Podcasts about enlitic

Good morning from Pharma and Biotech daily: the podcast that gives you only what's important to hear in Pharma e Biotech world. This week's commercialization news includes Dupixent's success in a chronic hives study, Lilly's development of a weekly insulin shot, and BioMarin's plans for growth. The House backs a bill restricting China's role in US biotech, while Lykos CEO is set to depart after FDA rejection and layoffs. The newsletter also discusses key developments in cell therapy and offers insights on utilizing a direct-to-patient model in the healthcare industry. Various resources and upcoming events in the biopharma industry are also highlighted. Biopharma Dive provides in-depth journalism and insights into the latest news and trends shaping the biotech and pharma industries.BridgeBio has reduced its gene therapy budget after data from a trial on an adrenal gland medicine did not meet the company's investment threshold. GlaxoSmithKline has discontinued a herpes vaccine after it did not meet efficacy goals in a phase 2 study. Roivant has launched a new 'vant' focused on a hypertension drug. Centessa's sleepiness drug has shown promising results in early studies, leading to a rise in the company's shares. Additionally, Dupixent has succeeded in a chronic hives study, giving Sanofi and Regeneron a chance to resubmit their application for approval. Investors are also paying attention to Centessa's sleepiness drug. This news comes alongside updates on other pharmaceutical developments, such as Saxenda's effectiveness for children as young as 6 and Roche's expansion of R&D labs. Additionally, the newsletter covers upcoming events and resources for biopharma professionals. Biopharma Dive provides in-depth coverage of news and trends in the biotech and pharma industries, including clinical trials, FDA approvals, gene therapy, drug pricing, and research partnerships.Iowa has awarded Centene's subsidiary, Iowa Total Care, a Medicaid managed care contract worth $2.8 billion. Telehealth groups are urging Congress and the White House to extend controlled substance virtual prescribing before pandemic-era flexibilities expire. The Biden administration has finalized a rule raising mental health coverage standards for private plans. Steward Health Care received court approval to sell its three most valuable hospitals to Orlando Health for $439 million. The importance of data quality in realizing value from medical imaging data is emphasized by Enlitic. Payers are encouraged to optimize quality and grow revenue through key strategies in an upcoming webinar. Healthcare Dive provides in-depth journalism and insight into the most impactful news and trends shaping healthcare across various sectors like health IT, policy & regulation, insurance, digital health, payer-provider partnerships, and value-based care.Novo Nordisk showcased its investigational GLP-1 pill that resulted in a remarkable 13% weight loss. This comes after positive Phase I results for the pill, which analysts compared to weight loss pills being developed by Lilly and Pfizer. Expanded coverage for cardiovascular disease under Medicare could have significant implications for Novo's obesity drug, Wegovy. Analysts estimate that the expansion of Wegovy's label beyond obesity could lead to an annual Medicare spending of $145 billion. Meanwhile, GSK has abandoned the development of its herpes vaccine after disappointing Phase I/II results, and Crispr Therapeutics and Vertex Pharmaceuticals are facing challenges in making their sickle cell gene therapy profitable. Novo's other drug, Saxenda, was found to effectively and safely lower BMI in children, according to a study published in NEJM. Additionally, Lilly continues to make progress with its once-weekly insulin, while Bain has raised $3 billion for a fund supporting life sciences companies. The biopharmaceutical industry continues to see changes, with Biomarin facing challenges and Terns moving forward in the obesity spac

27 Nov 2023 - Enlitic CEO Michael Sistenich discusses the company's AI solutions for healthcare data management as the company gears up for its December IPO.

