinking about ways to make it better, I tried training, doing the fine tune with a larger BART model. And this BART model, yeah, it was fine tuned on this Reddit data set called Eli5. So basically... [00:08:02]Alessio: Subreddit. [00:08:03]Swyx: Yeah, subreddit. [00:08:04]Alessio: Yeah. [00:08:05]Michael: And put it into like a well-formatted, relatively clean data set of like human questions and human answers. And that was a really great bootstrap for that model to be able to answer these types of questions. And so Eli5 actually turned out to be a good data set for training these types of question answering models, because the question is written by a human, the answer is written by a human, and at least helps the model get the format right, even if the model is still very small and it can't really think super well, at least it gets the format right. And so it ends up acting as kind of a glorified summarization model, where if it's fed in high quality context from the retrieval system, it's able to have a reasonably high quality output. And so once I made the model as big as I can, just fine tuning on BART large, I started looking for ways to improve the index. So in the demo, in the notebook, there were instructions for how to make an Elasticsearch index just for Wikipedia. And I was like, why not do all of Common Crawl? So I downloaded Common Crawl, and thankfully, I had like 10 or $15,000 worth of AWS credits left over from the SmartLens project. And that's what really allowed me to do this, because there's no other funding. I was still in college, not a lot of money, and so I was able to spin up a bunch of instances and just process all of Common Crawl, which is massive. So it's roughly like, it's terabytes of text. I went to Alexa to get the top 1,000 websites or 10,000 websites in the world, then filtered only by those websites, and then indexed those websites, because the web pages were already included in Dump. [00:09:38]Swyx: You mean to supplement Common Crawl or to filter Common Crawl? [00:09:41]Michael: Filter Common Crawl. [00:09:42]Alessio: Oh, okay. [00:09:43]Michael: Yeah, sorry. So we filtered Common Crawl just by the top, I think, 10,000, just to limit this, because obviously there's this massive long tail of small sites that are really cool, actually. There's other projects like, shout out to Marginalia Nu, which is a search engine specialized on the long tail. I think they actually exclude the top 10,000. [00:10:03]Swyx: That's what they do. [00:10:04]Alessio: Yeah. [00:10:05]Swyx: I've seen them around, I just don't really know what their pitch is. Okay, that makes sense. [00:10:08]Michael: So they exclude all the top stuff. So the long tail is cool, but for this, that was kind of out of the question, and that was most of the data anyway. So we've removed that. And then I indexed the remaining approximately 350 million webpages through Elasticsearch. So I built this index running on AWS with these webpages, and it actually worked quite well. You can ask it general common knowledge, history, politics, current events, questions, and it would be able to do a fast lookup in the index, feed it into the model, and it would give a surprisingly good result. And so when I saw that, I thought that this is definitely doable. And it kind of shocked me that no one else was doing this. And so this was now the fall of 2020. And yeah, I was kind of shocked no one was doing this, but it costs a lot of money to keep it up. I was still in college. There are things going on. I got bogged down by classes. And so I ended up shelving this for almost a full year, actually. When I returned to it in fall of 2021, when BigScience released T0, when BigScience released the T0 models, that was a massive jump in the reasoning ability of the model. And it was better at reasoning, it was better at summarization, it was still a glorified summarizer basically. [00:11:26]Swyx: Was this a precursor to Bloom? Because Bloom's the one that I know. [00:11:29]Alessio: Yeah. [00:11:30]Michael: Actually coming out in 2022. But Bloom had other problems where for whatever reason, the Bloom models just were never really that good, which is so sad because I really wanted to use them. But I think they didn't turn on that much data. I think they used like the original, they were trying to replicate GPT-3. So they just use those numbers, which we now know are like far below Chinchilla Optimal and even Chinchilla Optimal, which we can like talk about later, like what we're currently doing with MIMO goes, yeah, it goes way beyond that. But they weren't trying enough data. I'm not sure how that data was clean, but it probably wasn't super clean. And then they didn't really do any fine tuning until much later. So T0 worked well because they took the T5 models, which were closer to Chinchilla Optimal because I think they were trained on also like 300 something billion tokens, similar to GPT-3, but the models were much smaller. I think T0 is the first model that did large scale instruction tuning from diverse data sources in the fall of 2021. This is before Instruct GPT. This is before Flan T5, which came out in 2022. This is the very, very first, at least well-known example of that. And so it came out and then I did, on top of T0, I also did the Reddit Eli5 fine tune. And that was the first model and system that actually worked well enough to where I didn't get discouraged like I did previously, because the failure cases of the BART based system was so egregious. Sometimes it would just miss a question so horribly that it was just extremely discouraging. But for the first time, it was working reasonably well. Also using a much bigger model. I think the BART model is like 800 million parameters, but T0, we were using 3B. So it was T0, 3B, bigger model. And that was the very first iteration of Hello. So I ended up doing a show HN on Hacker News in January 2022 of that system. Our fine tune T0 model connected to our Elasticsearch index of those 350 million top 10,000 common crawl websites. And to the best of my knowledge, I think that's the first example that I'm aware of a LLM search engine model that's effectively connected to like a large enough index that I consider like an internet scale. So I think we were the first to release like an internet scale LLM powered rag search system In January 2022, around the time me and my future co-founder, Justin, we were like, this seems like the future. [00:14:02]Alessio: This is really cool. [00:14:03]Michael: I couldn't really sleep even like I was going to bed and I was like, I was thinking about it. Like I would say up until like 2.30 AM, like reading papers on my phone in bed, go to sleep, wake up the next morning at like eight and just be super excited to keep working. And I was also doing my thesis at the same time, my senior honors thesis at UT Austin about something very similar. We were researching factuality in abstractive question answering systems. So a lot of overlap with this project and the conclusions of my research actually kind of helped guide the development path of Hello. In the research, we found that LLMs, they don't know what they don't know. So the conclusion was, is that you always have to do a search to ensure that the model actually knows what it's talking about. And my favorite example of this even today is kind of with chat GPT browsing, where you can ask chat GPT browsing, how do I run llama.cpp? And chat GPT browsing will think that llama.cpp is some file on your computer that you can just compile with GCC and you're all good. It won't even bother doing a lookup, even though I'm sure somewhere in their internal prompts they have something like, if you're not sure, do a lookup. [00:15:13]Alessio: That's not good enough. So models don't know what they don't know. [00:15:15]Michael: You always have to do a search. And so we approached LLM powered question answering from the search angle. We pivoted to make this for programmers in June of 2022, around the time that we were getting into YC. We realized that what we're really interested in is the case where the models actually have to think. Because up until then, the models were kind of more glorified summarization models. We really thought of them like the Google featured snippets, but on steroids. And so we saw a future where the simpler questions would get commoditized. And I still think that's going to happen with like Google SGE and like it's nowadays, it's really not that hard to answer the more basic kind of like summarization, like current events questions with lightweight models that'll only continue to get cheaper over time. And so we kind of started thinking about this trade off where LLM models are going to get both better and cheaper over time. And that's going to force people who run them to make a choice. Either you can run a model of the same intelligence that you could previously for cheaper, or you can run a better model for the same price. So someone like Google, once the price kind of falls low enough, they're going to deploy and they're already doing this with SGE, they're going to deploy a relatively basic glorified summarizer model that can answer very basic questions about like current events, who won the Super Bowl, like, you know, what's going on on Capitol Hill, like those types of things. The flip side of that is like more complex questions where like you have to reason and you have to solve problems and like debug code. And we realized like we're much more interested in kind of going along the bleeding edge of that frontier case. And so we've optimized everything that we do for that. And that's a big reason of why we've built Phind specifically for programmers, as opposed to saying like, you know, we're kind of a search engine for everyone because as these models get more capable, we're very interested in seeing kind of what the emergent properties are in terms of reasoning, in terms of being able to solve complex multi-step problems. And I think that some of those emerging capabilities like we're starting to see, but we don't even fully understand. So I think there's always an opportunity for us to become more general if we wanted, but we've been along this path of like, what is the best, most advanced reasoning engine that's connected to your code base, that's connected to the internet that we can just provide. [00:17:39]Alessio: What is Phind today, pragmatically, from a product perspective, how do people interact with it? Yeah. Or does it plug into your workflow? [00:17:46]Michael: Yeah. [00:17:47]Alessio: So Phind is really a system. [00:17:48]Michael: Phind is a system for programmers when they have a question or when they're frustrated or when something's not working. [00:17:54]Swyx: When they're frustrated. [00:17:55]Alessio: Yeah. [00:17:56]Michael: For them to get on block. I think like the single, the most abstract page for Phind is like, if you're experiencing really any kind of issue as a programmer, we'll solve that issue for you in 15 seconds as opposed to 15 minutes or longer. Phind has an interface on the web. It has an interface in VS code and more IDEs to come, but ultimately it's just a system where a developer can paste in a question or paste in code that's not working and Phind will do a search on the internet or they will find other code in your code base perhaps that's relevant. And then we'll find the context that it needs to answer your question and then feed it to a reasoning engine powerful enough to actually answer it. So that's really the philosophy behind Phind. It's a system for getting developers the answers that they're looking for. And so right now from a product perspective, this means that we're really all about getting the right context. So the VS code extension that we launched recently is a big part of this because you can just ask a question and it knows where to find the right code context in your code. It can do an internet search as well. So it's up to date and it's not just reliant on what the model knows and it's able to figure out what it needs by itself and answer your question based on that. If it needs some help, you can also get yourself kind of just, there's opportunities for you yourself to put in all that context in. But the issue is also like not everyone wants these VS code. Some people like are real Neovim sticklers or they're using like PyCharm or other IDEs, JetBrains. And so for those people, they're actually like okay with switching tabs, at least for now, if it means them getting their answer. Because really like there's been an explosion of all these like startups doing code, doing search, etc. But really who everyone's competing with is ChatGPT, which only has like that one web interface. Like ChatGPT is really the bar. And so that's what we're up against. [00:19:50]Alessio: And so your idea, you know, we have Amman from Cursor on the podcast and they've gone through the we need to own the IDE thing. Yours is more like in order to get the right answer, people are happy to like go somewhere else basically. They're happy to get out of their IDE. [00:20:05]Michael: That was a great podcast, by the way. But yeah, so part of it is that people sometimes perhaps aren't even in an IDE. So like the whole task of software engineering goes way beyond just running code, right? There's also like a design stage. There's a planning stage. A lot of this happens like on whiteboards. It happens in notebooks. And so the web part also exists for that where you're not even coding it and you're just trying to get like a more conceptual understanding of what you're trying to build first. The podcast with Amman was great, but somewhere where I disagree with him is that you need to own the IDE. I think like he made some good points about not having platform risk in the long term. But some of the features that were mentioned like suggesting diffs, for example, those are all doable with an extension. We haven't yet seen with VS Code in particular any functionality that we'd like to do yet in the IDE that we can't either do through directly supported VS Code functionality or something that we kind of hack into there, which we've also done a fair bit of. And so I think it remains to be seen where that goes. But I think what we're looking to be is like we're not trying to just be in an IDE or be an IDE. Like Phind is a system that goes beyond the IDE and like is really meant to cover the entire lifecycle of a developer's thought process in going about like, hey, like I have this idea and I want to get from that idea to a working product. And so then that's what the long term vision of Phind is really about is starting with that. In the future, I think programming is just going to be really just the problem solving. Like you come up with an idea, you come up with like the basic design for the algorithm in your head, and you just tell the AI, hey, just like just do it, just make it work. And that's what we're building towards. [00:21:51]Swyx: I think we might want to give people an impression about like type of traffic that you have, because when you present it with a text box, you could type in anything. And I don't know if you have some mental categorization of like what are like the top three use cases that people tend to coalesce around. [00:22:08]Alessio: Yeah, that's a great question. [00:22:09]Michael: The two main types of searches that we see are how-to questions, like how to do X using Y tool. And this historically has been our bread and butter, because with our embeddings, like we're really, really good at just going over a bunch of developer documentation and figuring out exactly the part that's relevant and just telling you, OK, like you can use this method. But as LLMs have gotten better, and as we've really transitioned to using GPT-4 a lot in our product, people organically just started pasting in code that's not working and just said, fix it for me. [00:22:42]Swyx: Fix this. [00:22:43]Alessio: Yeah. [00:22:44]Michael: And what really shocks us is that a lot of the people who do that, they're coming from chat GPT. So they tried it in chat GPT with chat GPT-4. It didn't work. Maybe it required like some multi-step reasoning. Maybe it required some internet context or something found in either a Stack Overflow post or some documentation to solve it. And so then they paste it into find and then find works. So those are really those two different cases. Like, how can I build this conceptually or like remind me of this one detail that I need to build this thing? Or just like, here's this code. Fix it. And so that's what a big part of our VS Code extension is, is like enabling a much smoother here just like fix it for me type of workflow. That's really its main benefits. Like it's in your code base. It's in the IDE. It knows how to find the relevant context to answer that question. But at the end of the day, like I said previously, that's still a relatively, not to say it's a small part, but it's a limited part of the entire mental life cycle of a programmer. [00:23:47]Swyx: Yep. So you launched in Feb and then you launched V2 in August. You had a couple other pretty impactful posts slash feature launches. The web search one was massive. So you were mostly a GPT-4 wrapper. We were for a long time. [00:24:03]Michael: For a long time until recently. Yeah. [00:24:05]Alessio: Until recently. [00:24:06]Swyx: So like people coming over from ChatGPT were saying, we're going to say model with your version of web search. Would that be the primary value proposition? [00:24:13]Michael: Basically yeah. And so what we've seen is that any model plus web search is just significantly better than [00:24:18]Alessio: that model itself. Do you think that's what you got right in April? [00:24:21]Swyx: Like so you got 1500 points on Hacking News in April, which is like, if you live on Hacking News a lot, that is unheard of for someone so early on in your journey. [00:24:31]Alessio: Yeah. [00:24:32]Michael: We're super, super grateful for that. Definitely was not expecting it. So what we've done with Hacker News is we've just kept launching. [00:24:38]Alessio: Yeah. [00:24:39]Michael: Like what they don't tell you is that you can just keep launching. That's what we've been doing. So we launched the very first version of Find in its current incarnation after like the previous demo connected to our own index. Like once we got into YC, we scrapped our own index because it was too cumbersome at the time. So we moved over to using Bing as kind of just the raw source data. We launched as Hello Cognition. Over time, every time we like added some intelligence to the product, a better model, we just keep launching. And every additional time we launched, we got way more traffic. So we actually silently rebranded to Find in late December of last year. But like we didn't have that much traffic. Nobody really knew who we were. [00:25:18]Swyx: How'd you pick the name out of it? [00:25:19]Michael: Paul Graham actually picked it for us. [00:25:21]Swyx: All right. [00:25:22]Alessio: Tell the story. Yeah. So, oh boy. [00:25:25]Michael: So this is the biggest side. Should we go for like the full Paul Graham story or just the name? [00:25:29]Swyx: Do you want to do it now? Or do you want to do it later? I'll give you a choice. [00:25:32]Alessio: Hmm. [00:25:33]Michael: I think, okay, let's just start with the name for now and then we can do the full Paul Graham story later. But basically, Paul Graham, when we were lucky enough to meet him, he saw our name and our domain was at the time, sayhello.so and he's just like, guys, like, come on, like, what is this? You know? And we were like, yeah, but like when we bought it, you know, we just kind of broke college students. Like we didn't have that much money. And like, we really liked hello as a name because it was the first like conversational search engine. And that's kind of, that's the angle that we were approaching it from. And so we had sayhello.so and he's like, there's so many problems with that. Like, like, like the say hello, like, what does that even mean? And like .so, like, it's gotta be like a .com. And so we did some time just like with Paul Graham in the room. We just like looked at different domain names, like different things that like popped into our head. And one of the things that popped into like Paul Graham said was fine with the Phind spelling in particular. [00:26:33]Swyx: Yeah. Which is not typical naming advice, right? Yes. Because it's not when people hear it, they don't spell it that way. [00:26:38]Michael: Exactly. It's hard to spell. And also it's like very 90s. And so at first, like, we didn't like, I was like, like, ah, like, I don't know. But over time it kept growing on us. And eventually we're like, okay, we like the name. It's owned by this elderly Canadian gentleman who we got to know, and he was willing to sell it to us. [00:26:57]Michael: And so we bought it and we changed the name. Yeah. [00:27:01]Swyx: Anyways, where were you? [00:27:02]Alessio: I had to ask. [00:27:03]Swyx: I mean, you know, everyone who looks at you is wondering. [00:27:06]Michael: And a lot of people actually pronounce it Phind, which, you know, by now it's part of the game. But eventually we want to buy Phind.com and then just have that redirect to Phind. So Phind is like definitely the right spelling. But like, we'll just, yeah, we'll have all the cases addressed. [00:27:23]Swyx: Cool. So Bing web search, and then August you launched V2. Is V2 the Phind as a system pitch? Or have you moved, evolved since then? [00:27:31]Michael: Yeah, so I don't, like the V2 moniker, like, I don't really think of it that way in my mind. There's like, there's the version we launched during, last summer during YC, which was the Bing version directed towards programmers. And that's kind of like, that's why I call it like the first incarnation of what we currently are. Because it was already directed towards programmers. We had like a code snippet search built in as well, because at the time, you know, the models we were using weren't good enough to generate code snippets. Even GPT, like the text DaVinci 2 was available at the time, wasn't that good at generating code and it would generate like very, very short, very incomplete code snippets. And so we launched that last summer, got some traction, but really like we were only doing like, I don't know, maybe like 10,000 searches a day. [00:28:15]Alessio: Some people knew about it. [00:28:16]Michael: Some people used it, which is impressive because looking back, the product like was not that good. And every time we've like made an improvement to the way that we retrieve context through better embeddings, more intelligent, like HTML parsers, and importantly, like better underlying models. Every major version after that was when we introduced a better underlying answering model. Like in February, we had to swallow a bit of our pride when we were like, okay, our own models aren't good enough. We have to go to open AI. And actually that did lead to kind of like our first decent bump of traffic in February. And people kept using it, like our attention was way better too. But we were still kind of running into problems of like more advanced reasoning. Some people tried it, but people were leaving because even like GPT 3.5, both turbo and non-turbo, like still not that great at doing like code related reasoning beyond the how do you do X, like documentation search type of use case. And so it was really only when GPT 4 came around in April that we were like, okay, like this is like our first real opportunity to really make this thing like the way that it should have been all along. And having GPT 4 as the brain is what led to that Hacker News post. And so what we did was we just let anyone use GPT 4 on Fyne for free without a login, [00:29:43]Alessio: which I actually don't regret. [00:29:45]Michael: So it was very expensive, obviously. But like at that stage, all we needed to do was show like, we just needed to like show people here's what Fyne can do. That was the main thing. And so that worked. That worked. [00:29:58]Alessio: Like we got a lot of users. [00:29:59]Michael: Do you know Fireship? [00:30:01]Swyx: Yeah. YouTube, Jeff Delaney. [00:30:03]Michael: Yeah. He made a short about Fyne. [00:30:06]Alessio: Oh. [00:30:07]Michael: And that's on top of the Hacker News post. And that's what like really, really made it blow up. It got millions of views in days. And he's just funny. Like what I love about Fireship is like he like you guys, yeah, like humor goes a long a long way towards like really grabbing people's attention. And so that blew up. [00:30:25]Swyx: Something I would be anxious about as a founder during that period, so obviously we all remember that pretty closely. So there were a couple of people who had access to the GPT-4 API doing this, which is unrestricted access to GPT-4. And I have to imagine OpenAI wasn't that happy about that because it was like kind of de facto access to GPT-4 before they released it. [00:30:46]Alessio: No, no. [00:30:47]Michael: GPT-4 was in chat GPT from day one. I think. OpenAI actually came to our support because what happened was we had people building unofficial APIs around to try to get free access to it. And I think OpenAI actually has the right perspective on this where they're like, OK, people can do whatever they want with the API if they're paying for it, like they can do whatever they want, but it's like not OK if, you know, paying customers are being exploite by these other actors. They actually got in touch with us and they helped us like set up better Cloudflare bot monitoring controls to effectively like crack down on those unofficial APIs, which we're very happy about. But yeah, so we launched GPT-4. A lot of people come to the product and yeah, for a long time, we're just we're figuring out like what do we make of this, right? How do we a make it better, but also deal with like our costs, which have just like massively, massively ballooned. Over time, it's become more clear with the release of Llama 2 and Llama 3 on the horizon that we will once again see a return to vertical applications running their own models. As was true last year and before, I think that GPT-4, my hypothesis is that the jump from 4 to 4.5 or 4 to 5 will be smaller than the jump from 3 to 4. And the reason why is because there were a lot of different things. Like there was two plus, effectively two, two and a half years of research that went into going from 3 to 4. Like more data, bigger model, all of the instruction tuning techniques, RLHF, all of that is known. And like Meta, for example, and now there's all these other startups like Mistral too, like there's a bunch of very well-funded open source players that are now working on just like taking the recipe that's now known and scaling it up. So I think that even if a delta exists, the delta between in 2024, the delta between proprietary and open source won't be large enough that a startup like us with a lot of data that we've collected can take the data that we have, fine tune an open source model, and like be able to have it be better than whatever the proprietary model is at the time. That's my hypothesis.Michael: But we'll once again see a return to these verticalized models. And that's something that we're super excited about because, yeah, that brings us to kind of the fine model because the plan from kind of the start was to be able to return to that if that makes sense. And I think now we're definitely at a point where it does make sense because we have requests from users who like, they want longer context in the model, basically, like they want to be able to ask questions about their entire code base without, you know, context and retrieval and taking a chance of that. Like, I think it's generally been shown that if you have the space to just put the raw files inside of a big context window, that is still better than chunking and retrieval. So there's various things that we could do with longer context, faster speed, lower cost. Super excited about that. And that's the direction that we're going with the fine model. And our big hypothesis there is precisely that we can take a really good open source model and then just train it on absolutely all of the high quality data that we can find. And there's a lot of various, you know, interesting ideas for this. We have our own techniques that we're kind of playing with internally. One of the very interesting ideas that I've seen, I think it's called Octopack from BigCode. I don't think that it made that big waves when it came out, I think in August. But the idea is that they have this data set that maps GitHub commits to a change. So basically there's all this really high quality, like human made, human written diff data out there on every time someone makes a commit in some repo. And you can use that to train models. Take the file state before and like given a commit message, what should that code look like in the future? [00:34:52]Swyx: Got it. [00:34:53]Alessio: Do you think your HumanEval is any good?Michael: So we ran this experiment. We trained the Phind model. And if you go to the BigCode leaderboard, as of today, October 5th, all of our models are at the top of the BigCode leaderboard by far. It's not close, particularly in languages other than Python. We have a 10 point gap between us and the next best model on JavaScript. I think C sharp, multilingual. And what we kind of learned from that whole experience releasing those models is that human eval doesn't really matter. Not just that, but GPT-4 itself has been trained on human eval. And we know this because GPT-4 is able to predict the exact docstring in many of the problems. I've seen it predict like the specific example values in the docstring, which is extremely improbable. So I think there's a lot of dataset contamination and it only captures a very limited subset of what programmers are actually doing. What we do internally for evaluations are we have GPT-4 score answers. GPT-4 is a really good evaluator. I mean, obviously it's by really good, I mean, it's the best that we have. I'm sure that, you know, a couple of months from now, next year, we'll be like, oh, you know, like GPT-4.5, GPT-5, it's so much better. Like GPT-4 is terrible, but like right now it's the best that we have short of humans. And what we found is that when doing like temperature zero evals, it's actually mostly deterministic GPT-4 across runs in assigning scores to two different answers. So we found it to be a very useful tool in comparing our model to say, GPT-4, but yeah, on our like internal real world, here's what people will be asking this model dataset. And the other thing that we're running is just like releasing the model to our users and just seeing what they think. Because that's like the only thing that really matters is like releasing it for the application that it's intended for, and then seeing how people react. And for the most part, the incredible thing is, is that people don't notice a difference between our model and GPT-4 for the vast majority of searches. There's some reasoning problems that GPT-4 can still do better. We're working on addressing that. But in terms of like the types of questions that people are asking on find, there's not that much difference. And in fact, I've been running my own kind of side by side comparisons, shout out to GodMode, by the way. [00:37:16]Michael: And I've like myself, I've kind of confirmed this to be the case. And even sometimes it gives a better answer, perhaps like more concise or just like better implementation than GPT-4, which that's what surprises me. And by now we kind of have like this reasoning is all you need kind of hypothesis where we've seen emerging capabilities in the find model, whereby training it on high quality code, it can actually like reason better. It went from not being able to solve world problems, where riddles were like with like temporal placement of objects and moving and stuff like that, that GPT-4 can do pretty well. We went from not being able to do those at all to being able to do them just by training on more code, which is wild. So we're already like starting to see like these emerging capabilities. [00:37:59]Swyx: So I just wanted to make sure that we have the, I guess, like the model card in our heads. So you started from Code Llama? [00:38:07]Alessio: Yes. [00:38:08]Swyx: 65, 34? 34. [00:38:10]Michael: So unfortunately, there's no Code Llama 70b. If there was, that would be super cool. But there's not. [00:38:15]Swyx: 34. And then, which in itself was Llama 2, which is on 2 trillion tokens and the added 500 billion code tokens. Yes. [00:38:22]Michael: And you just added a bunch more. [00:38:23]Alessio: Yeah. [00:38:24]Michael: And they also did a couple of things. So they did, I think they did 500 billion, like general pre-training and then they did an extra 20 billion long context pre-training. So they actually increased the like max position tokens to 16k up from 8k. And then they changed the theta parameter for the ROPE embeddings as well to give it theoretically better long context support up to 100k tokens. But yeah, but otherwise it's like basically Llama 2. [00:38:50]Swyx: And so you just took that and just added data. [00:38:52]Michael: Exactly. [00:38:53]Swyx: You didn't do any other fundamental. [00:38:54]Michael: Yeah. So we didn't actually, we haven't yet done anything with the model architecture and we just trained it on like many, many more billions of tokens on our own infrastructure. And something else that we're taking a look at now is using reinforcement learning for correctness. One of the interesting pitfalls that we've noticed with the Phind model is that in cases where it gets stuff wrong, it sometimes is capable of getting the right answer. It's just, there's a big variance problem. It's wildly inconsistent. There are cases when it is able to get the right chain of thought and able to arrive [00:39:25]Alessio: at the right answer, but not always. [00:39:27]Michael: And so like one of our hypotheses is something that we're going to try is that like we can actually do reinforcement learning on, for a given problem, generate a bunch of completions and then like use the correct answer as like a loss basically to try to get it to be more correct. And I think there's a high chance I think of this working because it's very similar to the like RLHF method where you basically show pairs of completions for a given question except the criteria is like which one is like less harmful. But here we have a different criteria. But if the model is already capable of getting the right answer, which it is, we're just, we just need to cajole it into being more consistent. [00:40:06]Alessio: There were a couple of things that I noticed in the product that were not strange but unique. So first of all, the model can talk multiple times in a row, like most other applications is like human model, human model. And then you had outside of the thumbs up, thumbs down, you have things like have DLLM prioritize this message and its answers or then continue from this message to like go back. How does that change the flow of the user and like in terms of like prompting it, yeah, what are like some tricks or learnings you've had? [00:40:37]Michael: So yeah, that's specifically in our pair programmer mode, which is a more conversational mode that also like asks you clarifying questions back if it doesn't fully understand what you're doing and it kind of it holds your hand a bit more. And so from user feedback, we had requests to make more of an auto GPT where you can kind of give it this problem that might take multiple searches or multiple different steps like multiple reasoning steps to solve. And so that's the impetus behind building that product. Being able to do multiple steps and also be able to handle really long conversations. Like people are really trying to use the pair programmer to go from like sometimes really from like basic idea to like complete working code. And so we noticed was is that we were having like these very, very long threads, sometimes with like 60 messages, like 100 messages. And like those become really, really challenging to manage the appropriate context window of what should go inside of the context and how to preserve the context so that the model can continue or the product can continue giving good responses, even if you're like 60 messages deep in a conversation. So that's where the prioritized user messages like comes from. It's like people have asked us to just like let them pin messages that they want to be left in the conversation. And yeah, and then that seems to have like really gone a long way towards solving that problem, yeah. [00:41:54]Alessio: And then you have a run on Replit thing. Are you planning to build your own repl? Like learning some people trying to run the wrong code, unsafe code? [00:42:03]Michael: Yes. Yes. So I think like in the long term vision of like being a place where people can go from like idea to like fully working code, having a code sandbox, like a natively integrated code sandbox makes a lot of sense. And replit is great and people use that feature. But yeah, I think there's more we can do in terms of like having something a bit closer to code interpreter where it's able to run the code and then like recursively iterate on it. Exactly. [00:42:31]Swyx: So you're working on APIs to enable you to do that? Yep. So Amjad has specifically told me in person that he wants to enable that for people at the same time. He's also working on his own models, and Ghostwriter and you know, all the other stuff. So it's going to get interesting. Like he wants to power you, but also compete with you. Yeah. [00:42:47]Michael: And like, and we love replit. I think that a lot of the companies in our space, like we're all going to converge to solving a very similar problem, but from a different angle. So like replit approaches this problem from the IDE side. Like they started as like this IDE that you can run in the browser. And they started from that side, making coding just like more accessible. And we're approaching it from the side of like an LLM that's just like connected to everything that it needs to be connected to, which includes your code context. So that's why we're kind of making inroads into IDEs, but we're kind of, we're approaching this problem from different sides. And I think it'll be interesting to see where things end up. But I think that in the long, long term, we have an opportunity to also just have like this general technical reasoning engine product that's potentially also not just for, not just for programmers. It's also powered in this web interface, like where there's potential, I think other things that we will build that eventually might go beyond like our current scope. [00:43:49]Swyx: Exciting. We'll look forward to that. We're going to zoom out a little bit into sort of AI ecosystem stories, but first we got to get the Paul Graham, Ron Conway story. [00:43:59]Alessio: Yeah. [00:44:00]Michael: So flashback to last summer, we're in the YC batch. We're doing the summer batch, summer 22. So the summer batch runs from June to September, approximately. And so this was late July, early August, right around the time that many like YC startups start like going out, like during up, here's how we're going to pitch investors and everything. And at the same time, me and my co-founder, Justin, we were planning on moving to New York. So for a long time, actually, we were thinking about building this company in New York, mainly for personal reasons, actually, because like during the pandemic, pre-ChatGPT, pre last year, pre the AI boom, SF unfortunately really kind of, you know, like lost its luster. Yeah. Like no one was here. It was far from clear, like if there would be an AI boom, if like SF would be like... [00:44:49]Alessio: Back. [00:44:50]Michael: Yeah, exactly. Back. As everyone is saying these days, it was far from clear. And so, and all of our friends, we were graduating college because like we happened to just graduate college and immediately start YC, like we didn't even have, I think we had a week in between. [00:45:06]Swyx: You didn't bother looking for jobs. You were just like, this is what we want to do. [00:45:08]Michael: Well, actually both me and my co-founder, we had jobs that we secured in 2021 from previous internships, but we both, funny enough, when I spoke to my boss's boss at the company at where I reneged my offer, I told him we got into YC, they actually said, yeah, you should do YC. [00:45:27]Swyx: Wow. [00:45:28]Alessio: That's very selfless. [00:45:29]Swyx: That was really great that they did that. But in San Francisco, they would have offered to invest as well. [00:45:33]Michael: Yes, they would have. But yeah, but we were both planning to be in New York and all of our friends were there from college at this point, like we have this whole plan where like on August 1st, we're going to move to New York and we had like this Airbnb for the month of New York. We're going to stay there and we're going to work and like all of that. The day before we go to New York, I called Justin and I just, I tell him like, why are we doing this? Because in our batch, by the time August 1st rolled around, all of our mentors at YC were saying like, hey, like you should really consider staying in SF. [00:46:03]Swyx: It's the hybrid batch, right? [00:46:04]Michael: Yeah, it was the hybrid batch, but like there were already signs that like something was kind of like afoot in SF, even if like we didn't fully want to admit it yet. And so we were like, I don't know, I don't know. Something kind of clicked when the rubber met the road and it was time to go to New York. We're like, why are we doing this? And like, we didn't have any good reasons for staying in New York at that point beyond like our friends are there. So we still go to New York because like we have the Airbnb, like we don't have any other kind of place to go for the next few weeks. We're in New York and New York is just unfortunately too much fun. Like all of my other friends from college who are just, you know, basically starting their jobs, starting their lives as adults. They just stepped into these jobs, they're making all this money and they're like partying and like all these things are happening. And like, yeah, it's just a very distracting place to be. And so we were just like sitting in this like small, you know, like cramped apartment, terrible posture, trying to get as much work done as we can, too many distractions. And then we get this email from YC saying that Paul Graham is in town in SF and he is doing office hours with a certain number of startups in the current batch. And whoever signs up first gets it. And I happen to be super lucky. I was about to go for a run, but I just, I saw the email notification come across the street. I immediately clicked on the link and like immediately, like half the spots were gone, but somehow the very last spot was still available. And so I picked the very, very last time slot at 7 p.m. semi-strategically, you know, so we would have like time to go over. And also because I didn't really know how we're going to get to SF yet. And so we made a plan that we're going to fly from New York to SF and back to New York in one day and do like the full round trip. And we're going to meet with PG at the YC Mountain View office. And so we go there, we do that, we meet PG, we tell him about the startup. And one thing I love about PG is that he gets like, he gets so excited. Like when he gets excited about something, like you can see his eyes like really light up. And he'll just start asking you questions. In fact, it's a little challenging sometimes to like finish kind of like the rest of like the description of your pitch because like, he'll just like asking all these questions about how it works. And I'm like, you know, what's going on? [00:48:19]Swyx: What was the most challenging question that he asked you? [00:48:21]Michael: I think that like really how it worked. Because like as soon as like we told him like, hey, like we think that the future of search is answers, not links. Like we could really see like the gears turning in his head. I think we were like the first demo of that. [00:48:35]Swyx: And you're like 10 minutes with him, right? [00:48:37]Michael: We had like 45, yeah, we had a decent chunk of time. And so we tell him how it works. Like he's very excited about it. And I just like, I just blurted out, I just like asked him to invest and he hasn't even seen the product yet. We just asked him to invest and he says, yeah. And like, we're super excited about that. [00:48:55]Swyx: You haven't started your batch. [00:48:56]Michael: No, no, no. This is about halfway through the batch or two, two, no, two thirds of the batch. [00:49:02]Swyx: And you're like not technically fundraising yet. We're about to start fundraising. Yeah. [00:49:06]Michael: So we have like this demo and like we showed him and like there was still a lot of issues with the product, but I think like it must have like still kind of like blown his mind in some way. So like we're having fun. He's having fun. We have this dinner planned with this other friend that we had in SF because we were only there for that one day. So we thought, okay, you know, after an hour we'll be done, you know, we'll grab dinner with our friend and we'll fly back to New York. But PG was like, like, I'm having so much fun. Do you want to have dinner? Yeah. Come to my house. Or he's like, I gotta go have dinner with my wife, Jessica, who's also awesome, by the way. [00:49:40]Swyx: She's like the heart of YC. Yeah. [00:49:42]Michael: Jessica does not get enough credit as an aside for her role. [00:49:46]Swyx: He tries. [00:49:47]Michael: He understands like the technical side and she understands people and together they're just like a phenomenal team. But he's like, yeah, I got to go see Jessica, but you guys are welcome to come with. Do you want to come with? And we're like, we have this friend who's like right now outside of like literally outside the door who like we also promised to get dinner with. It's like, we'd love to, but like, I don't know if we can. He's like, oh, he's welcome to come too. So all of us just like hop in his car and we go to his house and we just like have this like we have dinner and we have this just chat about the future of search. Like I remember him telling Jessica distinctly, like our kids as kids are not going to know what like a search result is. Like they're just going to like have answers. That was really like a mind blowing, like inflection point moment for sure. [00:50:34]Swyx: Wow, that email changed your life. [00:50:35]Michael: Absolutely. [00:50:36]Swyx: And you also just spoiled the booking system for PG because now everyone's just going to go after the last slot. Oh man. [00:50:42]Michael: Yeah. But like, I don't know if he even does that anymore. [00:50:46]Swyx: He does. He does. Yeah. I've met other founders that he did it this year. [00:50:49]Michael: This year. Gotcha. But when we told him about how we did it, he was like, I am like frankly shocked that YC just did like a random like scheduling system. [00:50:55]Alessio: They didn't like do anything else. But, um. [00:50:58]Swyx: Okay. And then he introduces Duron Conway. Yes. Who is one of the most legendary angels in Silicon Valley. [00:51:04]Michael: Yes.So after PG invested, the rest of our round came together pretty quickly. [00:51:10]Swyx: I'm, by the way, I'm surprised. Like it's, it