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

Array Cast - July 8, 2022 Show Notes[01] 00:01:15 Dyalog Problem /solving Contest https://contest.dyalog.com/?goto=welcome[02] 00:01:35 Dyalog Early Bird Discount https://www.dyalog.com/user-meetings/dyalog22.htm[03] 00:02:40 Jeremy Howard wikipedia https://en.wikipedia.org/wiki/Jeremy_Howard_(entrepreneur) Fastmail https://www.fastmail.com/ Optimal Decisions Group https://www.finextra.com/newsarticle/18047/choicepoint-acquires-insurance-analytics-firm-optimal-decisions[04] 00:04:30 APL Study Group https://forums.fast.ai/t/apl-array-programming/97188[05] 00:05:50 McKinsey and Company https://en.wikipedia.org/wiki/McKinsey_%26_Company[06] 00:10:20 AT Kearney https://en.wikipedia.org/wiki/AT_Kearney[07] 00:12:33 MKL (Intel) https://en.wikipedia.org/wiki/Math_Kernel_Library[08] 00:13:00 BLAS http://www.netlib.org/blas/[09] 00:13:11 Perl BQN https://mlochbaum.github.io/BQN/running.html[10] 00:14:06 Raku https://en.wikipedia.org/wiki/Raku_%28programming_language%29[11] 00:15:45 kaggle https://www.kaggle.com/ kaggle competition https://www.kaggle.com/competitions/unimelb/leaderboard[12] 00:16:52 R https://en.wikipedia.org/wiki/R_(programming_language)[13] 00:18:50 Neural Networks https://en.wikipedia.org/wiki/Artificial_neural_network[14] 00:19:50 Enlitic https://www.enlitic.com/[15] 00:20:01 Fast.ai https://www.fast.ai/about/[16] 00:21:02 Numpy https://numpy.org/[17] 00:21:26 Leading Axis Theory https://aplwiki.com/wiki/Leading_axis_theory[18] 00:21:31 Rank Conjunction https://code.jsoftware.com/wiki/Vocabulary/quote[19] 00:21:40 Einstein notation https://en.wikipedia.org/wiki/Einstein_notation[20] 00:22:30 GPU https://en.wikipedia.org/wiki/Graphics_processing_unit[21] 00:22:55 CUDA https://en.wikipedia.org/wiki/CUDA[22] 00:23:30 Map https://en.wikipedia.org/wiki/Map_(higher-order_function)[23] 00:24:05 Data Science https://en.wikipedia.org/wiki/Data_science[24] 00:25:15 First Neural Network https://en.wikipedia.org/wiki/Frank_Rosenblatt[25] 00:28:51 Numpy Another Iverson Ghost https://dev.to/bakerjd99/numpy-another-iverson-ghost-9mc[26] 00:30:11 Pivot Tables https://en.wikipedia.org/wiki/Pivot_table[27] 00:30:36 SQL https://en.wikipedia.org/wiki/SQL[28] 00:31:25 Larry Wall "The three chief virtues of a programmer are: Laziness, Impatience and Hubris." From the glossary of the first Programming Perl book.[29] 00:32:00 Python https://www.python.org/[30] 00:36:25 Regular Expressions https://en.wikipedia.org/wiki/Regular_expression[31] 00:36:50 PyTorch https://pytorch.org/[32] 00:37:39 Notation as Tool of Thought https://www.jsoftware.com/papers/tot.htm[33] 00:37:55 Aaron Hsu codfns https://scholarworks.iu.edu/dspace/handle/2022/24749[34] 00:38:40 J https://www.jsoftware.com/#/[35] 00:39:06 Eric Iverson on Array Cast https://www.arraycast.com/episodes/episode10-eric-iverson[36] 00:40:18 Triangulation Jeremy Howard https://www.youtube.com/watch?v=hxB-rEQvBeM[37] 00:41:48 Google Brain https://en.wikipedia.org/wiki/Google_Brain[38] 00:42:30 RAPIDS https://rapids.ai/[39] 00:43:40 Julia https://julialang.org/[40] 00:43:50 llvm https://llvm.org/[41] 00:44:07 JAX https://jax.readthedocs.io/en/latest/notebooks/quickstart.html[42] 00:44:21 XLA https://www.tensorflow.org/xla[43] 00:44:32 MILAR https://www.tensorflow.org/mlir[44] 00:44:42 Chris Lattner https://en.wikipedia.org/wiki/Chris_Lattner[45] 00:44:53 Tensorflow https://www.tensorflow.org/[46] 00:49:33 torchscript https://pytorch.org/tutorials/beginner/Intro_to_TorchScript_tutorial.html[47] 00:50:09 Scheme https://en.wikipedia.org/wiki/Scheme_(programming_language)[48] 00:50:28 Swift https://en.wikipedia.org/wiki/Swift_(programming_language)[49] 00:51:10 DragonBox Algebra https://dragonbox.com/products/algebra-12[50] 00:52:47 APL Glyphs https://aplwiki.com/wiki/Glyph[51] 00:53:24 Dyalog APL https://www.dyalog.com/[52] 00:54:24 Jupyter https://jupyter.org/[53] 00:55:44 Jeremy's tweet of Meta Math https://twitter.com/jeremyphoward/status/1543738953391800320[54] 00:56:37 Power function https://aplwiki.com/wiki/Power_(function)[55] 01:03:06 Reshape ⍴ https://aplwiki.com/wiki/Reshape[56] 01:03:40 Stallman 'Rho, rho, rho' https://stallman.org/doggerel.html#APL[57] 01:04:20 APLcart https://aplcart.info/ BQNcrate https://mlochbaum.github.io/bqncrate/[58] 01:06:12 J for C programmers https://www.jsoftware.com/help/jforc/contents.htm[59] 01:07:54 Transpose episode https://www.arraycast.com/episodes/episode29-transpose[60] 01:10:00 APLcart video https://www.youtube.com/watch?v=r3owA7tfKE8[61] 01:12:28 Functional Programming https://en.wikipedia.org/wiki/Functional_programming[62] 01:13:00 List Comprehensions https://docs.python.org/3/tutorial/datastructures.html#list-comprehensions[63] 01:13:30 BQN to J https://mlochbaum.github.io/BQN/doc/fromJ.html BQN to Dyalog APL https://mlochbaum.github.io/BQN/doc/fromDyalog.html[64] 01:18:15 Einops https://cgarciae.github.io/einops/1-einops-basics/[65] 01:19:30 April Fools APL https://ci.tc39.es/preview/tc39/ecma262/sha/efb411f2f2a6f0e242849a8cc8d7e21bbcdff543/#sec-apl-expression-rules[66] 01:20:35 Flask library https://flask.palletsprojects.com/en/2.1.x/[67] 01:21:22 JuliaCon 2022 https://juliacon.org/2022/[68] 01:28:05 Myelination https://en.wikipedia.org/wiki/Myelin[69] 01:29:15 Sanyam Bhutani interview https://www.youtube.com/watch?v=g_6nQBsE4pU&t=2150s[70] 01:31:27 Jo Boaler Growth Mindset https://www.youcubed.org/resource/growth-mindset/[71] 01:33:45 Discovery Learning https://en.wikipedia.org/wiki/Discovery_learning[72] 01:37:05 Iverson Bracket https://en.wikipedia.org/wiki/Iverson_bracket[73] 01:39:14 Radek Osmulski Meta Learning https://rosmulski.gumroad.com/l/learn_machine_learning[74] 01:40:12 Top Down Learning https://medium.com/@jacksonbull1987/top-down-learning-4743f16d63d3[75] 01:41:20 Anki https://apps.ankiweb.net/[76] 01:43:50 Lex Fridman Interview https://www.youtube.com/watch?v=J6XcP4JOHmk Deep Talks #54 https://www.youtube.com/watch?v=n7YVlPszaWc0

In episode 10 of The Gradient Podcast, we interview data scientist, researcher, developer, educator, and entrepreneur Jeremy Howard.Subscribe to The Gradient Podcast: Apple Podcasts | Spotify | Pocket Casts | RSSJeremy Howard is a data scientist, researcher, developer, educator, and entrepreneur. Jeremy is a founding researcher at fast.ai, a research institute dedicated to making deep learning more accessible. He is also a Distinguished Research Scientist at the University of San Francisco, the chair of WAMRI, and is Chief Scientist at platform.ai. Previously, Jeremy was the founding CEO Enlitic, which was the first company to apply deep learning to medicine, was the President and Chief Scientist of the data science platform Kaggle, and was the founding CEO of two successful Australian startups.Podcast Theme: “MusicVAE: Trio 16-bar Sample #2” from "MusicVAE: A Hierarchical Latent Vector Model for Learning Long-Term Structure in Music". Get full access to The Gradient at thegradientpub.substack.com/subscribe