might feel like playing favorites right within the current batch to be like, yo, PG invested in this one. Right. [00:51:17]Alessio: Too bad for the others. [00:51:18]Swyx: Too bad for the others, I guess. [00:51:19]Michael: I think this is a bigger point about YC and like these accelerators in general is like YC gets like a lot of criticism from founders who feel like they didn't get value out of it. But like, in my view, YC is what you make of it. And YC tells you this. They're like, you really got to grab this opportunity, like buy the balls and make the most of it. And if you do, then it could be the best thing in the world. And if you don't, and if you're just kind of like a passive, even like an average founder in YC, you're still going to fail. And they tell you that. They're like, if you're average in your batch, you're going to fail. Like you have to just be exceptional in every way. With that in mind, perhaps that's even part of the reason why we asked PG to invest. And so yeah, after PG invested, the rest of our round came together pretty quickly, which I'm very fortunate for. And yeah, he introduced us to Ron. And after he did, I get a call from Ron. And then Ron says like, hey, like PG tells me what you're working on. I'd love to come meet you guys. And I'm like, wait, no way. And then we're just holed up in this like little house in San Mateo, which is a little small, but you know, it had a nice patio. In fact, we had like a monitor set up outside on the deck out there. And so Ron Conway comes over, we go over to the patio where like our workstation is. And Ron Conway, he's known for having like this notebook that he goes around with where he like sits down with the notebook and like takes very, very detailed notes. So he never like forgets anything. So he sits down with his notebook and he asks us like, hey guys, like, what do you need? And we're like, oh, we need GPUs. Back then, the GPU shortage wasn't even nearly as bad as it is now. But like even then, it was still challenging to get like the quota that we needed. And he's like, okay, no problem. And then like he leaves a couple hours later, we get an email and we're CC'd on an email that Ron wrote to Jensen, the CEO of Nvidia, saying like, hey, these guys need GPUs. [00:53:02]Swyx: You didn't say how much? It was just like, just give them GPUs. [00:53:04]Alessio: Basically, yeah. [00:53:05]Michael: Ron is known for writing these like one-liner emails that are like very short, but very to the point. And I think that's why like everyone responds to Ron. Everyone loves Ron. And so Jensen responds. He responds quickly, like tagging this VP of AI at Nvidia. And we start working with Nvidia, which is great. And something that I love about Nvidia, by the way, is that after that intro, we got matched with like a dedicated team. And at Nvidia, they know that they're going to win regardless. So they don't care where you get the GPUs from. They're like, they're truly neutral, unlike various sales reps that you might encounter at various like clouds and, you know, hardware companies, et cetera. They actually just want to help you because they know they don't care. Like regardless, they know that if you're getting Nvidia GPUs, they're still winning. So I guess that's a tip is that like if you're looking for GPUs like Nvidia, they'll help you do it. [00:53:54]Swyx: So just to tie up this thing, because so first of all, that's a fantastic story. And I just wanted to let you tell that because it's special. That is a strategic shift, right? That you already decided to make by the time you met Ron, which is we are going to have our own hardware. We're going to rack him in a data center somewhere. [00:54:11]Michael: Well, not even that we need our own hardware because actually we don't. Right. But we just we just need GPUs, period. And like every cloud loves like they have their own sales tactics and like they want to make you commit to long terms and like very non-flexible terms. And like there's a web of different things that you kind of have to navigate. Nvidia will kind of be to the point like, OK, you can do this on this cloud, this on this cloud. Like this is your budget. Maybe you want to consider buying as well. Like they'll help you walk through what the options are. And the reason why they're helpful is because like they look at the full picture. So they'll help you with the hardware. And in terms of software, they actually implemented a custom feature for us in Faster Transformer, which is one of their libraries.Swyx: For you? [00:54:53]Michael: For us. Yeah. Which is wild. I don't think they would have done it otherwise. They implemented streaming generation for T5 based models, which we were running at the time up until we switched to GPT in February, March of this year. So they implemented that just for us, actually, in Faster Transformer. And so like they'll help you like look at the complete picture and then just help you get done what you need to get done. I know one of your interests is also local models, open source models and hardware kind of goes hand in hand.Alessio: Any fun projects, explorations in the space that you want to share with local llamas and stuff? [00:55:27]Michael: Yeah, it's something that we're very interested in because something that kind of we're hearing a lot about is like people want something like find, especially comp

Thanks to the over 17,000 people who have joined the first AI Engineer Summit! A full recap is coming. Last call to fill out the State of AI Engineering survey! See our Community page for upcoming meetups in SF, Paris and NYC.This episode had good interest on Twitter.Fast.ai's “Practical Deep Learning” courses been watched by over >6,000,000 people, and the fastai library has over 25,000 stars on Github. Jeremy Howard, one of the creators of Fast, is now one of the most prominent and respected voices in the machine learning industry; but that wasn't always the case. Being non-consensus and right In 2018, Jeremy and Sebastian Ruder published a paper on ULMFiT (Universal Language Model Fine-tuning), a 3-step transfer learning technique for NLP tasks: The paper demonstrated that pre-trained language models could be fine-tuned on a specific task with a relatively small amount of data to achieve state-of-the-art results. They trained a 24M parameters model on WikiText-103 which was beat most benchmarks.While the paper had great results, the methods behind weren't taken seriously by the community: “Everybody hated fine tuning. Everybody hated transfer learning. I literally did tours trying to get people to start doing transfer learning and nobody was interested, particularly after GPT showed such good results with zero shot and few shot learning […] which I was convinced was not the right direction, but who's going to listen to me, cause as you said, I don't have a PhD, not at a university… I don't have a big set of computers to fine tune huge transformer models.”Five years later, fine-tuning is at the center of most major discussion topics in AI (we covered some like fine tuning vs RAG and small models fine tuning), and we might have gotten here earlier if Jeremy had OpenAI-level access to compute and distribution. At heart, Jeremy has always been “GPU poor”:“I've always been somebody who does not want to build stuff on lots of big computers because most people don't have lots of big computers and I hate creating stuff that most people can't use.”This story is a good reminder of how some of the best ideas are hiding in plain sight; we recently covered RWKV and will continue to highlight the most interesting research that isn't being done in the large labs. Replacing fine-tuning with continued pre-trainingEven though fine-tuning is now mainstream, we still have a lot to learn. The issue of “catastrophic forgetting” and potential solutions have been brought up in many papers: at the fine-tuning stage, the model can forget tasks it previously knew how to solve in favor of new ones. The other issue is apparent memorization of the dataset even after a single epoch, which Jeremy covered Can LLMs learn from a single example? but we still don't have the answer to. Despite being the creator of ULMFiT, Jeremy still professes that there are a lot of open questions on finetuning:“So I still don't know how to fine tune language models properly and I haven't found anybody who feels like they do.”He now advocates for "continued pre-training" - maintaining a diversity of data throughout the training process rather than separate pre-training and fine-tuning stages. Mixing instructional data, exercises, code, and other modalities while gradually curating higher quality data can avoid catastrophic forgetting and lead to more robust capabilities (something we covered in Datasets 101).“Even though I originally created three-step approach that everybody now does, my view is it's actually wrong and we shouldn't use it… the right way to do this is to fine-tune language models, is to actually throw away the idea of fine-tuning. There's no such thing. There's only continued pre-training. And pre-training is something where from the very start, you try to include all the kinds of data that you care about, all the kinds of problems that you care about, instructions, exercises, code, general purpose document completion, whatever. And then as you train, you gradually curate that, you know, you gradually make that higher and higher quality and more and more specific to the kinds of tasks you want it to do. But you never throw away any data….So yeah, that's now my view, is I think ULMFiT is the wrong approach. And that's why we're seeing a lot of these so-called alignment tax… I think it's actually because people are training them wrong.An example of this phenomena is CodeLlama, a LLaMA2 model finetuned on 500B tokens of code: while the model is much better at code, it's worse on generic tasks that LLaMA2 knew how to solve well before the fine-tuning. In the episode we also dive into all the places where open source model development and research is happening (academia vs Discords - tracked on our Communities list and on our survey), and how Jeremy recommends getting the most out of these diffuse, pseudonymous communities (similar to the Eleuther AI Mafia).Show Notes* Jeremy's Background* FastMail* Optimal Decisions* Kaggle* Enlitic* fast.ai* Rachel Thomas* Practical Deep Learning* fastai for PyTorch* nbdev* fastec2 (the underrated library we describe)* Can LLMs learn from a single example?* the Kaggle LLM Science Exam competition, which “challenges participants to answer difficult science-based questions written by a Large Language Model”.* Sebastian Ruder* Alec Radford* Sylvain Gugger* Stephen Merity* Chris Lattner* Modular.ai / Mojo* Jono Whittaker* Zeiler and Fergus paper* ULM Fit* DAWNBench* Phi-1* Code Llama* AlexNetTimestamps* [00:00:00] Intros and Jeremy's background* [00:05:28] Creating ULM Fit - a breakthrough in NLP using transfer learning* [00:06:32] The rise of GPT and the appeal of few-shot learning over fine-tuning* [00:10:00] Starting Fast.ai to distribute AI capabilities beyond elite academics* [00:14:30] How modern LMs like ChatGPT still follow the ULM Fit 3-step approach* [00:17:23] Meeting with Chris Lattner on Swift for TensorFlow at Google* [00:20:00] Continued pre-training as a fine-tuning alternative* [00:22:16] Fast.ai and looking for impact vs profit maximization* [00:26:39] Using Fast.ai to create an "army" of AI experts to improve their domains* [00:29:32] Fast.ai's 3 focus areas - research, software, and courses* [00:38:42] Fine-tuning memorization and training curve "clunks" before each epoch* [00:46:47] Poor training and fine-tuning practices may be causing alignment failures* [00:48:38] Academia vs Discords* [00:53:41] Jeremy's high hopes for Chris Lattner's Mojo and its potential* [01:05:00] Adding capabilities like SQL generation through quick fine-tuning* [01:10:12] Rethinking Fast.ai courses for the AI-assisted coding era* [01:14:53] Rapid model development has created major technical debt* [01:17:08] Lightning RoundAI Summary (beta)This is the first episode we're trying this. Here's an overview of the main topics before you dive in the transcript. * Jeremy's background and philosophies on AI* Studied philosophy and cognitive science in college* Focused on ethics and thinking about AI even 30 years ago* Believes AI should be accessible to more people, not just elite academics/programmers* Created fast.ai to make deep learning more accessible* Development of transfer learning and ULMFit* Idea of transfer learning critical for making deep learning accessible* ULMFit pioneered transfer learning for NLP* Proposed training general language models on large corpora then fine-tuning - this became standard practice* Faced skepticism that this approach would work from NLP community* Showed state-of-the-art results on text classification soon after trying it* Current open questions around fine-tuning LLMs* Models appear to memorize training data extremely quickly (after 1 epoch)* This may hurt training dynamics and cause catastrophic forgetting* Unclear how best to fine-tune models to incorporate new information/capabilities* Need more research on model training dynamics and ideal data mixing* Exciting new developments* Mojo and new programming languages like Swift could enable faster model innovation* Still lots of room for improvements in computer vision-like innovations in transformers* Small models with fine-tuning may be surprisingly capable for many real-world tasks* Prompting strategies enable models like GPT-3 to achieve new skills like playing chess at superhuman levels* LLMs are like computer vision in 2013 - on the cusp of huge new breakthroughs in capabilities* Access to AI research* Many key convos happen in private Discord channels and forums* Becoming part of these communities can provide great learning opportunities* Being willing to do real work, not just talk about ideas, is key to gaining access* The future of practical AI* Coding becoming more accessible to non-programmers through AI assistance* Pre-requisite programming experience for learning AI may no longer be needed* Huge open questions remain about how to best train, fine-tune, and prompt LLMsTranscriptAlessio: Hey everyone, welcome to the Latent Space Podcast. This is Alessio, partner and CTO at Residence at Decibel Partners, and I'm joined by my co-host Swyx, founder of Smol AI. [00:00:21]Swyx: Hey, and today we have in the remote studio, Jeremy Howard all the way from Australia. Good morning. [00:00:27]Jeremy: The remote studio, also known as my house. Good morning. Nice to see you. [00:00:32]Swyx: Nice to see you too. I'm actually very used to seeing you in your mask as a message to people, but today we're mostly audio. But thank you for doing the very important public service of COVID awareness. It was a pleasure. [00:00:46]Jeremy: It was all very annoying and frustrating and tedious, but somebody had to do it. [00:00:52]Swyx: Somebody had to do it, especially somebody with your profile. I think it really drives home the message. So we tend to introduce people for them and then ask people to fill in the blanks on the personal side. Something I did not know about you was that you graduated with a BA in philosophy from the University of Melbourne. I assumed you had a PhD. [00:01:14]Jeremy: No, I mean, I barely got through my BA because I was working 80 to 100 hour weeks at McKinsey and Company from 19 years old onwards. So I actually didn't attend any lectures in second and third year university. [00:01:35]Swyx: Well, I guess you didn't need it or you're very sort of self-driven and self-motivated. [00:01:39]Jeremy: I took two weeks off before each exam period when I was working at McKinsey. And then, I mean, I can't believe I got away with this in hindsight, I would go to all my professors and say, oh, I was meant to be in your class this semester and I didn't quite turn up. Were there any assignments I was meant to have done, whatever. I can't believe all of them let me basically have it. They basically always would say like, okay, well, if you can have this written by tomorrow, I'll accept it. So yeah, stressful way to get through university, but. [00:02:12]Swyx: Well, it shows that, I guess, you min-maxed the opportunities. That definitely was a precursor. [00:02:18]Jeremy: I mean, funnily, like in as much as I, you know, in philosophy, the things I found interesting and focused on in the little bit of time I did spend on it was ethics and cognitive science. And it's kind of really amazing that it's now come back around and those are actually genuinely useful things to know about, which I never thought would happen. [00:02:38]Swyx: A lot of, yeah, a lot of relevant conversations there. So you were a consultant for a while and then in the magical month of June 1989, you founded both Optimal Decisions and Fastmeal, which I also briefly used. So thank you for that. [00:02:53]Jeremy: Oh, good for you. Yeah. Cause I had read the statistics, which is that like 90% or something of small businesses fail. So I thought if I start two businesses, I have a higher chance. In hindsight, I was thinking of it as some kind of stochastic thing I didn't have control over, but it's a bit odd, but anyway. [00:03:10]Swyx: And then you were president and chief scientist at Kaggle, which obviously is the sort of composition platform of machine learning. And then Enlitic, where you were working on using deep learning to improve medical diagnostics and clinical decisions. Yeah. [00:03:28]Jeremy: I was actually the first company to use deep learning in medicine, so I kind of founded the field. [00:03:33]Swyx: And even now that's still like a pretty early phase. And I actually heard you on your new podcast with Tanish, where you went very, very deep into the stuff, the kind of work that he's doing, such a young prodigy at his age. [00:03:47]Jeremy: Maybe he's too old to be called a prodigy now, ex-prodigy. No, no. [00:03:51]Swyx: I think he still counts. And anyway, just to round out the bio, you have a lot more other credentials, obviously, but most recently you started Fast.ai, which is still, I guess, your primary identity with Rachel Thomas. So welcome. [00:04:05]Jeremy: Yep. [00:04:06]Swyx: Thanks to my wife. Thank you. Yeah. Doing a lot of public service there with getting people involved in AI, and I can't imagine a better way to describe it than fast, fast.ai. You teach people from nothing to stable diffusion in seven weeks or something, and that's amazing. Yeah, yeah. [00:04:22]Jeremy: I mean, it's funny, you know, when we started that, what was that, like 2016 or something, the idea that deep learning was something that you could make more accessible was generally considered stupid. Everybody knew that deep learning was a thing that you got a math or a computer science PhD, you know, there was one of five labs that could give you the appropriate skills and that you would join, yeah, basically from one of those labs, you might be able to write some papers. So yeah, the idea that normal people could use that technology to do good work was considered kind of ridiculous when we started it. And we weren't sure if it was possible either, but we kind of felt like we had to give it a go because the alternative was we were pretty sure that deep learning was on its way to becoming, you know, the most or one of the most, you know, important technologies in human history. And if the only people that could use it were a handful of computer science PhDs, that seemed like A, a big waste and B, kind of dangerous. [00:05:28]Swyx: Yeah. [00:05:29]Alessio: And, you know, well, I just wanted to know one thing on your bio that at Kaggle, you were also the top rank participant in both 2010 and 2011. So sometimes you see a lot of founders running companies that are not really in touch with the problem, but you were clearly building something that you knew a lot about, which is awesome. Talking about deep learning, you created, published a paper on ULM fit, which was kind of the predecessor to multitask learning and a lot of the groundwork that then went to into Transformers. I've read back on the paper and you turned this model, AWD LSTM, which I did the math and it was like 24 to 33 million parameters, depending on what training data set you use today. That's kind of like not even small, it's like super small. What were some of the kind of like contrarian takes that you had at the time and maybe set the stage a little bit for the rest of the audience on what was kind of like the state of the art, so to speak, at the time and what people were working towards? [00:06:32]Jeremy: Yeah, the whole thing was a contrarian take, you know. So okay, so we started Fast.ai, my wife and I, and we thought, yeah, so we're trying to think, okay, how do we make it more accessible? So when we started thinking about it, it was probably 2015 and then 2016, we started doing something about it. Why is it inaccessible? Okay, well, A, no one knows how to do it other than a few number of people. And then when we asked those few number of people, well, how do you actually get good results? They would say like, oh, it's like, you know, a box of tricks that aren't published. So you have to join one of the labs and learn the tricks. So a bunch of unpublished tricks, not much software around, but thankfully there was Theano and rappers and particularly Lasagna, the rapper, but yeah, not much software around, not much in the way of data sets, you know, very hard to get started in terms of the compute. Like how do you get that set up? So yeah, no, everything was kind of inaccessible. And you know, as we started looking into it, we had a key insight, which was like, you know what, most of the compute and data for image recognition, for example, we don't need to do it. You know, there's this thing which nobody knows about, nobody talks about called transfer learning, where you take somebody else's model, where they already figured out like how to detect edges and gradients and corners and text and whatever else, and then you can fine tune it to do the thing you want to do. And we thought that's the key. That's the key to becoming more accessible in terms of compute and data requirements. So when we started Fast.ai, we focused from day one on transfer learning. Lesson one, in fact, was transfer learning, literally lesson one, something not normally even mentioned in, I mean, there wasn't much in the way of courses, you know, the courses out there were PhD programs that had happened to have recorded their lessons and they would rarely mention it at all. We wanted to show how to do four things that seemed really useful. You know, work with vision, work with tables of data, work with kind of recommendation systems and collaborative filtering and work with text, because we felt like those four kind of modalities covered a lot of the stuff that, you know, are useful in real life. And no one was doing anything much useful with text. Everybody was talking about word2vec, you know, like king plus queen minus woman and blah, blah, blah. It was like cool experiments, but nobody's doing anything like useful with it. NLP was all like lemmatization and stop words and topic models and bigrams and SPMs. And it was really academic and not practical. But I mean, to be honest, I've been thinking about this crazy idea for nearly 30 years since I had done cognitive science at university, where we talked a lot about the CELS Chinese room experiment. This idea of like, what if there was somebody that could kind of like, knew all of the symbolic manipulations required to answer questions in Chinese, but they didn't speak Chinese and they were kind of inside a room with no other way to talk to the outside world other than taking in slips of