Coffee Sessions #47 with Jeremy Howard, fast.ai, AutoML, Software Engineering for ML. //Abstract Advancement in ML Workflows: You've been around the ML world for long enough to have seen how much workflows, tooling, frameworks, etc. have matured and allowed for greater scale and access. We'd love to reflect on your personal journey in this regard and hear about your early experiences putting models into production, as well as how you appreciate/might improve the process now. Data Professional Diversity and MLOps: Your work at fast.ai, Kaggle, and now with NBDEV has played a huge part in supercharging a diverse ecosystem of professionals that contribute to ML-like ML/data scientists, researchers, and ML engineers. As the attention turns to putting models into production, how do you think this range of professionals will evolve and work together? How will things around building models change as we build more? Turning Research into Practice: You've consistently been a leader in applying cutting-edge ideas from academia into practical code others can use. It's one of the things I appreciate most about the fast.ai course and package. How do you go about picking which ideas to invest in? What advice would you give to industry practitioners charged with a similar task at their company? // Bio Jeremy Howard is a data scientist, researcher, developer, educator, and entrepreneur. Jeremy is a founding researcher at fast.ai, a research institute dedicated to making deep learning more accessible. He is also a Distinguished Research Scientist at the University of San Francisco, the chair of WAMRI, and is Chief Scientist at platform.ai. Previously, Jeremy was the founding CEO of Enlitic, which was the first company to apply deep learning to medicine, and was selected as one of the world's top 50 smartest companies by MIT Tech Review two years running. He was the President and Chief Scientist of the data science platform Kaggle, where he was the top-ranked participant in international machine learning competitions for 2 years running. He was the founding CEO of two successful Australian startups (FastMail, and Optimal Decisions Group–purchased by Lexis-Nexis). Before that, he spent 8 years in management consulting, at McKinsey & Co, and at AT Kearney. Jeremy has invested in, mentored, and advised many startups, and contributed to many open-source projects. He has many media appearances, including writing for the Guardian, USA Today, and The Washington Post, appearing on ABC (Good Morning America), MSNBC (Joy Reid), CNN, Fox News, BBC, and was a regular guest on Australia's highest-rated breakfast news program. His talk on TED.com, “The wonderful and terrifying implications of computers that can learn”, has over 2.5 million views. He is a co-founder of the global Masks4All movement. // Other Links: jhoward.fastmail.fm enlitic.com jphoward.wordpress.com/ --------------- ✌️Connect With Us ✌️ ------------- Join our slack community: https://go.mlops.community/slack Follow us on Twitter: @mlopscommunity Sign up for the next meetup: https://go.mlops.community/register Connect with Demetrios on LinkedIn: https://www.linkedin.com/in/dpbrinkm/ Connect with Vishnu on LinkedIn: https://www.linkedin.com/in/vrachakonda/ Connect with Jeremy on LinkedIn: https://www.linkedin.com/in/howardjeremy/

Kevin Lyman was once the world’s highest ranked Warlock in Worlds of Warcraft and a professional Halo2 player. But that wasn’t his original plan. In fact, growing up in New Jersey, Kevin always wanted to be a scientist, even before he was sure of what that meant. While a student at Renselaer Polytechnic, Kevin took a number of jobs, including toy designer at Hasbro, sensor designer on the Falcon rocket for SpaceX and on the Excel team at Microsoft. But it was his first full time job as an engineer at Enlitic in 2015 that made him realize he wanted to apply his scientific ingenuity to healthcare, a field that he views as one of the few where you can help do something that really helps people. Enlitic took a number oof twists and turns as it built its imaging analytics products, and when those roads came back together as a result of his leadership, Kevin became the CEO in 2018. Kevin talks about what it’s like to be an under-30 CEO, the good and the bad of AI, and how one effectively balances intuition with the logical model inherent in an AI-focused company. He also talks about his current creative outlet – drawing – a sample of which you can see in evidence behind him in his photo. We were delighted to have Kevin on the show. We are grateful to Manatt Health for sponsoring today’s episode of Tech Tonics. Manatt Health integrates strategic business consulting, public policy acumen, legal excellence and deep analytics capabilities to better serve the complex needs of clients across America’s healthcare system. Together with its parent company, Manatt, Phelps & Phillips, the firm’s multidisciplinary team is dedicated to helping its clients across all industries grow and prosper.

Devon Bernard is an incredible full-stack engineer, manager, and entrepreneur. He's co-founded multiple companies including FlowActive, Jowl, and Rollio in addition to holding top engineering roles at Enlitic, Axgen, and now Somml. Learn more about Devon: https://www.linkedin.com/in/devonbernard/ (https://www.linkedin.com/in/devonbernard/) Want to level-up your skills in machine learning and software engineering? Subscribe to our newsletter: https://mlengineered.ck.page/943aa3fd46 (https://mlengineered.ck.page/943aa3fd46) Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ (https://www.givingwhatwecan.org/) Subscribe to ML Engineered: https://mlengineered.com/listen (https://mlengineered.com/listen) Follow Charlie on Twitter: https://twitter.com/CharlieYouAI (https://twitter.com/CharlieYouAI) Timestamps: (00:00) Intro (02:00) How he got started in CS (03:03) Working for GOOG and MSFT while running 2 startups (05:10) Learning to program (10:25) How he got started in entrepreneurship (17:30) Building an animal crossing trading exchange (21:37) Designing scalable and maintainable backends (25:55) Does he use formal design methods (DDD)? (28:24) What makes for a great engineer? (36:43) Functional programming (39:43) Increasing productivity of engineering teams (45:56) Managing up as an individual contributor (49:59) Consulting advice (01:04:20) Health-tech startups (01:19:27) Exciting opportunities outside of healthcare (01:28:34) Rapid Fire Questions Links: https://www.amazon.com/Never-Split-Difference-Negotiating-Depended-ebook/dp/B014DUR7L2/ (Never Split the Difference) https://www.amazon.com/Radical-Candor-Revised-Kick-Ass-Humanity-ebook/dp/B07P9LPXPT/ (Radical Candor)