paper with Chinese written on them and then they do all their rules and then they pass back a piece of paper with Chinese back. And this room with a person in is actually fantastically good at answering any question you give them written in Chinese. You know, do they understand Chinese? And is this, you know, something that's intelligently working with Chinese? Ever since that time, I'd say the most thought, to me, the most thoughtful and compelling philosophical response is yes. You know, intuitively it feels like no, because that's just because we can't imagine such a large kind of system. But you know, if it looks like a duck and acts like a duck, it's a duck, you know, or to all intents and purposes. And so I always kind of thought, you know, so this is basically a kind of analysis of the limits of text. And I kind of felt like, yeah, if something could ingest enough text and could use the patterns it saw to then generate text in response to text, it could appear to be intelligent, you know. And whether that means it is intelligent or not is a different discussion and not one I find very interesting. Yeah. And then when I came across neural nets when I was about 20, you know, what I learned about the universal approximation theorem and stuff, and I started thinking like, oh, I wonder if like a neural net could ever get big enough and take in enough data to be a Chinese room experiment. You know, with that background and this kind of like interest in transfer learning, you know, I'd been thinking about this thing for kind of 30 years and I thought like, oh, I wonder if we're there yet, you know, because we have a lot of text. Like I can literally download Wikipedia, which is a lot of text. And I thought, you know, how would something learn to kind of answer questions or, you know, respond to text? And I thought, well, what if we used a language model? So language models are already a thing, you know, they were not a popular or well-known thing, but they were a thing. But language models exist to this idea that you could train a model to fill in the gaps. Or actually in those days it wasn't fill in the gaps, it was finish a string. And in fact, Andrej Karpathy did his fantastic RNN demonstration from this at a similar time where he showed like you can have it ingest Shakespeare and it will generate something that looks a bit like Shakespeare. I thought, okay, so if I do this at a much bigger scale, using all of Wikipedia, what would it need to be able to do to finish a sentence in Wikipedia effectively, to do it quite accurately quite often? I thought, geez, it would actually have to know a lot about the world, you know, it'd have to know that there is a world and that there are objects and that objects relate to each other through time and cause each other to react in ways and that causes proceed effects and that, you know, when there are animals and there are people and that people can be in certain positions during certain timeframes and then you could, you know, all that together, you can then finish a sentence like this was signed into law in 2016 by US President X and it would fill in the gap, you know. So that's why I tried to create what in those days was considered a big language model trained on the entirety on Wikipedia, which is that was, you know, a bit unheard of. And my interest was not in, you know, just having a language model. My interest was in like, what latent capabilities would such a system have that would allow it to finish those kind of sentences? Because I was pretty sure, based on our work with transfer learning and vision, that I could then suck out those latent capabilities by transfer learning, you know, by fine-tuning it on a task data set or whatever. So we generated this three-step system. So step one was train a language model on a big corpus. Step two was fine-tune a language model on a more curated corpus. And step three was further fine-tune that model on a task. And of course, that's what everybody still does today, right? That's what ChatGPT is. And so the first time I tried it within hours, I had a new state-of-the-art academic result on IMDB. And I was like, holy s**t, it does work. And so you asked, to what degree was this kind of like pushing against the established wisdom? You know, every way. Like the reason it took me so long to try it was because I asked all my friends in NLP if this could work. And everybody said, no, it definitely won't work. It wasn't like, oh, maybe. Everybody was like, it definitely won't work. NLP is much more complicated than vision. Language is a much more vastly complicated domain. You know, and you've got problems like the grounding problem. We know from like philosophy and theory of mind that it's actually impossible for it to work. So yeah, so don't waste your time. [00:15:10]Alessio: Jeremy, had people not tried because it was like too complicated to actually get the data and like set up the training? Or like, were people just lazy and kind of like, hey, this is just not going to work? [00:15:20]Jeremy: No, everybody wasn't lazy. So like, so the person I thought at that time who, you know, there were two people I thought at that time, actually, who were the strongest at language models were Stephen Merity and Alec Radford. And at the time I didn't know Alec, but I, after we had both, after I'd released ULM Fit and he had released GPT, I organized a chat for both of us with Kate Metz in the New York Times. And Kate Metz answered, sorry, and Alec answered this question for Kate. And Kate was like, so how did, you know, GPT come about? And he said, well, I was pretty sure that pre-training on a general large corpus wouldn't work. So I hadn't tried it. And then I read ULM Fit and turns out it did work. And so I did it, you know, bigger and it worked even better. And similar with, with Stephen, you know, I asked Stephen Merity, like, why don't we just find, you know, take your AWD-ASTLM and like train it on all of Wikipedia and fine tune it? And he's kind of like, well, I don't think that's going to really lie. Like two years before I did a very popular talk at KDD, the conference where everybody in NLP was in the audience. I recognized half the faces, you know, and I told them all this, I'm sure transfer learning is the key. I'm sure ImageNet, you know, is going to be an NLP thing as well. And, you know, everybody was interested and people asked me questions afterwards and, but not just, yeah, nobody followed up because everybody knew that it didn't work. I mean, even like, so we were scooped a little bit by Dai and Lee, Kwok Lee at Google. They had, they had, I already, I didn't even realize this, which is a bit embarrassing. They had already done a large language model and fine tuned it. But again, they didn't create a general purpose, large language model on a general purpose corpus. They only ever tested a domain specific corpus. And I haven't spoken to Kwok actually about that, but I assume that the reason was the same. It probably just didn't occur to them that the general approach could work. So maybe it was that kind of 30 years of mulling over the, the cell Chinese room experiment that had convinced me that it probably would work. I don't know. Yeah. [00:17:48]Alessio: Interesting. I just dug up Alec announcement tweet from 2018. He said, inspired by Cobe, Elmo, and Yola, I'm fit. We should have a single transformer language model can be fine tuned to a wide variety. It's interesting because, you know, today people think of AI as the leader, kind of kind of like the research lab pushing forward the field. What was that at the time? You know, like kind of like going back five years, people think of it as an overnight success, but obviously it took a while. [00:18:16]Swyx: Yeah. Yeah. [00:18:17]Jeremy: No, I mean, absolutely. And I'll say like, you know, it's interesting that it mentioned Elmo because in some ways that was kind of diametrically opposed to, to ULM fit. You know, there was these kind of like, so there was a lot of, there was a lot of activity at the same time as ULM fits released. So there was, um, so before it, as Brian McCann, I think at Salesforce had come out with this neat model that did a kind of multitask learning, but again, they didn't create a general fine tune language model first. There was Elmo, um, which I think was a lip, you know, actually quite a few months after the first ULM fit example, I think. Um, but yeah, there was a bit of this stuff going on. And the problem was everybody was doing, and particularly after GPT came out, then everybody wanted to focus on zero shot and few shot learning. You know, everybody hated fine tuning. Everybody hated transfer learning. And like, I literally did tours trying to get people to start doing transfer learning and people, you know, nobody was interested, particularly after GPT showed such good results with zero shot and few shot learning. And so I actually feel like we kind of went backwards for years and, and not to be honest, I mean, I'm a bit sad about this now, but I kind of got so disappointed and dissuaded by like, it felt like these bigger lab, much bigger labs, you know, like fast AI had only ever been just me and Rachel were getting all of this attention for an approach I thought was the wrong way to do it. You know, I was convinced was the wrong way to do it. And so, yeah, for years people were really focused on getting better at zero shot and few shots and it wasn't until, you know, this key idea of like, well, let's take the ULM fit approach, but for step two, rather than fine tuning on a kind of a domain corpus, let's fine tune on an instruction corpus. And then in step three, rather than fine tuning on a reasonably specific task classification, let's fine tune on a, on a RLHF task classification. And so that was really, that was really key, you know, so I was kind of like out of the NLP field for a few years there because yeah, it just felt like, I don't know, pushing uphill against this vast tide, which I was convinced was not the right direction, but who's going to listen to me, you know, cause I, as you said, I don't have a PhD, not at a university, or at least I wasn't then. I don't have a big set of computers to fine tune huge transformer models. So yeah, it was definitely difficult. It's always been hard. You know, it's always been hard. Like I've always been somebody who does not want to build stuff on lots of big computers because most people don't have lots of big computers and I hate creating stuff that most people can't use, you know, and also stuff that's created on lots of big computers has always been like much more media friendly. So like, it might seem like a recent thing, but actually throughout my 30 years in data science, the attention's always been on, you know, the big iron results. So when I first started, everybody was talking about data warehouses and it was all about Teradata and it'd be like, oh, this big bank has this huge room full of computers and they have like terabytes of data available, you know, at the press of a button. And yeah, that's always what people want to talk about, what people want to write about. And then of course, students coming out of their PhDs and stuff, that's where they want to go work because that's where they read about. And to me, it's a huge distraction, you know, because like I say, most people don't have unlimited compute and I want to help most people, not the small subset of the most well-off people. [00:22:16]Alessio: That's awesome. And it's great to hear, you do such a great job educating that a lot of times you're not telling your own story, you know? So I love this conversation. And the other thing before we jump into Fast.AI, actually, a lot of people that I know, they run across a new architecture and whatnot, they're like, I got to start a company and raise a bunch of money and do all of this stuff. And say, you were like, I want everybody to have access to this. Why was that the case for you? Was it because you already had a successful venture in like FastMail and you were more interested in that? What was the reasoning? [00:22:52]Jeremy: It's a really good question. So I guess the answer is yes, that's the reason why. So when I was a teenager, I thought it would be really cool to like have my own company. You know, I didn't know the word startup. I didn't know the word entrepreneur. I didn't know the word VC. And I didn't really know what any of those things were really until after we started Kaggle, to be honest. Even the way it started to what we now call startups. I just thought they were just small businesses. You know, they were just companies. So yeah, so those two companies were FastMail and Optimal Decisions. FastMail was the first kind of synchronized email provider for non-businesses. So something you can get your same email at home, on your laptop, at work, on your phone, whatever. And then Optimal Decisions invented a new approach to insurance pricing. Something called profit-optimized insurance pricing. So I saw both of those companies, you know, after 10 years. And at that point, I had achieved the thing that as a teenager I had wanted to do. You know, it took a lot longer than it should have because I spent way longer in management consulting than I should have because I got caught up in that stupid rat race. But, you know, eventually I got there and I remember my mom saying to me, you must be so proud. You know, because she remembered my dream. She's like, you've done it. And I kind of reflected and I was like, I'm not proud at all. You know, like people quite liked FastMail. You know, it's quite nice to have synchronized email. It probably would have happened anyway. Yeah, I'm certainly not proud that I've helped some insurance companies suck more money out of their customers. Yeah, no, I'm not proud. You know, it's actually, I haven't really helped the world very much. You know, maybe in the insurance case I've made it a little bit worse. I don't know. So, yeah, I was determined to not waste more years of my life doing things, working hard to do things which I could not be reasonably sure would have a lot of value. So, you know, I took some time off. I wasn't sure if I'd ever work again, actually. I didn't particularly want to, because it felt like, yeah, it felt like such a disappointment. And, but, you know, and I didn't need to. I had enough money. Like, I wasn't super rich, but I had enough money. I didn't need to work. And I certainly recognized that amongst the other people I knew who had enough money that they didn't need to work, they all worked ridiculously hard, you know, and constantly put themselves in extremely stressful situations. And I thought, I don't want to be one of those idiots who's tied to, you know, buying a bigger plane than the next guy or whatever. You know, Kaggle came along and I mainly kind of did that just because it was fun and interesting to hang out with interesting people. But, you know, with Fast.ai in particular, you know, Rachel and I had a very explicit, you know, long series of conversations over a long period of time about like, well, how can we be the most helpful to society as a whole, and particularly to those people who maybe need more help, you know? And so we definitely saw the world going in a potentially pretty dystopian direction if the world's most powerful technology was controlled by a small group of elites. So we thought, yeah, we should focus on trying to help that not happen. You know, sadly, it looks like it still is likely to happen. But I mean, I feel like we've helped make it a little bit less likely. So we've done our bit. [00:26:39]Swyx: You've shown that it's possible. And I think your constant advocacy, your courses, your research that you publish, you know, just the other day you published a finding on, you know, learning that I think is still something that people are still talking about quite a lot. I think that that is the origin story of a lot of people who are going to be, you know, little Jeremy Howards, furthering your mission with, you know, you don't have to do everything by yourself is what I'm saying. No, definitely. Definitely. [00:27:10]Jeremy: You know, that was a big takeaway from like, analytic was analytic. It definitely felt like we had to do everything ourselves. And I kind of, I wanted to solve medicine. I'll say, yeah, okay, solving medicine is actually quite difficult. And I can't do it on my own. And there's a lot of other things I'd like to solve, and I can't do those either. So that was definitely the other piece was like, yeah, you know, can we create an army of passionate domain experts who can change their little part of the world? And that's definitely happened. Like I find nowadays, at least half the time, probably quite a bit more that I get in contact with somebody who's done really interesting work in some domain. Most of the time I'd say, they say, yeah, I got my start with fast.ai. So it's definitely, I can see that. And I also know from talking to folks at places like Amazon and Adobe and stuff, which, you know, there's lots of alumni there. And they say, oh my God, I got here. And like half of the people are fast.ai alumni. So it's fantastic. [00:28:13]Swyx: Yeah. [00:28:14]Jeremy: Actually, Andre Kapathy grabbed me when I saw him at NeurIPS a few years ago. And he was like, I have to tell you, thanks for the fast.ai courses. When people come to Tesla and they need to know more about deep learning, we always send them to your course. And the OpenAI Scholars Program was doing the same thing. So it's kind of like, yeah, it's had a surprising impact, you know, that's just one of like three things we do is the course, you know. [00:28:40]Swyx: Yes. [00:28:40]Jeremy: And it's only ever been at most two people, either me and Rachel or me and Sylvia nowadays, it's just me. So yeah, I think it shows you don't necessarily need a huge amount of money and a huge team of people to make an impact. [00:28:56]Swyx: Yeah. So just to reintroduce fast.ai for people who may not have dived into it much, there is the courses that you do. There is the library that is very well loved. And I kind of think of it as a nicer layer on top of PyTorch that people should start with by default and use it as the basis for a lot of your courses. And then you have like NBDev, which I don't know, is that the third one? [00:29:27]Jeremy: Oh, so the three areas were research, software, and courses. [00:29:32]Swyx: Oh, sorry. [00:29:32]Jeremy: So then in software, you know, fast.ai is the main thing, but NBDev is not far behind. But then there's also things like FastCore, GHAPI, I mean, dozens of open source projects that I've created and some of them have been pretty popular and some of them are still a little bit hidden, actually. Some of them I should try to do a better job of telling people about. [00:30:01]Swyx: What are you thinking about? Yeah, what's on the course of my way? Oh, I don't know, just like little things. [00:30:04]Jeremy: Like, for example, for working with EC2 and AWS, I created a FastEC2 library, which I think is like way more convenient and nice to use than anything else out there. And it's literally got a whole autocomplete, dynamic autocomplete that works both on the command line and in notebooks that'll like auto-complete your instance names and everything like that. You know, just little things like that. I try to make like, when I work with some domain, I try to make it like, I want to make it as enjoyable as possible for me to do that. So I always try to kind of like, like with GHAPI, for example, I think that GitHub API is incredibly powerful, but I didn't find it good to work with because I didn't particularly like the libraries that are out there. So like GHAPI, like FastEC2, it like autocompletes both at the command line or in a notebook or whatever, like literally the entire GitHub API. The entire thing is like, I think it's like less than 100K of code because it actually, as far as I know, the only one that grabs it directly from the official open API spec that GitHub produces. And like if you're in GitHub and you just type an API, you know, autocomplete API method and hit enter, it prints out the docs with brief docs and then gives you a link to the actual documentation page. You know, GitHub Actions, I can write now in Python, which is just so much easier than writing them in TypeScript and stuff. So, you know, just little things like that. [00:31:40]Swyx: I think that's an approach which more developers took to publish some of their work along the way. You described the third arm of FastAI as research. It's not something I see often. Obviously, you do do some research. And how do you run your research? What are your research interests? [00:31:59]Jeremy: Yeah, so research is what I spend the vast majority of my time on. And the artifacts that come out of that are largely software and courses. You know, so to me, the main artifact shouldn't be papers because papers are things read by a small exclusive group of people. You know, to me, the main artifacts should be like something teaching people, here's how to use this insight and here's software you can use that builds it in. So I think I've only ever done three first-person papers in my life, you know, and none of those are ones I wanted to do. You know, they were all ones that, like, so one was ULM Fit, where Sebastian Ruder reached out to me after seeing the course and said, like, you have to publish this as a paper, you know. And he said, I'll write it. He said, I want to write it because if I do, I can put it on my PhD and that would be great. And it's like, okay, well, I want to help you with your PhD. And that sounds great. So like, you know, one was the masks paper, which just had to exist and nobody else was writing it. And then the third was the Fast.ai library paper, which again, somebody reached out and said, please, please write this. We will waive the fee for the journal and everything and actually help you get it through publishing and stuff. So yeah, so I don't, other than that, I've never written a first author paper. So the research is like, well, so for example, you know, Dawn Bench was a competition, which Stanford ran a few years ago. It was kind of the first big competition of like, who can train neural nets the fastest rather than the most accurate. And specifically it was who can train ImageNet the fastest. And again, this was like one of these things where it was created by necessity. So Google had just released their TPUs. And so I heard from my friends at Google that they had put together this big team to smash Dawn Bench so that they could prove to people that they had to use Google Cloud and use their TPUs and show how good their TPUs were. And we kind of thought, oh s**t, this would be a disaster if they do that, because then everybody's going to be like, oh, deep learning is not accessible. [00:34:20]Swyx: You know, to actually be good at it, [00:34:21]Jeremy: you have to be Google and you have to use special silicon. And so, you know, we only found out about this 10 days before the competition finished. But, you know, we basically got together an emergency bunch of our students and Rachel and I and sat for the next 10 days and just tried to crunch through and try to use all of our best ideas that had come from our research. And so particularly progressive resizing, just basically train mainly on small things, train on non-square things, you know, stuff like that. And so, yeah, we ended up winning, thank God. And so, you know, we turned it around from being like, like, oh s**t, you know, this is going to show that you have to be Google and have TPUs to being like, oh my God, even the little guy can do deep learning. So that's an example of the kind of like research artifacts we do. And yeah, so