Jordan Dunne works as a Technical Program Manager at Google Payments. He previously worked as a Program Manager at Microsoft, Lead Forward-Deployed Engineer at Enlitic, and Product Manager at Vim. Learn more about Jordan: https://www.linkedin.com/in/jordanwdunne/ (https://www.linkedin.com/in/jordanwdunne/) Want to level-up your skills in machine learning and software engineering? Subscribe to our newsletter: https://mlengineered.ck.page/943aa3fd46 (https://mlengineered.ck.page/943aa3fd46) Take the Giving What We Can Pledge: https://www.givingwhatwecan.org/ (https://www.givingwhatwecan.org/) Subscribe to ML Engineered: https://www.mlengineered.com/listen (https://www.mlengineered.com/listen) Follow Charlie on Twitter: https://twitter.com/CharlieYouAI (https://twitter.com/CharlieYouAI) Timestamps: (02:00) How were you exposed to CS and why did you pursue it? (03:25) Is software engineering actually engineering? (06:40) How do you define product management? (11:05) When did you realize you wanted to be a PM instead of a developer? (16:40) Project vs Program vs Product Management (18:35) Effective PM as leverage on a dev team (24:05) What can engineers do to make PM's lives easier? (26:10) Companies moving towards technical PMs? (30:20) Handling the added uncertainty from Data/ML products (42:00) ML models held to a higher standard than their human equivalents (45:15) Why are Xoogle PMs so successful? (52:10) Google's and Boeing's cultures influenced by their business models (56:00) "Needless complexity" in PM (59:00) Getting better at estimation (01:04:00) Knowing ML evaluation metrics as a PM (01:06:30) Getting better at communication (01:14:20) Prioritizing what to learn (01:16:50) Keeping the big picture in mind (01:20:00) Rapid fire questions Links: https://www.amazon.com/Thanks-Feedback-Science-Receiving-Well/dp/0670014664 (Thanks for the Feedback) https://www.amazon.com/Better-Angels-Our-Nature-Violence/dp/0143122010 (Better Angels of Our Nature) https://www.amazon.com/Crucial-Conversations-Talking-Stakes-Second/dp/1469266822 (Crucial Conversations) https://www.imdb.com/title/tt0057115/ (The Great Escape)

Jeremy Howard is a founding researcher at fast.ai, a research institute dedicated to making Deep Learning more accessible. Previously, he was the CEO and Founder at Enlitic, an advanced machine learning company in San Francisco, California. Howard is a faculty member at Singularity University, where he teaches data science. He is also a Young Global Leader with the World Economic Forum, and spoke at the World Economic Forum Annual Meeting 2014 on "Jobs For The Machines." Howard advised Khosla Ventures as their Data Strategist, identifying the biggest opportunities for investing in data-driven startups and mentoring their portfolio companies to build data-driven businesses. Howard was the founding CEO of two successful Australian startups, FastMail and Optimal Decisions Group. Before that, he spent eight years in management consulting, at McKinsey & Company and AT Kearney. TOPICS COVERED: 0:00 Introduction 0:52 Dad things 2:40 The story of Fast.ai 4:57 How the courses have evolved over time 9:24 Jeremy’s top down approach to teaching 13:02 From Fast.ai the course to Fast.ai the library 15:08 Designing V2 of the library from the ground up 21:44 The ingenious type dispatch system that powers Fast.ai 25:52 Were you able to realize the vision behind v2 of the library 28:05 Is it important to you that Fast.ai is used by everyone in the world, beyond the context of learning 29:37 Real world applications of Fast.ai, including animal husbandry 35:08 Staying ahead of the new developments in the field 38:50 A bias towards learning by doing 40:02 What’s next for Fast.ai 40.35 Python is not the future of Machine Learning 43:58 One underrated aspect of machine learning 45:25 Biggest challenge of machine learning in the real world Follow Jeremy on Twitter: https://twitter.com/jeremyphoward Links: Deep learning R&D & education: http://fast.ai Software: http://docs.fast.ai Book: http://up.fm/book Course: http://course.fast.ai Papers: The business impact of deep learning https://dl.acm.org/doi/10.1145/2487575.2491127 De-identification Methods for Open Health Data https://www.jmir.org/2012/1/e33/ Visit our podcasts homepage for transcripts and more episodes! www.wandb.com/podcast

This week on the show, Cam and Yonas welcome Tara Adams, a Product Manager at Enlitic! Tara recently transitioned into a new product role, and she chats with the guys about changing roles and starting fresh during quarantine, the differences in product teams across different organizations, and what it's like being an experienced product hire at a new company.

Longevity, eternal youth or even immortality have been an aspiration in religion and culture throughout history. Today, people adopt all sorts of approaches to increase their wellbeing, delay aging and avoid diseases. Efforts are increasingly quantified with sensors, wearables, or even biohacking - interventions to influence body biology. The new hope for advancements in longevity is seen in artificial intelligence, which is becoming increasingly powerful. Alex Zhavoronkov has been researching the use of AI in aging for years. He is the CEO of Insilico Medicine, a Baltimore-based leader in the next-generation artificial intelligence technologies for drug discovery and aging biomarkers discovery. He truly is a well of knowledge - since 2012 he published over 130 peer-reviewed research papers and 2 books including "The Ageless Generation: How Biomedical Advances Will Transform the Global Economy" (Palgrave Macmillan, 2013). In this episode, he talks about the complexity of aging as a biological process, types of artificial intelligence and the role of AI in research advancements.   Some of his latest research articles include:  Blood Biochemistry Analysis to Detect Smoking Status and Quantify Accelerated Aging in Smokers - https://www.nature.com/articles/s41598-018-35704-w#author-information Artificial intelligence for aging and longevity research: Recent advances and perspectives - https://www.sciencedirect.com/science/article/pii/S156816371830240X?via%3Dihub Artificial Intelligence for Drug Discovery, Biomarker Development, and Generation of Novel Chemistry - https://pubs.acs.org/doi/10.1021/acs.molpharmaceut.8b00930 Listen also:  F013 What to expect from artificial intelligence in healthcare in the next 10 years? (Sally Daub, Enlitic) https://medium.com/faces-of-digital-health/f013-what-to-expect-from-artificial-intelligence-in-healthcare-in-the-next-10-years-fdaf2edf32f8

Creative destruction occurs when a disruptive industry supplants a legacy industry, causing the loss of some jobs and the creation of others. Marc explains the need to get ahead of the disruptions in your industry, using examples from industrial giants who quickly became insignificant or who vanished as a result of unexpected market or social movements. Marc shares current technological changes and views of more drastic changes soon to come.   Listen in for a sample of the helpful advice in the new edition of Repurpose Your Career.   Key Takeaways: [1:14] Marc welcomes you to Episode 123 of the Repurpose Your Career podcast. Career Pivot brings you this podcast. CareerPivot.com is one of the very few websites dedicated to those of us in the second half of life and our careers. Take a moment to check out the blog and the other resources delivered to you, free of charge. [1:44] If you are enjoying this podcast, please share it with other like-minded souls. Subscribe on CareerPivot.com, iTunes, or any of the other apps that supply podcasts. Share it on social media or just tell your neighbors, and colleagues. The more people they reach, the more people they can help.[2:06] Next week, Marc will be interviewing Patti Temple Rocks, author of I’m Not Done: It's Time to Talk About Ageism in the Workplace, a great book on ageism. Marc thinks you will like this great interview. [2:20] If you are a regular listener to this show, you probably noticed that Marc has stopped talking about the next edition of his book, Repurpose Your Career. Susan Lahey and Marc are back on track and a draft of the third edition just got sent to the copy editor. [2:35] Marc’s plan is to release the third edition of the book in September of this year. [2:41] This week, Marc will read the pre-release chapter, “Learn to Embrace Creative Destruction.” He plans to release this chapter in PDF form to the review team within a week. [2:54] If you are interested in being on the release team and get early access to chapters in the new edition, go to careerpivot.com/rycteam. Marc hopes you enjoy this episode.[3:12] The pre-release chapter of “Learn to Embrace Creative Destruction.” In his book, Antifragile: Things That Gain From Disorder, writer Nassim Nicholas Taleb explains the problem of turkeys. A butcher feeds a turkey for 1,000 days. Every day that that turkey’s life remains constant confirms the surety of his current existence. [3:40] “This is the way it goes. This is the way it has always gone. This is the way it always will go.” All of the data confirms that butchers love turkeys. The turkey can rest confident in this idea because he has 999 days of benevolent treatment to back it up. [4:01] Then, a few days before Thanksgiving, everything in his worldview is upturned. This is what Taleb calls a ‘black swan event.’ All of the evidence proves it can’t happen, until it does. [4:16] The truth is that this is the normal course of things in human existence. A sudden rain shower hits the picnic. A car accident ruins travel plans. A financial windfall or unexpected romance changes your trajectory. Death comes unexpectedly. This is how life is. [4:36] In the world of work, the force behind these changes is often the power of creative destruction. One thing is destroyed and another is created. The turkey’s life is over. Dinner is served. [4:52] If the change is in our favor, we think it’s a good change. If the change is not in our favor, we think it’s a bad change. Regardless of how we feel about it, though, it’s going to happen. We need not be taken by surprise, like the turkey. [5:10] I was listening to a rebroadcast of a Freakonomics Radio podcast called “How Safe Is Your Job?” The hosts were talking about pianos. In 1905, they said, 400,000 pianos were made in America. If you wanted music in your house, you learned to play the piano. [5:31] The phonograph had been created 30 years before, in 1877 but phonograph sales didn’t take off until 1915. A decade later, the radio became popular. Then, eventually, the tape player, the eight-track, the CD player, and streaming and… [5:49] Today only about 30,000 pianos are made each year, about eight percent of the number made in 1905. [5:58] Each new iteration of musical enjoyment was a form of creative destruction. Each caused people in the previous industry to lose jobs or pivot. [6:09] In 1975, an employee of the Kodak company created a digital camera. But instead of developing it, Kodak concluded it was a non-starter because they didn’t think people wanted to look at their pictures on their TVs. So the company continued on focusing on chemical film until it became clear that they had bet on the wrong horse. [6:31] In 2001, Kodak had the second-most-popular digital camera on the market but lost $60 on every sale. A decade later, Kodak declared bankruptcy. [6:47] In these cases, creative destruction took 20, 30, or 40 years to bring down one giant and birth another. Now, that pace is accelerating. [6:58] Amazon.com was founded in 1994 and, initially, just sold books. They were credited with the demise of several brick-and-mortar bookselling chains. Over the next 11 years, Amazon moved into retailing pretty much everything and by 2015, it passed Walmart to be the most valuable retailer in the world, by market capitalization. [7:24] It took them and their online retail competitors only a few years to bring down what had been a staple of the world economy, the brick-and-mortar store. [7:36] In 2018, Amazon started buying surviving brick-and-mortar retailers, including Whole Foods, presumably to collect data on people who still shop there and further strengthen their market presence. [7:50] Now, Amazon is opening brick-and-mortar stores around the country, including convenience and book stores. They’re remaking retail, Amazon-style. [8:00] The iPhone was created only 11 years ago, in 2007, but at that time, I used my phone for talking to people. [8:10] Today, this is what I use my phone for: the weather report from the Weather Channel app; manage my social media with LinkedIn and Twitter. I removed the Facebook app after the last presidential elections. [8:23] I take and view pictures, edit files in Google Drive or Dropbox, communicate with clients over Skype, check scores on the ESPN app, find my keys, using the Tile app, listen to podcasts and audiobooks (as I no longer listen to the radio), find the new coffee shop via Google Maps or Apple Maps, … [8:45] … enter the YMCA by swiping the barcode in the YMCA app, manage multiple credit cards and bank accounts, show the police officer my proof of insurance via the State Farm app, check airline schedules to see if my son’s flight home is on time, … [9:04] … search Google to answer the question my wife just asked me, and watch House Hunters International on HGTV via the Sling TV app. Oh, and a lot of people use them to listen to music. [9:17] Because of the technology we have now, everything is being reimagined, reconfigured, reinvented, at a pace our parents never could have conceived of. One way to say it is the world is being ‘SMACed.’ [9:38] S = Social media: LinkedIn Facebook, Twitter, Pinterest, Instagram, and Snapchat. Today, people go to social media for everything. It’s the U.S. Mail, the telephone, the photo album, the gossip chain, the opinion column, the news, the entertainment, education, and job board, all rolled in one. [10:02] It’s also one place employers go to find you and find out whether you are the kind of candidate they want. [10:10] M = Mobile. Roughly 60% of adults get their news on a mobile device. According to the research by the Pew Research Foundation, mobile apps track our behavior and our preferences as well as give us a means to pay for things. People use mobile devices to shop, to bank, and to date. [10:32] If your career isn’t mobile-friendly, you will be left in the dust. [10:39] A = Analytics. More data has been collected in the last few years than was collected in the previous century. A lot of it is coming voluntarily from our activities via social media and mobile. [10:55] How we shop, where we shop, what we pay with, where we go online, and even how long it takes to get somewhere are some of the things that inform this data. Do you remember the movie, Minority Report, where Tom Cruise walks through the mall and hyper-customized ads display everywhere? [11:15] Analytics will affect how you are hired. [11:19] C = Cloud. Cloud is changing everything in the technology world. Most of the major technology hardware vendors are seeing portions of their business collapse because data isn’t being stored on their hardware. It’s being stored in the Cloud. [11:39] A classic example is IBM, who missed the shift and is seeing massive changes in their business. Their hardware business is collapsing. Cloud computing is sometimes referred to as SaaS or Software as a Service. [11:55] With SaaS, you don’t have to buy a disc. You don’t have to save data on your computer. You don’t have to have a photo album or a filing cabinet. You can keep everything in the Cloud. [12:10] Also, you can get services in the Cloud, rather than hiring someone to do them, like bookkeeping, record keeping, customer relationship management, and marketing. [12:19] You can book travel on the Cloud, make appointments in the Cloud, even hold conversations in the Cloud. SMAC is a representation of what we’ve long called the Robot Invasion. Articles have said for decades that robots are going to take our jobs. And SMAC is robots doing just that. [12:40] Some people assume the jobs robots can do are severely limited. I’m here to say, “Nope.” [12:48] Surprising jobs a robot can do: journalism. [12:52] An article in Wired called, “What News-Writing Bots Mean to the Future of Journalism,” leads with “When Republican Steve King beat back Democratic challenger Kim Weaver in the race for Iowa’s 4th District seat in November, the Washington Post snapped into action, covering both