all of my research is always, how do we do more with less, you know? So how do we get better results with less data, with less compute, with less complexity, with less education, you know, stuff like that. So ULM fits obviously a good example of that. [00:35:37]Swyx: And most recently you published, can LLMs learn from a single example? Maybe could you tell the story a little bit behind that? And maybe that goes a little bit too far into the learning of very low resource, the literature. [00:35:52]Jeremy: Yeah, yeah. So me and my friend, Jono Whittaker, basically had been playing around with this fun Kaggle competition, which is actually still running as we speak, which is, can you create a model which can answer multiple choice questions about anything that's in Wikipedia? And the thing that makes it interesting is that your model has to run on Kaggle within nine hours. And Kaggle's very, very limited. So you've only got 14 gig RAM, only two CPUs, and a small, very old GPU. So this is cool, you know, if you can do well at this, then this is a good example of like, oh, you can do more with less. So yeah, Jono and I were playing around with fine tuning, of course, transfer learning, pre-trained language models. And we saw this, like, so we always, you know, plot our losses as we go. So here's another thing we created. Actually, Sylvain Guuger, when he worked with us, created called fast progress, which is kind of like TQEDM, but we think a lot better. So we look at our fast progress curves, and they kind of go down, down, down, down, down, down, down, a little bit, little bit, little bit. And then suddenly go clunk, and they drop. And then down, down, down, down, down a little bit, and then suddenly clunk, they drop. We're like, what the hell? These clunks are occurring at the end of each epoch. So normally in deep learning, this would be, this is, you know, I've seen this before. It's always been a bug. It's always turned out that like, oh, we accidentally forgot to turn on eval mode during the validation set. So I was actually learning then, or, oh, we accidentally were calculating moving average statistics throughout the epoch. So, you know, so it's recently moving average or whatever. And so we were using Hugging Face Trainer. So, you know, I did not give my friends at Hugging Face the benefit of the doubt. I thought, oh, they've fucked up Hugging Face Trainer, you know, idiots. Well, you'll use the Fast AI Trainer instead. So we switched over to Learner. We still saw the clunks and, you know, that's, yeah, it shouldn't really happen because semantically speaking in the epoch, isn't like, it's not a thing, you know, like nothing happens. Well, nothing's meant to happen when you go from ending one epoch to starting the next one. So there shouldn't be a clunk, you know. So I kind of asked around on the open source discords. That's like, what's going on here? And everybody was just like, oh, that's just what, that's just what these training curves look like. Those all look like that. Don't worry about it. And I was like, oh, are you all using Trainer? Yes. Oh, well, there must be some bug with Trainer. And I was like, well, we also saw it in Learner [00:38:42]Swyx: and somebody else is like, [00:38:42]Jeremy: no, we've got our own Trainer. We get it as well. They're just like, don't worry about it. It's just something we see. It's just normal. [00:38:48]Swyx: I can't do that. [00:38:49]Jeremy: I can't just be like, here's something that's like in the previous 30 years of neural networks, nobody ever saw it. And now suddenly we see it. [00:38:57]Swyx: So don't worry about it. [00:38:59]Jeremy: I just, I have to know why. [00:39:01]Swyx: Can I clarify? This is, was everyone that you're talking to, were they all seeing it for the same dataset or in different datasets? [00:39:08]Jeremy: Different datasets, different Trainers. They're just like, no, this is just, this is just what it looks like when you fine tune language models. Don't worry about it. You know, I hadn't seen it before, but I'd been kind of like, as I say, I, you know, I kept working on them for a couple of years after ULM fit. And then I kind of moved on to other things, partly out of frustration. So I hadn't been fine tuning, you know, I mean, Lama's only been out for a few months, right? But I wasn't one of those people who jumped straight into it, you know? So I was relatively new to the kind of Lama fine tuning world, where else these guys had been, you know, doing it since day one. [00:39:49]Swyx: It was only a few months ago, [00:39:51]Jeremy: but it's still quite a bit of time. So, so yeah, they're just like, no, this is all what we see. [00:39:56]Swyx: Don't worry about it. [00:39:56]Jeremy: So yeah, I, I've got a very kind of like, I don't know, I've just got this brain where I have to know why things are. And so I kind of, I ask people like, well, why, why do you think it's happening? And they'd be like, oh, it would pretty obviously, cause it's like memorize the data set. It's just like, that can't be right. It's only seen it once. Like, look at this, the loss has dropped by 0.3, 0.3, which is like, basically it knows the answer. And like, no, no, it's just, it is, it's just memorize the data set. So yeah. So look, Jono and I did not discover this and Jono and I did not come up with a hypothesis. You know, I guess we were just the ones, I guess, who had been around for long enough to recognize that like, this, this isn't how it's meant to work. And so we, we, you know, and so we went back and like, okay, let's just run some experiments, you know, cause nobody seems to have actually published anything about this. [00:40:51]Well, not quite true.Some people had published things, but nobody ever actually stepped back and said like, what the hell, you know, how can this be possible? Is it possible? Is this what's happening? And so, yeah, we created a bunch of experiments where we basically predicted ahead of time. It's like, okay, if this hypothesis is correct, that it's memorized in the training set, then we ought to see blah, under conditions, blah, but not under these conditions. And so we ran a bunch of experiments and all of them supported the hypothesis that it was memorizing the data set in a single thing at once. And it's a pretty big data set, you know, which in hindsight, it's not totally surprising because the theory, remember, of the ULMFiT theory was like, well, it's kind of creating all these latent capabilities to make it easier for it to predict the next token. So if it's got all this kind of latent capability, it ought to also be really good at compressing new tokens because it can immediately recognize it as like, oh, that's just a version of this. So it's not so crazy, you know, but it is, it requires us to rethink everything because like, and nobody knows like, okay, so how do we fine tune these things? Because like, it doesn't even matter. Like maybe it's fine. Like maybe it's fine that it's memorized the data set after one go and you do a second go and okay, the validation loss is terrible because it's now really overconfident. [00:42:20]Swyx: That's fine. [00:42:22]Jeremy: Don't, you know, don't, I keep telling people, don't track validation loss, track validation accuracy because at least that will still be useful. Just another thing that's got lost since ULMFiT, nobody tracks accuracy of language models anymore. But you know, it'll still keep learning and it does, it does keep improving. But is it worse? You know, like, is it like, now that it's kind of memorized it, it's probably getting a less strong signal, you know, I don't know. So I still don't know how to fine tune language models properly and I haven't found anybody who feels like they do, like nobody really knows whether this memorization thing is, it's probably a feature in some ways. It's probably some things that you can do usefully with it. It's probably, yeah, I have a feeling it's messing up training dynamics as well. [00:43:13]Swyx: And does it come at the cost of catastrophic forgetting as well, right? Like, which is the other side of the coin. [00:43:18]Jeremy: It does to some extent, like we know it does, like look at Code Llama, for example. So Code Llama was a, I think it was like a 500 billion token fine tuning of Llama 2 using code. And also pros about code that Meta did. And honestly, they kind of blew it because Code Llama is good at coding, but it's bad at everything else, you know, and it used to be good. Yeah, I was pretty sure it was like, before they released it, me and lots of people in the open source discords were like, oh my God, you know, we know this is coming, Jan Lukinsk saying it's coming. I hope they kept at least like 50% non-code data because otherwise it's going to forget everything else. And they didn't, only like 0.3% of their epochs were non-code data. So it did, it forgot everything else. So now it's good at code and it's bad at everything else. So we definitely have catastrophic forgetting. It's fixable, just somebody has to do, you know, somebody has to spend their time training a model on a good mix of data. Like, so, okay, so here's the thing. Even though I originally created three-step approach that everybody now does, my view is it's actually wrong and we shouldn't use it. [00:44:36]Jeremy: And that's because people are using it in a way different to why I created it. You know, I created it thinking the task-specific models would be more specific. You know, it's like, oh, this is like a sentiment classifier as an example of a task, you know, but the tasks now are like a, you know, RLHF, which is basically like answer questions that make people feel happy about your answer. So that's a much more general task and it's a really cool approach. And so we see, for example, RLHF also breaks models like, you know, like GPT-4, RLHDEFT, we know from kind of the work that Microsoft did, you know, the pre, the earlier, less aligned version was better. And these are all kind of examples of catastrophic forgetting. And so to me, the right way to do this is to fine-tune language models, is to actually throw away the idea of fine-tuning. There's no such thing. There's only continued pre-training. And pre-training is something where from the very start, you try to include all the kinds of data that you care about, all the kinds of problems that you care about, instructions, exercises, code, general purpose document completion, whatever. And then as you train, you gradually curate that, you know, you gradually make that higher and higher quality and more and more specific to the kinds of tasks you want it to do. But you never throw away any data. You always keep all of the data types there in reasonably high quantities. You know, maybe the quality filter, you stop training on low quality data, because that's probably fine to forget how to write badly, maybe. So yeah, that's now my view, is I think ULM fit is the wrong approach. And that's why we're seeing a lot of these, you know, so-called alignment tacks and this view of like, oh, a model can't both code and do other things. And, you know, I think it's actually because people are training them wrong. [00:46:47]Swyx: Yeah, well, I think you have a clear [00:46:51]Alessio: anti-laziness approach. I think other people are not as good hearted, you know, they're like, [00:46:57]Swyx: hey, they told me this thing works. [00:46:59]Alessio: And if I release a model this way, people will appreciate it, I'll get promoted and I'll kind of make more money. [00:47:06]Jeremy: Yeah, and it's not just money. It's like, this is how citations work most badly, you know, so if you want to get cited, you need to write a paper that people in your field recognize as an advancement on things that we know are good. And so we've seen this happen again and again. So like I say, like zero shot and few shot learning, everybody was writing about that. Or, you know, with image generation, everybody just was writing about GANs, you know, and I was trying to say like, no, GANs are not the right approach. You know, and I showed again through research that we demonstrated in our videos that you can do better than GANs, much faster and with much less data. And nobody cared because again, like if you want to get published, you write a GAN paper that slightly improves this part of GANs and this tiny field, you'll get published, you know. So it's, yeah, it's not set up for real innovation. It's, you know, again, it's really helpful for me, you know, I have my own research lab with nobody telling me what to do and I don't even publish. So it doesn't matter if I get citations. And so I just write what I think actually matters. I wish there was, and, you know, and actually places like OpenAI, you know, the researchers there can do that as well. It's a shame, you know, I wish there was more academic, open venues in which people can focus on like genuine innovation. [00:48:38]Swyx: Twitter, which is unironically has become a little bit of that forum. I wanted to follow up on one thing that you mentioned, which is that you checked around the open source discords. I don't know if it's too, I don't know if it's a pusher to ask like what discords are lively or useful right now. I think that something I definitely felt like I missed out on was the early days of Luther AI, which is a very hard bit. And, you know, like what is the new Luther? And you actually shouted out the alignment lab AI discord in your blog post. And that was the first time I even knew, like I saw them on Twitter, never knew they had a discord, never knew that there was actually substantive discussions going on in there and that you were an active member of it. Okay, yeah. [00:49:23]Jeremy: And then even then, if you do know about that and you go there, it'll look like it's totally dead. And that's because unfortunately, nearly all the discords, nearly all of the conversation happens in private channels. You know, and that's, I guess. [00:49:35]Swyx: How does someone get into that world? Because it's obviously very, very instructive, right? [00:49:42]Jeremy: You could just come to the first AI discord, which I'll be honest with you, it's less bustling than some of the others, but it's not terrible. And so like, at least, to be fair, one of Emma's bustling channels is private. [00:49:57]Swyx: I guess. [00:49:59]Jeremy: So I'm just thinking. [00:50:01]Swyx: It's just the nature of quality discussion, right? Yeah, I guess when I think about it, [00:50:05]Jeremy: I didn't have any private discussions on our discord for years, but there was a lot of people who came in with like, oh, I just had this amazing idea for AGI. If you just thought about like, if you imagine that AI is a brain, then we, you know, this just, I don't want to talk about it. You know, I don't want to like, you don't want to be dismissive or whatever. And it's like, oh, well, that's an interesting comment, but maybe you should like, try training some models first to see if that aligns with your intuition. Like, oh, but how could I possibly learn? It's like, well, we have a course, just actually spend time learning. Like, you know, anyway. And there's like, okay, I know the people who always have good answers there. And so I created a private channel and put them all in it. And I got to admit, that's where I post more often because there's much less, you know, flight of fancy views about how we could solve AGI, blah, blah, blah. So there is a bit of that. But having said that, like, I think the bar is pretty low. Like if you join a Discord and you can hit the like participants or community or whatever button, you can see who's in it. And then you'll see at the top, who the admins or moderators or people in the dev role are. And just DM one of them and say like, oh, here's my GitHub. Well, here's some blog posts I wrote. You know, I'm interested in talking about this, you know, can I join the private channels? And I've never heard of anybody saying no. I will say, you know, Alutha's all pretty open. So you can do the Alutha Discord still. You know, one problem with the Alutha Discord is it's been going on for so long that it's like, it's very inside baseball. It's quite hard to get started. Yeah. Carpa AI looks, I think it's all open. That's just less stability. That's more accessible. [00:52:03]Swyx: Yeah. [00:52:04]Jeremy: There's also just recently, now it's research that does like the Hermes models and data set just opened. They've got some private channels, but it's pretty open, I think. You mentioned Alignment Lab, that one it's all the interesting stuff is on private channels. So just ask. If you know me, ask me, cause I've got admin on that one. There's also, yeah, OS Skunkworks, OS Skunkworks AI is a good Discord, which I think it's open. So yeah, they're all pretty good. [00:52:40]Swyx: I don't want you to leak any, you know, Discords that don't want any publicity, but this is all helpful. [00:52:46]Jeremy: We all want people, like we all want people. [00:52:49]Swyx: We just want people who like, [00:52:51]Jeremy: want to build stuff, rather than people who, and like, it's fine to not know anything as well, but if you don't know anything, but you want to tell everybody else what to do and how to do it, that's annoying. If you don't know anything and want to be told like, here's a really small kind of task that as somebody who doesn't know anything is going to take you a really long time to do, but it would still be helpful. Then, and then you go and do it. That would be great. The truth is, yeah, [00:53:19]Swyx: like, I don't know, [00:53:20]Jeremy: maybe 5% of people who come in with great enthusiasm and saying that they want to learn and they'll do anything. [00:53:25]Swyx: And then somebody says like, [00:53:25]Jeremy: okay, here's some work you can do. Almost nobody does that work. So if you're somebody who actually does the work and follows up, you will massively stand out. That's an extreme rarity. And everybody will then want to help you do more work. [00:53:41]Swyx: So yeah. [00:53:41]Jeremy: So just, yeah, just do work and people will want to support you. [00:53:47]Alessio: Our Discord used to be referral only for a long time. We didn't have a public invite and then we opened it and they're kind of like channel gating. Yeah. A lot of people just want to do, I remember it used to be like, you know, a forum moderator. [00:54:00]Swyx: It's like people just want to do [00:54:01]Alessio: like drive-by posting, [00:54:03]Swyx: you know, and like, [00:54:03]Alessio: they don't want to help the community. They just want to get their question answered. [00:54:07]Jeremy: I mean, the funny thing is our forum community does not have any of that garbage. You know, there's something specific about the low latency thing where people like expect an instant answer. And yeah, we're all somehow in a forum thread where they know it's like there forever. People are a bit more thoughtful, but then the forums are less active than they used to be because Discord has got more popular, you know? So it's all a bit of a compromise, you know, running a healthy community is, yeah, it's always a bit of a challenge. All right, we got so many more things [00:54:47]Alessio: we want to dive in, but I don't want to keep you here for hours. [00:54:50]Swyx: This is not the Lex Friedman podcast [00:54:52]Alessio: we always like to say. One topic I would love to maybe chat a bit about is Mojo, modular, you know, CrystalLiner, not many of you on the podcast. So we want to spend a little time there. You recently did a hacker's guide to language models and you ran through everything from quantized model to like smaller models, larger models, and all of that. But obviously modular is taking its own approach. Yeah, what got you excited? I know you and Chris have been talking about this for like years and a lot of the ideas you had, so. [00:55:23]Jeremy: Yeah, yeah, yeah, yeah, no, absolutely. So I met Chris, I think it was at the first TensorFlow Dev Summit. And I don't think he had even like, I'm not sure if he'd even officially started his employment with Google at that point. So I don't know, you know, certainly nothing had been mentioned. So I, you know, I admired him from afar with LLVM and Swift and whatever. And so I saw him walk into the courtyard at Google. It's just like, oh s**t, man, that's Chris Latner. I wonder if he would lower his standards enough to talk to me. Well, worth a try. So I caught up my courage because like nobody was talking to him. He looked a bit lost and I wandered over and it's like, oh, you're Chris Latner, right? It's like, what are you doing here? What are you doing here? And I was like, yeah, yeah, yeah. It's like, oh, I'm Jeremy Howard. It's like, oh, do you do some of this AI stuff? And I was like, yeah, yeah, I like this AI stuff. Are you doing AI stuff? It's like, well, I'm thinking about starting to do some AI stuff. Yeah, I think it's going to be cool. And it's like, wow. So like, I spent the next half hour just basically brain dumping all the ways in which AI was stupid to him. And he listened patiently. And I thought he probably wasn't even remember or care or whatever. But yeah, then I kind of like, I guess I re-caught up with him a few months later. And it's like, I've been thinking about everything you said in that conversation. And he like narrated back his response to every part of it, projects he was planning to do. And it's just like, oh, this dude follows up. Holy s**t. And I was like, wow, okay. And he was like, yeah, so we're going to create this new thing called Swift for TensorFlow. And it's going to be like, it's going to be a compiler with auto differentiation built in. And blah, blah, blah. And I was like, why would that help? [00:57:10]Swyx: You know, why would you? [00:57:10]Jeremy: And he was like, okay, with a compiler during the forward pass, you don't have to worry about saving context, you know, because a lot will be optimized in the backward. But I was like, oh my God. Because I didn't really know much about compilers. You know, I spent enough to kind of like, understand the ideas, but it hadn't occurred to me that a compiler basically solves a lot of the problems we have as end users. I was like, wow, that's amazing. Okay, you do know, right, that nobody's going to use this unless it's like usable. It's like, yeah, I know, right. So I was thinking you should create like a fast AI for this. So, okay, but I don't even know Swift. And he was like, well, why don't you start learning it? And if you have any questions, ask me. It's just like, holy s**t. Like, not only has Chris Latner lowered his standards enough to talk to me, but he's offering me personal tutoring on the programming language that he made. So I was just like, I'm not g

Want to help define the AI Engineer stack? Have opinions on the top tools, communities and builders? We're collaborating with friends at Amplify to launch the first State of AI Engineering survey! Please fill it out (and tell your friends)!If AI is so important, why is its software so bad?This was the motivating question for Chris Lattner as he reconnected with his product counterpart on Tensorflow, Tim Davis, and started working on a modular solution to the problem of sprawling, monolithic, fragmented platforms in AI development. They announced a $30m seed in 2022 and, following their successful double launch of Modular/Mojo

Thanks to the almost 30k people who tuned in to the last episode!Your podcast cohosts have been busy shipping:* Alessio open sourced smol-podcaster, which makes the show notes here! * swyx launched GodMode. Maybe someday the Cursor of browsers?* We're also helping organize a Llama Finetuning Hackameetup this Saturday in anticipation of the CodeLlama release. Lastly, more speakers were announced at AI Engineer Summit!

近3小时的硅谷AI重磅嘉宾现场对谈，下集光速奉上！如果你还没有听过上一期，赶紧去补课！ Hello World, who is OnBoard!? 简单介绍一下这次Monica 期待已久的嘉宾组合！ 两位都在OpenAI工作过的技术大牛，包括Nvidia资深研究员 Jim Fan, 除了对生成式agents 和机器人的具身智能有深度研究外，他的Twitter 连 Jeff bezos 都关注，是AI领域全球范围内的顶级大V。另一位嘉宾戴涵俊，Google Deepmind 的资深研究员，也是 Google 新一代大语言模型的深度参与者。最后，兼任主持和嘉宾的硅谷上市公司华人高管，硅谷徐老师, 每次来 Onboard! 串台都大受好评。 这是三个小时播客的第二部分。上一期的内容，我们深度讨论了最近AI领域最火的话题，Generative Agents, 生成式代理。这一期更是精彩纷呈，包含了AI领域更多核心话题，包括多模态大模型的研究进展，具备具身智能 embodied AI 的机器人如何打造，AI对saas的影响，我们对未来AI的商业和社会畅想等等。真的是非常尽兴的讨论，你也可以拿起笔记本做笔记了。 几位嘉宾都是长期在美国工作生活，夹杂英文在所难免，不接受抱怨。Enjoy! 嘉宾介绍 Jim Fan（推特：@DrJimFan），Nvidia 高级 AI 研究科学家，曾在OpenAI工作，Stanford PhD 李飞飞实验室 戴涵俊（推特：@hanjundai），Google Deepmind 资深研究员，深度参与 Google 大语言模型项目，曾在OpenAI工作，Georgia Tech PhD 硅谷徐老师（推特：@h0wie_xu），硅谷连续创业者、人工智能高管、斯坦福商学院客座讲师，「科技早知道」主播 ｜微信公众号：硅谷云｜ AI博客：howiexu.substack.com 主持：Monica（推特：@Monica_Yxie）：美元VC投资人，前 AWS 硅谷团队+AI创业公司打工人，公众号：M小姐研习录 (ID: MissMStudy) 主理人 | 即刻：莫妮卡同学 我们都聊了什么 ⁠01:55⁠ 为什么 Jim 觉得 Llama 2 作为语言模型，对于多模态模型和机器人有重大推动 ⁠05:24⁠ Hanjun 解读多模态大模型的两种实现方式 ⁠07:47⁠ 多模态大模型只是解锁了新的场景，还是能更大提升大模型本身的智能？如何理解大模型的智能？ ⁠12:34⁠ 为什么说机器人的多模态问题更有挑战？ ⁠16:35⁠ 处理多模态训练数据有哪些难点？ ⁠18:12⁠ 大模型训练还需要哪些工具？Infra/tooling 有哪些机会？ ⁠19:51⁠ 亲历OpenAI 的经历回顾和感受：2016-2020，OpenAI 都发生了什么 ⁠25:11⁠ OpenAI 近年的发展，哪个时刻震撼了你？ ⁠34:20⁠ 为什么说 Evaluation 是大语言模型最被低估的挑战之一？ ⁠39:54⁠ 未来1年和未来10年，你最期待人工智能领域带来什么？ ⁠46:17⁠ 我们自己和下一代应该如何为未来做准备？ ⁠59:33⁠ 有趣的 closing 和未来展望：被 Jeff Bezos 关注是什么感觉？！ 我们提到的内容 Llama 2: Meta 开源的大语言模型 Jim Fan 对于Llama 2 的解读 OpenAI 赢得DOTA 游戏比赛 LSTM (Long Short-term Memory) Jim Fan 对大猩猩玩Minecraft 的解读 DALL-E 2: DALL·E 2 is an AI system that can create realistic images and art from a description in natural language (by OpenAI) CLIP: Connecting text and image ImageNET: an image dataset organized according to the WordNet hierarchy. AlexNET: ImageNet Classification with Deep Convolutional Neural Networks 重点词汇 RLHF (Reinforcement Learning with Human Feedback): 人类反馈的强化学习 - 一种AI模型通过人类反馈与传统的强化学习结合来学习的方法。 Fine tuning: 微调 - 在特定的数据集上进一步训练预训练的机器学习模型，使其适应特定任务的过程。 Hallucination: 幻觉 - 在AI中，指的是模型生成不在输入中的信息，可能导致输出不准确。 Multi-modal model: 多模态模型 - 能够理解和处理多种类型数据（如文本、图像和声音）的模型。 Auto regressive model: 自回归模型 - 一种统计模型，它使用一个变量的过去值来预测其未来值。 Diffusion model: 扩散模型 - 用于描述信息、疾病或创新等东西如何在群体中传播的模型。 Tokenize: 分词 - 将文本分解成更小的部分（如单词或子词）的过程，通常在文本处理或自然语言处理中使用。 Intuitive physics: 直观物理 - 人类对物理现象的直观理解，例如物体如何移动或互相碰撞。 Embodied AI: 具体化的人工智能 - 通过物理或虚拟的身体与环境互动的AI系统，例如机器人或虚拟代理。 CVPR (Computer Vision and Pattern Recognition): 计算机视觉和模式识别 - 专门研究计算机如何“看”并从图像或视频中理解内容的领域。 Walkaround: 绕行 - 解决问题或障碍的方法 欢迎关注M小姐的微信公众号，了解更多中美软件、AI与创业投资的干货内容！ M小姐研习录 (ID: MissMStudy) 大家的点赞、评论、转发是对我们最好的鼓励！ 如果你能在小宇宙上点个赞，Apple Podcasts 上给个五星好评，就能让更多的朋友看到我们努力制作的内容，打赏请我们喝杯咖啡，就给你比心！ 有任何心得和建议，也欢迎在评论区跟我们互动~

We have just announced our first set of speakers at AI Engineer Summit! Sign up for the livestream or email sponsors@ai.engineer if you'd like to support.We are facing a massive GPU crunch. As both startups and VC's hoard Nvidia GPUs like countries count nuclear stockpiles, tweets about GPU shortages have become increasingly common. But what if we could run LLMs with AMD cards, or without a GPU at all? There's just one weird trick: compilation. And there's one person uniquely qualified to do it.We had the pleasure to sit down with Tianqi Chen, who's an Assistant Professor at CMU, where he both teaches the MLC course and runs the MLC group. You might also know him as the creator of XGBoost, Apache TVM, and MXNet, as well as the co-founder of OctoML. The MLC (short for Machine Learning Compilation) group has released a lot of interesting projects:* MLC Chat: an iPhone app that lets you run models like RedPajama-3B and Vicuna-7B on-device. It gets up to 30 tok/s!* Web LLM: Run models like LLaMA-70B in your browser (!!) to offer local inference in your product.* MLC LLM: a framework that allows any language models to be deployed natively on different hardware and software stacks.The MLC group has just announced new support for AMD cards; we previously talked about the shortcomings of ROCm, but using MLC you can get performance very close to the NVIDIA's counterparts. This is great news for founders and builders, as AMD cards are more readily available. Here are their latest results on AMD's 7900s vs some of top NVIDIA consumer cards.If you just can't get a GPU at all, MLC LLM also supports ARM and x86 CPU architectures as targets by leveraging LLVM. While speed performance isn't comparable, it allows for non-time-sensitive inference to be run on commodity hardware.We also enjoyed getting a peek into TQ's process, which involves a lot of sketching:With all the other work going on in this space with projects like ggml and Ollama, we're excited to see GPUs becoming less and less of an issue to get models in the hands of more people, and innovative software solutions to hardware problems!Show Notes* TQ's Projects:* XGBoost* Apache TVM* MXNet* MLC* OctoML* CMU Catalyst* ONNX* GGML* Mojo* WebLLM* RWKV* HiPPO* Tri Dao's Episode* George Hotz EpisodePeople:* Carlos Guestrin* Albert GuTimestamps* [00:00:00] Intros* [00:03:41] The creation of XGBoost and its surprising popularity* [00:06:01] Comparing tree-based models vs deep learning* [00:10:33] Overview of TVM and how it works with ONNX* [00:17:18] MLC deep dive* [00:28:10] Using int4 quantization for inference of language models* [00:30:32] Comparison of MLC to other model optimization projects* [00:35:02] Running large language models in the browser with WebLLM* [00:37:47] Integrating browser models into applications* [00:41:15] OctoAI and self-optimizing compute* [00:45:45] Lightning RoundTranscriptAlessio: Hey everyone, welcome to the Latent Space podcast. This is Alessio, Partner and CTO in Residence at Decibel Partners, and I'm joined by my co-host Swyx, writer and editor of Latent Space. [00:00:20]Swyx: Okay, and we are here with Tianqi Chen, or TQ as people call him, who is assistant professor in ML computer science at CMU, Carnegie Mellon University, also helping to run Catalyst Group, also chief technologist of OctoML. You wear many hats. Are those, you know, your primary identities these days? Of course, of course. [00:00:42]Tianqi: I'm also, you know, very enthusiastic open source. So I'm also a VP and PRC member of the Apache TVM project and so on. But yeah, these are the things I've been up to so far. [00:00:53]Swyx: Yeah. So you did Apache TVM, XGBoost, and MXNet, and we can cover any of those in any amount of detail. But maybe what's one thing about you that people might not learn from your official bio or LinkedIn, you know, on the personal side? [00:01:08]Tianqi: Let me say, yeah, so normally when I do, I really love coding, even though like I'm trying to run all those things. So one thing that I keep a habit on is I try to do sketchbooks. I have a book, like real sketchbooks to draw down the design diagrams and the sketchbooks I keep sketching over the years, and now I have like three or four of them. And it's kind of a usually a fun experience of thinking the design through and also seeing how open source project evolves and also looking back at the sketches that we had in the past to say, you know, all these ideas really turn into code nowadays. [00:01:43]Alessio: How many sketchbooks did you get through to build all this stuff? I mean, if one person alone built one of those projects, he'll be a very accomplished engineer. Like you built like three of these. What's that process like for you? Like it's the sketchbook, like the start, and then you think about the code or like. [00:01:59]Swyx: Yeah. [00:02:00]Tianqi: So, so usually I start sketching on high level architectures and also in a project that works for over years, we also start to think about, you know, new directions, like of course generative AI language model comes in, how it's going to evolve. So normally I would say it takes like one book a year, roughly at that rate. It's usually fun to, I find it's much easier to sketch things out and then gives a more like a high level architectural guide for some of the future items. Yeah. [00:02:28]Swyx: Have you ever published this sketchbooks? Cause I think people would be very interested on, at least on a historical basis. Like this is the time where XGBoost was born, you know? Yeah, not really. [00:02:37]Tianqi: I started sketching like after XGBoost. So that's a kind of missing piece, but a lot of design details in TVM are actually part of the books that I try to keep a record of. [00:02:48]Swyx: Yeah, we'll try to publish them and publish something in the journals. Maybe you can grab a little snapshot for visual aid. Sounds good. [00:02:57]Alessio: Yeah. And yeah, talking about XGBoost, so a lot of people in the audience might know it's a gradient boosting library, probably the most popular out there. And it became super popular because many people started using them in like a machine learning competitions. And I think there's like a whole Wikipedia page of like all state-of-the-art models. They use XGBoost and like, it's a really long list. When you were working on it, so we just had Tri Dao, who's the creator of FlashAttention on the podcast. And I asked him this question, it's like, when you were building FlashAttention, did you know that like almost any transform race model will use it? And so I asked the same question to you when you were coming up with XGBoost, like, could you predict it would be so popular or like, what was the creation process? And when you published it, what did you expect? We have no idea. [00:03:41]Tianqi: Like, actually, the original reason that we built that library is that at that time, deep learning just came out. Like that was the time where AlexNet just came out. And one of the ambitious mission that myself and my advisor, Carlos Guestrin, then is we want to think about, you know, try to test the hypothesis. Can we find alternatives to deep learning models? Because then, you know, there are other alternatives like, you know, support vector machines, linear models, and of course, tree-based models. And our question was, if you build those models and feed them with big enough data, because usually like one of the key characteristics of deep learning is that it's taking a lot [00:04:22]Swyx: of data, right? [00:04:23]Tianqi: So we will be able to get the same amount of performance. That's a hypothesis we're setting out to test. Of course, if you look at now, right, that's a wrong hypothesis, but as a byproduct, what we find out is that, you know, most of the gradient boosting library out there is not efficient enough for us to test that hypothesis. So I happen to have quite a bit of experience in the past of building gradient boosting trees and their variants. So Effective Action Boost was kind of like a byproduct of that hypothesis testing. At that time, I'm also competing a bit in data science challenges, like I worked on KDDCup and then Kaggle kind of become bigger, right? So I kind of think maybe it's becoming useful to others. One of my friends convinced me to try to do a Python binding of it. That tends to be like a very good decision, right, to be effective. Usually when I build it, we feel like maybe a command line interface is okay. And now we have a Python binding, we have R bindings. And then it realized, you know, it started getting interesting. People started contributing different perspectives, like visualization and so on. So we started to push a bit more on to building distributive support to make sure it works on any platform and so on. And even at that time point, when I talked to Carlos, my advisor, later, he said he never anticipated that we'll get to that level of success. And actually, why I pushed for gradient boosting trees, interestingly, at that time, he also disagreed. He thinks that maybe we should go for kernel machines then. And it turns out, you know, actually, we are both wrong in some sense, and Deep Neural Network was the king in the hill. But at least the gradient boosting direction got into something fruitful. [00:06:01]Swyx: Interesting. [00:06:02]Alessio: I'm always curious when it comes to these improvements, like, what's the design process in terms of like coming up with it? And how much of it is a collaborative with like other people that you're working with versus like trying to be, you know, obviously, in academia, it's like very paper-driven kind of research driven. [00:06:19]Tianqi: I would say the extra boost improvement at that time point was more on like, you know, I'm trying to figure out, right. But it's combining lessons. Before that, I did work on some of the other libraries on matrix factorization. That was like my first open source experience. Nobody knew about it, because you'll find, likely, if you go and try to search for the package SVD feature, you'll find some SVN repo somewhere. But it's actually being used for some of the recommender system packages. So I'm trying to apply some of the previous lessons there and trying to combine them. The later projects like MXNet and then TVM is much, much more collaborative in a sense that... But, of course, extra boost has become bigger, right? So when we started that project myself, and then we have, it's really amazing to see people come in. Michael, who was a lawyer, and now he works on the AI space as well, on contributing visualizations. Now we have people from our community contributing different things. So extra boost even today, right, it's a community of committers driving the project. So it's definitely something collaborative and moving forward on getting some of the things continuously improved for our community. [00:07:37]Alessio: Let's talk a bit about TVM too, because we got a lot of things to run through in this episode. [00:07:42]Swyx: I would say that at some point, I'd love to talk about this comparison between extra boost or tree-based type AI or machine learning compared to deep learning, because I think there is a lot of interest around, I guess, merging the two disciplines, right? And we can talk more about that. I don't know where to insert that, by the way, so we can come back to it later. Yeah. [00:08:04]Tianqi: Actually, what I said, when we test the hypothesis, the hypothesis is kind of, I would say it's partially wrong, because the hypothesis we want to test now is, can you run tree-based models on image classification tasks, where deep learning is certainly a no-brainer right [00:08:17]Swyx: now today, right? [00:08:18]Tianqi: But if you try to run it on tabular data, still, you'll find that most people opt for tree-based models. And there's a reason for that, in the sense that when you are looking at tree-based models, the decision boundaries are naturally rules that you're looking at, right? And they also have nice properties, like being able to be agnostic to scale of input and be able to automatically compose features together. And I know there are attempts on building neural network models that work for tabular data, and I also sometimes follow them. I do feel like it's good to have a bit of diversity in the modeling space. Actually, when we're building TVM, we build cost models for the programs, and actually we are using XGBoost for that as well. I still think tree-based models are going to be quite relevant, because first of all, it's really to get it to work out of the box. And also, you will be able to get a bit of interoperability and control monotonicity [00:09:18]Swyx: and so on. [00:09:19]Tianqi: So yes, it's still going to be relevant. I also sometimes keep coming back to think about, are there possible improvements that we can build on top of these models? And definitely, I feel like it's a space that can have some potential in the future. [00:09:34]Swyx: Are there any current projects that you would call out as promising in terms of merging the two directions? [00:09:41]Tianqi: I think there are projects that try to bring a transformer-type model for tabular data. I don't remember specifics of them, but I think even nowadays, if you look at what people are using, tree-based models are still one of their toolkits. So I think maybe eventually it's not even a replacement, it will be just an ensemble of models that you can call. Perfect. [00:10:07]Alessio: Next up, about three years after XGBoost, you built this thing called TVM, which is now a very popular compiler framework for models. Let's talk about, so this came out about at the same time as ONNX. So I think it would be great if you could maybe give a little bit of an overview of how the two things work together. Because it's kind of like the model, then goes to ONNX, then goes to the TVM. But I think a lot of people don't understand the nuances. I can get a bit of a backstory on that. [00:10:33]Tianqi: So actually, that's kind of an ancient history. Before XGBoost, I worked on deep learning for two years or three years. I got a master's before I started my PhD. And during my master's, my thesis focused on applying convolutional restricted Boltzmann machine for ImageNet classification. That is the thing I'm working on. And that was before AlexNet moment. So effectively, I had to handcraft NVIDIA CUDA kernels on, I think, a GTX 2070 card. I have a 22070 card. It took me about six months to get one model working. And eventually, that model is not so good, and we should have picked a better model. But that was like an ancient history that really got me into this deep learning field. And of course, eventually, we find it didn't work out. So in my master's, I ended up working on recommender system, which got me a paper, and I applied and got a PhD. But I always want to come back to work on the deep learning field. So after XGBoost, I think I started to work with some folks on this particular MXNet. At that time, it was like the frameworks of CAFE, Ciano, PyTorch haven't yet come out. And we're really working hard to optimize for performance on GPUs. At that time, I found it's really hard, even for NVIDIA GPU. It took me six months. And then it's amazing to see on different hardwares how hard it is to go and optimize code for the platforms that are interesting. So that gets me thinking, can we build something more generic and automatic? So that I don't need an entire team of so many people to go and build those frameworks. So that's the motivation of starting working on TVM. There is really too little about machine learning engineering needed to support deep learning models on the platforms that we're interested in. I think it started a bit earlier than ONNX, but once it got announced, I think it's in a similar time period at that time. So overall, how it works is that TVM, you will be able to take a subset of machine learning programs that are represented in what we call a computational graph. Nowadays, we can also represent a loop-level program ingest from your machine learning models. Usually, you have model formats ONNX, or in PyTorch, they have FX Tracer that allows you to trace the FX graph. And then it goes through TVM. We also realized that, well, yes, it needs to be more customizable, so it will be able to perform some of the compilation optimizations like fusion operator together, doing smart memory planning, and more importantly, generate low-level code. So that works for NVIDIA and also is portable to other GPU backends, even non-GPU backends [00:13:36]Swyx: out there. [00:13:37]Tianqi: So that's a project that actually has been my primary focus over the past few years. And it's great to see how it started from where I think we are the very early initiator of machine learning compilation. I remember there was a visit one day, one of the students asked me, are