the win and the wider electoral trend.” [13:15] “‘Republicans retain the control of the House and lost only a handful of seats from their commanding majority,’ the article read, ‘a stunning reversal of fortune after many GOP leaders feared double-digit losses.’” [13:30] “The dispatch came with the clarity and verve for which Post reporters are known, with one key difference: It was generated by Heliograf, a bot that made its debut on the Post’s website last year and marked the most sophisticated use of artificial intelligence in journalism to date.” [13:52] Any type of writing that is based on data can be replaced with automation and robots. In fact, artificial intelligence is working to take over creative writing, too. Another piece, in the Observer, is called “Will Robots That Can Write Steal Your Creative Job?” [14:12] The author writes, “So, could the machines eventually begin to analyze popular fiction and start to come up with all new narratives that fit our tastes? Indeed, to ever more narrow tastes? We have already seen greater individuation in fiction as the e-book market has made shelf space infinite.” [14:36] “Before e-books took off, novels about werewolves were already a healthy little Fantasy and Science Fiction sub-genre. Since e-books, though, billionaire werewolf romance novels are now a thing.” [14:52] Automation robots will have an incredible impact on medical professions. If a doctor wants an EKG, he can record it on your smartphone app. All of your medical data will be digitized, including X-ray images, CT Scans, and MRIs. [15:12] The Economist produced a special report called “Automation and Anxiety,” which discussed the impact on medicine of deep learning. A product from Enlitic can outperform doctors in reading diagnostic images. [15:27] It’s not just the images are sent to places like India or China to be evaluated by doctors who are paid less but automation and robots are actually doing the work that doctors have always done. [15:40] Jobs are being eliminated in retail at an alarming rate. Retail giants like Sears have shed legacy brands such as Craftsman and Lands End in an effort to survive. Many specialty chains are failing, like Tailored Brands (TLRD), owner of stores like Men’s Wearhouse and Jos. A. Bank. [16:05] Amazon is opening up stores like Amazon Go, where people can do their whole shopping trip without interacting with a single person. As the “Fight for Fifteen” movement works to raise the minimum wage to $15 an hour, one of the unintended consequences will be the deployment of automation and robots. [16:25] I’m already seeing fast food chains rolling out mobile apps and kiosks where you can order your food and never have to speak to a person. [16:35] I’m seeing lots of requests for career middle managers in the retail segment looking for assistance in getting out of the industry. A 2018 study by PWC predicts that nearly 40% of jobs in the U.S. may be vulnerable to replacement by robots in the next 15 years. [16:54] Hopefully, I’ve demonstrated to you that professions that one would have thought would be immune to automation and robots are at risk. Similarly, if the industry where you are working is at risk, you must be on the lookout. [17:10] If you think you are safe from automation and robots sabotaging your career, you must be smoking something! And yes, you are inhaling. [17:22] It is devastating to realize that the career you built — the skills you’ve honed, the seniority you’ve acquired — have all been wiped out because someone built a robot that can do what you do faster and cheaper if not better. [17:38] For many people, these changes have hit like an earthquake or hurricane. They are living in a career disaster area. They will recover but they’re not moving back into the old house. [17:52] Sally was 65 and was a consummate marketing professional. She had worked in a variety of different industries over the span of her career. At different times in her career, she worked freelance and she worked for some major agencies. [18:06] Like many of her peers, she took a hit in the great recession. Then her spouse passed away suddenly and Sally decided to move across the country to be closer to her children. [18:16] Now, she’s trying to re-establish herself in a new city where the culture and job market are very young and vibrant. Sally is taking courses in social media and digital marketing but the skills required to be productive marketing professional have made tectonic shifts in the direction of technology. [18:37] In the 1990s, when I was working in a marketing and sales support function in IBM marketing or in the executive briefing center, we produced presentations and marketing collateral; web content that supported the sale of IBM hardware and software. That world no longer exists. [18:56] The world that does exist today, as I launch the Career Pivot Online Community, requires a completely new set of skills. I’m learning about Facebook marketing, Google Adwords, re-marketing, re-targeting, pixeling strategies, ad networks, and other digital marketing approaches. [19:14] When I made the decision to leave the world of technology marketing, more than 15 years ago, I left a place that looks nothing like it does today. Can Sally shift into this new technological marketing world that’s populated with a very young workforce, at the age of 65? It’s possible but not probable. [19:37] Larry is also 65. He is an engineer who has worked for some of the top companies that designed and manufactured leading computer hardware through his career. He was a program and project manager for huge multinational, multi-company development projects with huge scope and complexity. That world is disappearing, fast. [19:59] Companies like HP, IBM and others have seen their hardware business almost completely disappear. Companies like Sun and DEC have been wiped off the map in a very short period of time. [20:14] There are many like Larry, who built their careers around designing large and ever-growing complex hardware systems. But in the last 10 years, the hardware market has been commoditized. The iPhone sitting next to me has more computing power and function than huge computers of just a few years ago. [20:35] Larry interviewed for a program management job with one of the leading Cloud infrastructure companies. And the first thing they asked him to do was to take a coding test. What?! A coding test? For a program management job? [20:51] Like Larry, I haven’t written a line of code in over 15 years. Could I pass a coding test? Probably not. Does it make sense that they want to see if he can code? Probably not. But that’s not the world we live in, now. [21:06] They moved my cheese. The complex world that Larry excelled and thrived in moved from hardware to software, at Warp speed. They moved Larry’s cheese —  referencing the book Who Moved My Cheese, an amazing way to deal with change in your work and in your life by Dr. Spencer Johnson — and he didn’t even realize it. [21:31] The career space that Larry and his peers lived in for so many years now looks like a career disaster area. Like Sally, he could retool but can he do it fast enough and be accepted in a very young, fast-moving market? It’s possible but not probable. [21:52] It’s now time to shift expectations and direction. People can and do rebuild after a disaster. Sometimes people have to walk away from the disaster scene because it’s just too risky to stay. This is the destruction part. But after a period of grieving all that, it’s time to move away from destruction and get on with creation. [22:15] From here out, there is no safe haven where you can just tuck yourself in and work as long as you want to work. Creative destruction is happening every day and you have to be constantly learning, evolving, and pivoting. How you do that is the subject of the next chapter. [22:35] Action steps: Is your industry in the process of being SMACed? Evaluate where you’re keeping up with changes. Research what skills you need to keep up with your current industry and how much of a challenge will that be? Does it mean going back to school or merely taking online classes? [22:55] Write down how your current skills might be useful in other emerging business types or industries. [23:04] Marc hopes you enjoyed this episode. It is imperative that you learn to embrace creative destruction, as it’s not going away. If anything, it’s going to accelerate. [23:15] The Career Pivot Community website has become a valuable resource for about 50 members who are participating in the Beta phase of this project. Marc is currently recruiting new members for the next cohort. [23:27] If you are interested in the endeavor and would like to be put on the waiting list, please go to CareerPivot.com/Community. When you sign up you’ll receive information about the community as it evolves. [23:43] Those who are in these initial cohorts set the direction of this endeavor. This is a paid membership community with group coaching and special content. More importantly, it’s a community where you can seek help. Go to CareerPivot.com/Community to learn more. [24:07] Marc invites you to connect with him on LinkedIn.com/in/mrmiller. Just include in the connection request that you heard Marc on this podcast. You can look for Career Pivot on Facebook, LinkedIn, or @CareerPivot on Twitter. [24:28] Please come back next week, when Marc will interview Patti Temple Rocks, author of I’m Not Done: It's Time to Talk About Ageism in the Workplace, a great book on ageism. [24:38] Marc thanks you for listening to the Repurpose Your Career podcast. [24:43] You will find the show notes for this episode at CareerPivot.com/episode-123. [24:57] Please hop over to CareerPivot.com and subscribe to get updates on this podcast and all the other happenings at Career Pivot. You can also subscribe to the podcast on iTunes, Stitcher, the Google Podcasts app, Podbean, the Overcast app, or the Spotify app.