you still working on deep learning frameworks? I tell them that I'm working on ML compilation. And they said, okay, compilation, that sounds very ancient. It sounds like a very old field. And why are you working on this? And now it's starting to get more traction, like if you say Torch Compile and other things. I'm really glad to see this field starting to pick up. And also we have to continue innovating here. [00:14:17]Alessio: I think the other thing that I noticed is, it's kind of like a big jump in terms of area of focus to go from XGBoost to TVM, it's kind of like a different part of the stack. Why did you decide to do that? And I think the other thing about compiling to different GPUs and eventually CPUs too, did you already see some of the strain that models could have just being focused on one runtime, only being on CUDA and that, and how much of that went into it? [00:14:50]Tianqi: I think it's less about trying to get impact, more about wanting to have fun. I like to hack code, I had great fun hacking CUDA code. Of course, being able to generate CUDA code is cool, right? But now, after being able to generate CUDA code, okay, by the way, you can do it on other platforms, isn't that amazing? So it's more of that attitude to get me started on this. And also, I think when we look at different researchers, myself is more like a problem solver type. So I like to look at a problem and say, okay, what kind of tools we need to solve that problem? So regardless, it could be building better models. For example, while we build extra boots, we build certain regularizations into it so that it's more robust. It also means building system optimizations, writing low-level code, maybe trying to write assembly and build compilers and so on. So as long as they solve the problem, definitely go and try to do them together. And I also see it's a common trend right now. Like if you want to be able to solve machine learning problems, it's no longer at Aggressor layer, right? You kind of need to solve it from both Aggressor data and systems angle. And this entire field of machine learning system, I think it's kind of emerging. And there's now a conference around it. And it's really good to see a lot more people are starting to look into this. [00:16:10]Swyx: Yeah. Are you talking about ICML or something else? [00:16:13]Tianqi: So machine learning and systems, right? So not only machine learning, but machine learning and system. So there's a conference called MLsys. It's definitely a smaller community than ICML, but I think it's also an emerging and growing community where people are talking about what are the implications of building systems for machine learning, right? And how do you go and optimize things around that and co-design models and systems together? [00:16:37]Swyx: Yeah. And you were area chair for ICML and NeurIPS as well. So you've just had a lot of conference and community organization experience. Is that also an important part of your work? Well, it's kind of expected for academic. [00:16:48]Tianqi: If I hold an academic job, I need to do services for the community. Okay, great. [00:16:53]Swyx: Your most recent venture in MLsys is going to the phone with MLCLLM. You announced this in April. I have it on my phone. It's great. I'm running Lama 2, Vicuña. I don't know what other models that you offer. But maybe just kind of describe your journey into MLC. And I don't know how this coincides with your work at CMU. Is that some kind of outgrowth? [00:17:18]Tianqi: I think it's more like a focused effort that we want in the area of machine learning compilation. So it's kind of related to what we built in TVM. So when we built TVM was five years ago, right? And a lot of things happened. We built the end-to-end machine learning compiler that works, the first one that works. But then we captured a lot of lessons there. So then we are building a second iteration called TVM Unity. That allows us to be able to allow ML engineers to be able to quickly capture the new model and how we demand building optimizations for them. And MLCLLM is kind of like an MLC. It's more like a vertical driven organization that we go and build tutorials and go and build projects like LLM to solutions. So that to really show like, okay, you can take machine learning compilation technology and apply it and bring something fun forward. Yeah. So yes, it runs on phones, which is really cool. But the goal here is not only making it run on phones, right? The goal is making it deploy universally. So we do run on Apple M2 Macs, the 17 billion models. Actually, on a single batch inference, more recently on CUDA, we get, I think, the most best performance you can get out there already on the 4-bit inference. Actually, as I alluded earlier before the podcast, we just had a result on AMD. And on a single batch, actually, we can get the latest AMD GPU. This is a consumer card. It can get to about 80% of the 4019, so NVIDIA's best consumer card out there. So it's not yet on par, but thinking about how diversity and what you can enable and the previous things you can get on that card, it's really amazing that what you can do with this kind of technology. [00:19:10]Swyx: So one thing I'm a little bit confused by is that most of these models are in PyTorch, but you're running this inside a TVM. I don't know. Was there any fundamental change that you needed to do, or was this basically the fundamental design of TVM? [00:19:25]Tianqi: So the idea is that, of course, it comes back to program representation, right? So effectively, TVM has this program representation called TVM script that contains more like computational graph and operational representation. So yes, initially, we do need to take a bit of effort of bringing those models onto the program representation that TVM supports. Usually, there are a mix of ways, depending on the kind of model you're looking at. For example, for vision models and stable diffusion models, usually we can just do tracing that takes PyTorch model onto TVM. That part is still being robustified so that we can bring more models in. On language model tasks, actually what we do is we directly build some of the model constructors and try to directly map from Hugging Face models. The goal is if you have a Hugging Face configuration, we will be able to bring that in and apply optimization on them. So one fun thing about model compilation is that your optimization doesn't happen only as a soft language, right? For example, if you're writing PyTorch code, you just go and try to use a better fused operator at a source code level. Torch compile might help you do a bit of things in there. In most of the model compilations, it not only happens at the beginning stage, but we also apply generic transformations in between, also through a Python API. So you can tweak some of that. So that part of optimization helps a lot of uplifting in getting both performance and also portability on the environment. And another thing that we do have is what we call universal deployment. So if you get the ML program into this TVM script format, where there are functions that takes in tensor and output tensor, we will be able to have a way to compile it. So they will be able to load the function in any of the language runtime that TVM supports. So if you could load it in JavaScript, and that's a JavaScript function that you can take in tensors and output tensors. If you're loading Python, of course, and C++ and Java. So the goal there is really bring the ML model to the language that people care about and be able to run it on a platform they like. [00:21:37]Swyx: It strikes me that I've talked to a lot of compiler people, but you don't have a traditional compiler background. You're inventing your own discipline called machine learning compilation, or MLC. Do you think that this will be a bigger field going forward? [00:21:52]Tianqi: First of all, I do work with people working on compilation as well. So we're also taking inspirations from a lot of early innovations in the field. Like for example, TVM initially, we take a lot of inspirations from Halide, which is just an image processing compiler. And of course, since then, we have evolved quite a bit to focus on the machine learning related compilations. If you look at some of our conference publications, you'll find that machine learning compilation is already kind of a subfield. So if you look at papers in both machine learning venues, the MLC conferences, of course, and also system venues, every year there will be papers around machine learning compilation. And in the compiler conference called CGO, there's a C4ML workshop that also kind of trying to focus on this area. So definitely it's already starting to gain traction and becoming a field. I wouldn't claim that I invented this field, but definitely I helped to work with a lot of folks there. And I try to bring a perspective, of course, trying to learn a lot from the compiler optimizations as well as trying to bring in knowledges in machine learning and systems together. [00:23:07]Alessio: So we had George Hotz on the podcast a few episodes ago, and he had a lot to say about AMD and their software. So when you think about TVM, are you still restricted in a way by the performance of the underlying kernel, so to speak? So if your target is like a CUDA runtime, you still get better performance, no matter like TVM kind of helps you get there, but then that level you don't take care of, right? [00:23:34]Swyx: There are two parts in here, right? [00:23:35]Tianqi: So first of all, there is the lower level runtime, like CUDA runtime. And then actually for NVIDIA, a lot of the mood came from their libraries, like Cutlass, CUDN, right? Those library optimizations. And also for specialized workloads, actually you can specialize them. Because a lot of cases you'll find that if you go and do benchmarks, it's very interesting. Like two years ago, if you try to benchmark ResNet, for example, usually the NVIDIA library [00:24:04]Swyx: gives you the best performance. [00:24:06]Tianqi: It's really hard to beat them. But as soon as you start to change the model to something, maybe a bit of a variation of ResNet, not for the traditional ImageNet detections, but for latent detection and so on, there will be some room for optimization because people sometimes overfit to benchmarks. These are people who go and optimize things, right? So people overfit the benchmarks. So that's the largest barrier, like being able to get a low level kernel libraries, right? In that sense, the goal of TVM is actually we try to have a generic layer to both, of course, leverage libraries when available, but also be able to automatically generate [00:24:45]Swyx: libraries when possible. [00:24:46]Tianqi: So in that sense, we are not restricted by the libraries that they have to offer. That's why we will be able to run Apple M2 or WebGPU where there's no library available because we are kind of like automatically generating libraries. That makes it easier to support less well-supported hardware, right? For example, WebGPU is one example. From a runtime perspective, AMD, I think before their Vulkan driver was not very well supported. Recently, they are getting good. But even before that, we'll be able to support AMD through this GPU graphics backend called Vulkan, which is not as performant, but it gives you a decent portability across those [00:25:29]Swyx: hardware. [00:25:29]Alessio: And I know we got other MLC stuff to talk about, like WebLLM, but I want to wrap up on the optimization that you're doing. So there's kind of four core things, right? Kernel fusion, which we talked a bit about in the flash attention episode and the tiny grab one memory planning and loop optimization. I think those are like pretty, you know, self-explanatory. I think the one that people have the most questions, can you can you quickly explain [00:25:53]Swyx: those? [00:25:54]Tianqi: So there are kind of a different things, right? Kernel fusion means that, you know, if you have an operator like Convolutions or in the case of a transformer like MOP, you have other operators that follow that, right? You don't want to launch two GPU kernels. You want to be able to put them together in a smart way, right? And as a memory planning, it's more about, you know, hey, if you run like Python code, every time when you generate a new array, you are effectively allocating a new piece of memory, right? Of course, PyTorch and other frameworks try to optimize for you. So there is a smart memory allocator behind the scene. But actually, in a lot of cases, it's much better to statically allocate and plan everything ahead of time. And that's where like a compiler can come in. We need to, first of all, actually for language model, it's much harder because dynamic shape. So you need to be able to what we call symbolic shape tracing. So we have like a symbolic variable that tells you like the shape of the first tensor is n by 12. And the shape of the third tensor is also n by 12. Or maybe it's n times 2 by 12. Although you don't know what n is, right? But you will be able to know that relation and be able to use that to reason about like fusion and other decisions. So besides this, I think loop transformation is quite important. And it's actually non-traditional. Originally, if you simply write a code and you want to get a performance, it's very hard. For example, you know, if you write a matrix multiplier, the simplest thing you can do is you do for i, j, k, c, i, j, plus, equal, you know, a, i, k, times b, i, k. But that code is 100 times slower than the best available code that you can get. So we do a lot of transformation, like being able to take the original code, trying to put things into shared memory, and making use of tensor calls, making use of memory copies, and all this. Actually, all these things, we also realize that, you know, we cannot do all of them. So we also make the ML compilation framework as a Python package, so that people will be able to continuously improve that part of engineering in a more transparent way. So we find that's very useful, actually, for us to be able to get good performance very quickly on some of the new models. Like when Lamato came out, we'll be able to go and look at the whole, here's the bottleneck, and we can go and optimize those. [00:28:10]Alessio: And then the fourth one being weight quantization. So everybody wants to know about that. And just to give people an idea of the memory saving, if you're doing FB32, it's like four bytes per parameter. Int8 is like one byte per parameter. So you can really shrink down the memory footprint. What are some of the trade-offs there? How do you figure out what the right target is? And what are the precision trade-offs, too? [00:28:37]Tianqi: Right now, a lot of people also mostly use int4 now for language models. So that really shrinks things down a lot. And more recently, actually, we started to think that, at least in MOC, we don't want to have a strong opinion on what kind of quantization we want to bring, because there are so many researchers in the field. So what we can do is we can allow developers to customize the quantization they want, but we still bring the optimum code for them. So we are working on this item called bring your own quantization. In fact, hopefully MOC will be able to support more quantization formats. And definitely, I think there's an open field that's being explored. Can you bring more sparsities? Can you quantize activations as much as possible, and so on? And it's going to be something that's going to be relevant for quite a while. [00:29:27]Swyx: You mentioned something I wanted to double back on, which is most people use int4 for language models. This is actually not obvious to me. Are you talking about the GGML type people, or even the researchers who are training the models also using int4? [00:29:40]Tianqi: Sorry, so I'm mainly talking about inference, not training, right? So when you're doing training, of course, int4 is harder, right? Maybe you could do some form of mixed type precision for inference. I think int4 is kind of like, in a lot of cases, you will be able to get away with int4. And actually, that does bring a lot of savings in terms of the memory overhead, and so on. [00:30:09]Alessio: Yeah, that's great. Let's talk a bit about maybe the GGML, then there's Mojo. How should people think about MLC? How do all these things play together? I think GGML is focused on model level re-implementation and improvements. Mojo is a language, super sad. You're more at the compiler level. Do you all work together? Do people choose between them? [00:30:32]Tianqi: So I think in this case, I think it's great to say the ecosystem becomes so rich with so many different ways. So in our case, GGML is more like you're implementing something from scratch in C, right? So that gives you the ability to go and customize each of a particular hardware backend. But then you will need to write from CUDA kernels, and you write optimally from AMD, and so on. So the kind of engineering effort is a bit more broadened in that sense. Mojo, I have not looked at specific details yet. I think it's good to start to say, it's a language, right? I believe there will also be machine learning compilation technologies behind it. So it's good to say, interesting place in there. In the case of MLC, our case is that we do not want to have an opinion on how, where, which language people want to develop, deploy, and so on. And we also realize that actually there are two phases. We want to be able to develop and optimize your model. By optimization, I mean, really bring in the best CUDA kernels and do some of the machine learning engineering in there. And then there's a phase where you want to deploy it as a part of the app. So if you look at the space, you'll find that GGML is more like, I'm going to develop and optimize in the C language, right? And then most of the low-level languages they have. And Mojo is that you want to develop and optimize in Mojo, right? And you deploy in Mojo. In fact, that's the philosophy they want to push for. In the ML case, we find that actually if you want to develop models, the machine learning community likes Python. Python is a language that you should focus on. So in the case of MLC, we really want to be able to enable, not only be able to just define your model in Python, that's very common, right? But also do ML optimization, like engineering optimization, CUDA kernel optimization, memory planning, all those things in Python that makes you customizable and so on. But when you do deployment, we realize that people want a bit of a universal flavor. If you are a web developer, you want JavaScript, right? If you're maybe an embedded system person, maybe you would prefer C++ or C or Rust. And people sometimes do like Python in a lot of cases. So in the case of MLC, we really want to have this vision of, you optimize, build a generic optimization in Python, then you deploy that universally onto the environments that people like. [00:32:54]Swyx: That's a great perspective and comparison, I guess. One thing I wanted to make sure that we cover is that I think you are one of these emerging set of academics that also very much focus on your artifacts of delivery. Of course. Something we talked about for three years, that he was very focused on his GitHub. And obviously you treated XGBoost like a product, you know? And then now you're publishing an iPhone app. Okay. Yeah. Yeah. What