Gamer turned AI maven, Kevin Lyman is implementing Jeremy Howard’s vision of harnessing the super chips that power gaming machines (GPUs) to interpret medical images. Deep learning algorithms are already making radiologists faster and better at their work. It’s a story of man plus machine not man versus machine. Don’t miss this accessible report from the frontier where machine is starting to make a tangible difference in medical care. Some highlights from the interview: Kevin Lyman Versions 1.0 & 2.0 Kevin Lyman, Gamer Turned AI Maven Stints at Hasbro, SpaceX & Microsoft The Inventors Guild Jeremy Howard of Kaggle Fame Decides AI Is Ready to Take on Medical Imaging-Founds Enlitic GPUs: Good for Gaming + Processing Medical Images “…deep learning has managed to turn AI into more of an engineering problem than a science problem.” “…you can see most AI setups today as being like a black box.” Kevin Lyman Dropped Everything to Join Jeremy Howard at Enlitic “…they didn't just give us the $10 million in our Series A. They gave us 100 terabytes of historic patient data-“ “…21% faster, they caught 11% more of the true positives and with 9% fewer false positives…” “…we should not be focused on man versus machine… We should be very focused on man plus machine…” AI – Compare & Contrast Kevin Lyman’s Personality Disposes Him to Work in Startups Kevin Lyman 2.0 Interview – Nine Months Later Headway During Nine Months Since 1.0 95% Body Coverage About to Deploy Beta with Capitol Health

“Every decision needs to be made with the mindset of how will this impact patient down the line”

AI is the buzzword startups are very keen on using when describing their products. For decades, movies are full of ideas on what artificial intelligence could do in a positive and negative way. What is AI, deep learning or a simple algorithm? What is the dream and what current reality around AI? How does AI look in practice? In this episode, you will hear from Sally Daub - the CEO of Enlitic talk about the market potential of AI, the current state of the market and more. Enlitic is a San Francisco based startup using deep learning to distill actionable insights from billions of clinical cases and help doctors leverage the collective intelligence of the medical community. At the moment, the use of AI is highest in the field of medical imaging and diagnostics, drug discovery and therapy planning, but Accenture predicts that by 2026 150 billion US dollars could be saved annually due to applications to robot-assisted surgery, virtual nursing assistants, followed by administrative workflow assistance, fraud detection and dosage error reduction, to name the first few areas with most significant savings.

Imagine if we could learn from the experience of every doctor in the world and the data collected from billions of clinical cases. The San Francisco-based company, Enlitic, is using deep learning techniques to harness the collective intelligence of the global medical community. The goal is to make doctors faster, more accurate and efficient at treating diseases. Kevin Lyman, Enlitic's Chief Operating Officer and lead scientist, graduated from the private research university, Rensselaer Polytechnic Institute (RPI) with a BS in Computer Science. He has worked for SpaceX, where he developed sensors and control circuits for the Falcon 9 rocket and Dragon space capsule, and Microsoft, where he designed new features for Excel and developed a real-time error monitoring system for Office Online. But now his focus is on healthcare and optimization of the human body’s potential. In this in-depth interview, recorded at TEDMED 2017, Kevin explains his devotion to self-experimentation and diagnostics. He also reveals details of his data-driven quest to lose a huge amount of body weight by applying the principles of engineering. ------- "The goal is not to replace doctors. The goal is to augment them. We like to say that it's not about man versus machine. It's about man plus machine." Kevin Lyman (@ktlyman) is the Chief Operating Officer and lead scientist at Enlitic (@enlitic) Kevin attended Rensselaer Polytechnic Institute RPI) in Upstate New York where he changed majors eight times before finally graduating in Computer Science. Kevin Lyman spoke to LLAMA host Peter Bowes at the 2017 TEDMED conference in La Quinta, near Palm Springs, California. His TED talk, as a ‘Hive innovator' will be published on the TEDMED site in the next few months. "My goal personally is really to advance technology in a meaningful way that will impact a large number of lives.”

Our guest in this episode is Kevin Lyman, an extremely talented entrepreneur and computer scientist based out in San Francisco. He’s worked at companies including Microsoft and SpaceX, and he’s even founded several companies as well. Now he’s applying his ability to create leading products to the healthcare sector- and he’s the COO and Lead Scientist for Enlitic, a startup that develops deep learning products to make doctors faster and more accurate. He’s going to help us break down some of the details and capabilities of Artificial Intelligence, and explain the differences between machine learning and deep learning. Don’t worry - you don’t have to be a computer scientist to understand it, but with so many terms floating around for a set of technologies that is on track to be the most significant development of our lifetime - we need to peel back the curtain a little bit to understand them a bit better. Then we get into how all this is making an impact on healthcare, and how people and organizations are collaborating to create and implement the solutions. Be sure to get all the show notes, videos, and links at our website, visit digitalhealthtoday.com/38, while youre there, please take a second to subscribe to the podcast. You can do that right from your phone, just visit the website, look for your preferred platform on the podcast page, and select the Apple, Soundcloud or Stitcher icons, whatever you prefer, and you can subscribe to this podcast right there.   Learn more about your ad choices. Visit megaphone.fm/adchoices

The Institute of Medicine estimates that diagnostic errors affect 12 million Americans every year. More accurate and efficient tools for doctors to better handle and analyze data could greatly reduce that number. Every time a doctor sees a patient, they are solving a complex data problem. The goal of each case is to arrive at an optimal diagnostic decision based on many forms of clinical data. Let’s listen as Kevin Lyman from Enlitic talks about how his history as a championship gamer and toy designer has led him to using deep learning to find medical insights from billions of clinical cases that will help doctors handle patient data more efficiently and successfully. Ultimately bringing better care and outcomes to millions of patients.

My guest this time is Diogo Almeida, senior data scientist at healthcare startup Enlitic. Diogo and I met at the O'Reilly AI conference, where he delivered a great presentation on in-the-trenches deep learning titled “Deep Learning: Modular in theory, inflexible in practice,” which we discuss in this interview. Diogo is also a past 1st place Kaggle competition winner, and we spend some time discussing the competition he competed in and the approach he took as well. The notes for this show can be found at twimlai.com/talk/8.

En este programa hablamos de la posibilidad de ofrecer consultoría privada para los oyentes de Hablando de Tecnología. También hablamos de algunas propiedades adicionales en la Internet que vamos a eliminar. Y nuestra primera noticia tiene que ver con “Big Data”. Entérate cómo una compañía de California llamada “Enlitic” se propone agregar cantidades masivas de información médica y utilizarlas para realizar diagnósticos más acertados. Además, siguiendo con el tema de la detección, hablamos de un sistema de detección temprana de terremotos, creado por la universidad de UC Berkeley, que funciona pero carece de fondos para implantarlo. También en California, el gobernador de ese estado Jerry Brown firmó la ley llamada “kill switch” que permite que un usuario inactive su teléfono remotamente en caso de que se lo roben. Y por último hablamos del programa de Apple para reemplazar la batería en algunos iPhone 5 que están dando problemas. ENLACES: Los Puertorriqueños... según Gabriel García Márquez ¿Podrá el “Big Data” ayudar a diagnosticar condiciones médicas? Sistema de detección temprana de terremotos funciona pero no tiene fondos Estatua de Colón descubre su hogar en Arecibo Gobernador de california firma la ley del “kilo switch” Apple reemplazará la batería en algunos iPhone 5 Programa de Reemplazo de Baterías de Apple [sc:FirmaOrlandoMergal2014 ]