is his thinking about academics getting involved in shipping products? [00:33:24]Tianqi: I think there are different ways of making impact, right? Definitely, you know, there are academics that are writing papers and building insights for people so that people can build product on top of them. In my case, I think the particular field I'm working on, machine learning systems, I feel like really we need to be able to get it to the hand of people so that really we see the problem, right? And we show that we can solve a problem. And it's a different way of making impact. And there are academics that are doing similar things. Like, you know, if you look at some of the people from Berkeley, right? A few years, they will come up with big open source projects. Certainly, I think it's just a healthy ecosystem to have different ways of making impacts. And I feel like really be able to do open source and work with open source community is really rewarding because we have a real problem to work on when we build our research. Actually, those research bring together and people will be able to make use of them. And we also start to see interesting research challenges that we wouldn't otherwise say, right, if you're just trying to do a prototype and so on. So I feel like it's something that is one interesting way of making impact, making contributions. [00:34:40]Swyx: Yeah, you definitely have a lot of impact there. And having experience publishing Mac stuff before, the Apple App Store is no joke. It is the hardest compilation, human compilation effort. So one thing that we definitely wanted to cover is running in the browser. You have a 70 billion parameter model running in the browser. That's right. Can you just talk about how? Yeah, of course. [00:35:02]Tianqi: So I think that there are a few elements that need to come in, right? First of all, you know, we do need a MacBook, the latest one, like M2 Max, because you need the memory to be big enough to cover that. So for a 70 million model, it takes you about, I think, 50 gigahertz of RAM. So the M2 Max, the upper version, will be able to run it, right? And it also leverages machine learning compilation. Again, what we are doing is the same, whether it's running on iPhone, on server cloud GPUs, on AMDs, or on MacBook, we all go through that same MOC pipeline. Of course, in certain cases, maybe we'll do a bit of customization iteration for either ones. And then it runs on the browser runtime, this package of WebLM. So that will effectively... So what we do is we will take that original model and compile to what we call WebGPU. And then the WebLM will be to pick it up. And the WebGPU is this latest GPU technology that major browsers are shipping right now. So you can get it in Chrome for them already. It allows you to be able to access your native GPUs from a browser. And then effectively, that language model is just invoking the WebGPU kernels through there. So actually, when the LATMAR2 came out, initially, we asked the question about, can you run 17 billion on a MacBook? That was the question we're asking. So first, we actually... Jin Lu, who is the engineer pushing this, he got 17 billion on a MacBook. We had a CLI version. So in MLC, you will be able to... That runs through a metal accelerator. So effectively, you use the metal programming language to get the GPU acceleration. So we find, okay, it works for the MacBook. Then we asked, we had a WebGPU backend. Why not try it there? So we just tried it out. And it's really amazing to see everything up and running. And actually, it runs smoothly in that case. So I do think there are some kind of interesting use cases already in this, because everybody has a browser. You don't need to install anything. I think it doesn't make sense yet to really run a 17 billion model on a browser, because you kind of need to be able to download the weight and so on. But I think we're getting there. Effectively, the most powerful models you will be able to run on a consumer device. It's kind of really amazing. And also, in a lot of cases, there might be use cases. For example, if I'm going to build a chatbot that I talk to it and answer questions, maybe some of the components, like the voice to text, could run on the client side. And so there are a lot of possibilities of being able to have something hybrid that contains the edge component or something that runs on a server. [00:37:47]Alessio: Do these browser models have a way for applications to hook into them? So if I'm using, say, you can use OpenAI or you can use the local model. Of course. [00:37:56]Tianqi: Right now, actually, we are building... So there's an NPM package called WebILM, right? So that you will be able to, if you want to embed it onto your web app, you will be able to directly depend on WebILM and you will be able to use it. We are also having a REST API that's OpenAI compatible. So that REST API, I think, right now, it's actually running on native backend. So that if a CUDA server is faster to run on native backend. But also we have a WebGPU version of it that you can go and run. So yeah, we do want to be able to have easier integrations with existing applications. And OpenAI API is certainly one way to do that. Yeah, this is great. [00:38:37]Swyx: I actually did not know there's an NPM package that makes it very, very easy to try out and use. I want to actually... One thing I'm unclear about is the chronology. Because as far as I know, Chrome shipped WebGPU the same time that you shipped WebILM. Okay, yeah. So did you have some kind of secret chat with Chrome? [00:38:57]Tianqi: The good news is that Chrome is doing a very good job of trying to have early release. So although the official shipment of the Chrome WebGPU is the same time as WebILM, actually, you will be able to try out WebGPU technology in Chrome. There is an unstable version called Canary. I think as early as two years ago, there was a WebGPU version. Of course, it's getting better. So we had a TVM-based WebGPU backhand two years ago. Of course, at that time, there were no language models. It was running on less interesting, well, still quite interesting models. And then this year, we really started to see it getting matured and performance keeping up. So we have a more serious push of bringing the language model compatible runtime onto the WebGPU. [00:39:45]Swyx: I think you agree that the hardest part is the model download. Has there been conversations about a one-time model download and sharing between all the apps that might use this API? That is a great point. [00:39:58]Tianqi: I think it's already supported in some sense. When we download the model, WebILM will cache it onto a special Chrome cache. So if a different web app uses the same WebILM JavaScript package, you don't need to redownload the model again. So there is already something there. But of course, you have to download the model once at least to be able to use it. [00:40:19]Swyx: Okay. One more thing just in general before we're about to zoom out to OctoAI. Just the last question is, you're not the only project working on, I guess, local models. That's right. Alternative models. There's gpt4all, there's olama that just recently came out, and there's a bunch of these. What would be your advice to them on what's a valuable problem to work on? And what is just thin wrappers around ggml? Like, what are the interesting problems in this space, basically? [00:40:45]Tianqi: I think making API better is certainly something useful, right? In general, one thing that we do try to push very hard on is this idea of easier universal deployment. So we are also looking forward to actually have more integration with MOC. That's why we're trying to build API like WebILM and other things. So we're also looking forward to collaborate with all those ecosystems and working support to bring in models more universally and be able to also keep up the best performance when possible in a more push-button way. [00:41:15]Alessio: So as we mentioned in the beginning, you're also the co-founder of Octomel. Recently, Octomel released OctoAI, which is a compute service, basically focuses on optimizing model runtimes and acceleration and compilation. What has been the evolution there? So Octo started as kind of like a traditional MLOps tool, where people were building their own models and you help them on that side. And then it seems like now most of the market is shifting to starting from pre-trained generative models. Yeah, what has been that experience for you and what you've seen the market evolve? And how did you decide to release OctoAI? [00:41:52]Tianqi: One thing that we found out is that on one hand, it's really easy to go and get something up and running, right? So if you start to consider there's so many possible availabilities and scalability issues and even integration issues since becoming kind of interesting and complicated. So we really want to make sure to help people to get that part easy, right? And now a lot of things, if we look at the customers we talk to and the market, certainly generative AI is something that is very interesting. So that is something that we really hope to help elevate. And also building on top of technology we build to enable things like portability across hardwares. And you will be able to not worry about the specific details, right? Just focus on getting the model out. We'll try to work on infrastructure and other things that helps on the other end. [00:42:45]Alessio: And when it comes to getting optimization on the runtime, I see when we run an early adopters community and most enterprises issue is how to actually run these models. Do you see that as one of the big bottlenecks now? I think a few years ago it was like, well, we don't have a lot of machine learning talent. We cannot develop our own models. Versus now it's like, there's these great models you can use, but I don't know how to run them efficiently. [00:43:12]Tianqi: That depends on how you define by running, right? On one hand, it's easy to download your MLC, like you download it, you run on a laptop, but then there's also different decisions, right? What if you are trying to serve a larger user request? What if that request changes? What if the availability of hardware changes? Right now it's really hard to get the latest hardware on media, unfortunately, because everybody's trying to work on the things using the hardware that's out there. So I think when the definition of run changes, there are a lot more questions around things. And also in a lot of cases, it's not only about running models, it's also about being able to solve problems around them. How do you manage your model locations and how do you make sure that you get your model close to your execution environment more efficiently? So definitely a lot of engineering challenges out there. That we hope to elevate, yeah. And also, if you think about our future, definitely I feel like right now the technology, given the technology and the kind of hardware availability we have today, we will need to make use of all the possible hardware available out there. That will include a mechanism for cutting down costs, bringing something to the edge and cloud in a more natural way. So I feel like still this is a very early stage of where we are, but it's already good to see a lot of interesting progress. [00:44:35]Alessio: Yeah, that's awesome. I would love, I don't know how much we're going to go in depth into it, but what does it take to actually abstract all of this from the end user? You know, like they don't need to know what GPUs you run, what cloud you're running them on. You take all of that away. What was that like as an engineering challenge? [00:44:51]Tianqi: So I think that there are engineering challenges on. In fact, first of all, you will need to be able to support all the kind of hardware backhand you have, right? On one hand, if you look at the media library, you'll find very surprisingly, not too surprisingly, most of the latest libraries works well on the latest GPU. But there are other GPUs out there in the cloud as well. So certainly being able to have know-hows and being able to do model optimization is one thing, right? Also infrastructures on being able to scale things up, locate models. And in a lot of cases, we do find that on typical models, it also requires kind of vertical iterations. So it's not about, you know, build a silver bullet and that silver bullet is going to solve all the problems. It's more about, you know, we're building a product, we'll work with the users and we find out there are interesting opportunities in a certain point. And when our engineer will go and solve that, and it will automatically reflect it in a service. [00:45:45]Swyx: Awesome. [00:45:46]Alessio: We can jump into the lightning round until, I don't know, Sean, if you have more questions or TQ, if you have more stuff you wanted to talk about that we didn't get a chance to [00:45:54]Swyx: touch on. [00:45:54]Alessio: Yeah, we have talked a lot. [00:45:55]Swyx: So, yeah. We always would like to ask, you know, do you have a commentary on other parts of AI and ML that is interesting to you? [00:46:03]Tianqi: So right now, I think one thing that we are really pushing hard for is this question about how far can we bring open source, right? I'm kind of like a hacker and I really like to put things together. So I think it's unclear in the future of what the future of AI looks like. On one hand, it could be possible that, you know, you just have a few big players, you just try to talk to those bigger language models and that can do everything, right? On the other hand, one of the things that Wailing Academic is really excited and pushing for, that's one reason why I'm pushing for MLC, is that can we build something where you have different models? You have personal models that know the best movie you like, but you also have bigger models that maybe know more, and you get those models to interact with each other, right? And be able to have a wide ecosystem of AI agents that helps each person while still being able to do things like personalization. Some of them can run locally, some of them, of course, running on a cloud, and how do they interact with each other? So I think that is a very exciting time where the future is yet undecided, but I feel like there is something we can do to shape that future as well. [00:47:18]Swyx: One more thing, which is something I'm also pursuing, which is, and this kind of goes back into predictions, but also back in your history, do you have any idea, or are you looking out for anything post-transformers as far as architecture is concerned? [00:47:32]Tianqi: I think, you know, in a lot of these cases, you can find there are already promising models for long contexts, right? There are space-based models, where like, you know, a lot of some of our colleagues from Albert, who he worked on this HIPPO models, right? And then there is an open source version called RWKV. It's like a recurrent models that allows you to summarize things. Actually, we are bringing RWKV to MOC as well, so maybe you will be able to see one of the models. [00:48:00]Swyx: We actually recorded an episode with one of the RWKV core members. It's unclear because there's no academic backing. It's just open source people. Oh, I see. So you like the merging of recurrent networks and transformers? [00:48:13]Tianqi: I do love to see this model space continue growing, right? And I feel like in a lot of cases, it's just that attention mechanism is getting changed in some sense. So I feel like definitely there are still a lot of things to be explored here. And that is also one reason why we want to keep pushing machine learning compilation, because one of the things we are trying to push in was productivity. So that for machine learning engineering, so that as soon as some of the models came out, we will be able to, you know, empower them onto those environments that's out there. [00:48:43]Swyx: Yeah, it's a really good mission. Okay. Very excited to see that RWKV and state space model stuff. I'm hearing increasing chatter about that stuff. Okay. Lightning round, as always fun. I'll take the first one. Acceleration. What has already happened in AI that you thought would take much longer? [00:48:59]Tianqi: Emergence of more like a conversation chatbot ability is something that kind of surprised me before it came out. This is like one piece that I feel originally I thought would take much longer, but yeah, [00:49:11]Swyx: it happens. And it's funny because like the original, like Eliza chatbot was something that goes all the way back in time. Right. And then we just suddenly came back again. Yeah. [00:49:21]Tianqi: It's always too interesting to think about, but with a kind of a different technology [00:49:25]Swyx: in some sense. [00:49:25]Alessio: What about the most interesting unsolved question in AI? [00:49:31]Swyx: That's a hard one, right? [00:49:32]Tianqi: So I can tell you like what kind of I'm excited about. So, so I think that I have always been excited about this idea of continuous learning and lifelong learning in some sense. So how AI continues to evolve with the knowledges that have been there. It seems that we're getting much closer with all those recent technologies. So being able to develop systems, support, and be able to think about how AI continues to evolve is something that I'm really excited about. [00:50:01]Swyx: So specifically, just to double click on this, are you talking about continuous training? That's like a training. [00:50:06]Tianqi: I feel like, you know, training adaptation and it's all similar things, right? You want to think about entire life cycle, right? The life cycle of collecting data, training, fine tuning, and maybe have your local context that getting continuously curated and feed onto models. So I think all these things are interesting and relevant in here. [00:50:29]Swyx: Yeah. I think this is something that people are really asking, you know, right now we have moved a lot into the sort of pre-training phase and off the shelf, you know, the model downloads and stuff like that, which seems very counterintuitive compared to the continuous training paradigm that people want. So I guess the last question would be for takeaways. What's basically one message that you want every listener, every person to remember today? [00:50:54]Tianqi: I think it's getting more obvious now, but I think one of the things that I always want to mention in my talks is that, you know, when you're thinking about AI applications, originally people think about algorithms a lot more, right? Our algorithm models, they are still very important. But usually when you build AI applications, it takes, you know, both algorithm side, the system optimizations, and the data curations, right? So it takes a connection of so many facades to be able to bring together an AI system and be able to look at it from that holistic perspective is really useful when we start to build modern applications. I think it's going to continue going to be more important in the future. [00:51:35]Swyx: Yeah. Thank you for showing the way on this. And honestly, just making things possible that I thought would take a lot longer. So thanks for everything you've done. [00:51:46]Tianqi: Thank you for having me. [00:51:47]Swyx: Yeah. [00:51:47]Alessio: Thanks for coming on TQ. [00:51:49]Swyx: Have a good one. [00:51:49] Get full access to Latent Space at www.latent.space/subscribe