The following is a conversation with Jeremy Howard.
He's the founder of Fast AI, a research institute dedicated
to making deep learning more accessible.
He's also a distinguished research scientist
at the University of San Francisco,
a former president of Kegel, as well as
a top-ranking competitor there.
And in general, he's a successful entrepreneur,
educator, researcher, and an inspiring personality
in the AI community.
When someone asked me, how do I get started
with deep learning, Fast AI is one of the top places
I point them to.
It's free.
It's easy to get started.
It's insightful and accessible.
And if I may say so, it has very little BS.
It can sometimes dilute the value of educational content
on popular topics like deep learning.
Fast AI has a focus on practical application
of deep learning and hands-on exploration
of the cutting edge that is incredibly
both accessible to beginners and useful to experts.
This is the Artificial Intelligence Podcast.
If you enjoy it, subscribe on YouTube,
give it five stars on iTunes, support it on Patreon,
or simply connect with me on Twitter.
Alex Friedman, spelled F-R-I-D-M-A-N.
And now, here's my conversation with Jeremy Howard.
What's the first program you ever written?
First program I wrote that I remember
would be at high school.
I did an assignment where I decided
to try to find out if there were some better musical scales
than the normal 12-tone, 12-interval scale.
So I wrote a program on my Commodore 64 in BASIC
that searched through other scale sizes
to see if it could find one where there were
more accurate harmonies.
Like mid-tone?
Like you want an actual exactly three-to-two ratio,
whereas with a 12-interval scale,
it's not exactly three-to-two, for example.
So that's well-tempered, as they say.
And BASIC on a Commodore 64.
Where was the interest in music from?
Or is it just technical?
I did music all my life, so I played saxophone and clarinet
and piano and guitar and drums and whatever.
How does that thread go through your life?
Where's music today?
It's not where I wish it was.
For various reasons, couldn't really keep it going,
particularly because I had a lot of problems with RSI,
with my fingers.
And so I had to cut back anything that used hands
and fingers.
I hope one day I'll be able to get back to it health-wise.
So there's a love for music underlying it all.
Sure, yeah.
What's your favorite instrument?
Saxophone.
Baritone saxophone.
Well, probably bass saxophone, but they're awkward.
Well, I always love it when music is
coupled with programming.
There's something about a brain that
utilizes those that emerges with creative ideas.
So you've used and studied quite a few programming languages.
Can you give an overview of what you've used?
What are the pros and cons of each?
Well, my favorite programming environment almost certainly
was Microsoft Access back in the earliest days.
So that was a special basic for applications, which
is not a good programming language,
but the programming environment is fantastic.
It's like the ability to create user interfaces and tied data
and actions to them and create reports and all that,
as I've never seen anything as good.
So things nowadays like Airtable,
which are like small subsets of that, which people love
for good reason, but unfortunately, nobody's ever
achieved anything like that.
What is that if you could pause on that for a second?
Oh, Access.
Access, is it a fundamentally database?
It was a database program that Microsoft produced,
part of Office, and it kind of withered.
But basically, it lets you in a totally graphical way
create tables and relationships and queries
and tie them to forms and set up event handlers and calculations.
And it was a very complete, powerful system designed
for not massive scalable things, but for useful little applications
that I loved.
So what's the connection between Excel and Access?
So very close.
So Access was the relational database equivalent,
if you like.
So people still do a lot of that stuff that
should be in Access in Excel because they know it.
Excel's great as well.
But it's just not as rich a programming model as VBA
combined with a relational database.
And so I've always loved relational databases.
But today, programming on top of relational databases
is just a lot more of a headache.
You generally either need to kind of you
need something that connects, that runs on a database server
unless you use SQLite, which has its own issues.
Then you kind of often, if you want
to get a nice programming model, you
need to create an ORM on top.
And then I don't know, there's all these pieces tied together.
And it's just a lot more awkward than it should be.
There are people that are trying to make it easier,
so that in particular, I think of F-sharp, you know,
Don Syme, who him and his team have done a great job of making
something like a database appear in the type system
so you actually get like tab completion for fields and tables
and stuff like that.
Anyway, so that was kind of anyway.
So like that whole VBA office thing, I guess,
was a starting point, which is your miss.
And I got into standard visual basic, which
that's interesting just to pause on that for a second.
And it's interesting that you're connecting programming
languages to the ease of management of data.
So in your use of programming languages,
you always had a love and a connection with data.
I've always been interested in doing useful things for myself
and for others, which generally means getting some data
and doing something with it and putting it out there again.
So that's been my interest throughout.
So I also did a lot of stuff with Apple script
back in the early days.
So it's kind of nice being able to get the computer
and computers to talk to each other and to do things for you.
And then I think that one night, the programming language
I most loved then would have been Delphi, which
was Object Pascal created by Anders Halsberg, who previously
did Turbo Pascal and then went on to create.net
and then went on to create TypeScript.
Delphi was amazing because it was like a compiled, fast language
that was as easy to use as Visual Basic.
Delphi, what is it similar to in more modern languages?
Visual Basic.
Visual Basic.
Yeah, but a compiled, fast version.
So I'm not sure there's anything quite like it anymore.
If you took C-Sharp or Java and got rid of the virtual machine
and replaced it with something, you could compile a small type
binary.
I feel like it's where Swift could get to with the new Swift
UI and the cross-platform development going on.
That's one of my dreams is that we'll hopefully get back
to where Delphi was.
There is actually a free Pascal project nowadays
called Lazarus, which is also attempting
to recreate Delphi.
They're making good progress.
So OK, Delphi, that's one of your favorite programming languages.
Well, it's programming environments.
Again, say Pascal's not a nice language.
If you wanted to know specifically
about what languages I like, I would definitely
pick Jay as being an amazingly wonderful language.
What's Jay?
Jay, are you aware of APL?
I am not, except from doing a little research on the work
you've done.
OK, so not at all surprising you're not
familiar with it, because it's not well-known.
But it's actually one of the main families
of programming languages going back
to the late 50s, early 60s.
So there was a couple of major directions.
One was the kind of lambda-calculus-salonzo-church
direction, which I guess kind of Lisbon scheme
and whatever, which has a history going back
to the early days of computing.
The second was the kind of imperative slash
OO, algo, similar, going on to C, C++, so forth.
There was a third, which are called array-oriented languages,
which started with a paper by a guy called Ken Iverson, which
was actually a math theory paper, not a programming paper.
It was called Notation as a Tool for Thought.
And it was the development of a new type of math notation.
And the idea is that this math notation was much more
flexible, expressive, and also well-defined.
Then traditional math notation, which
is none of those things, math notation is awful.
And so he actually turned that into a programming language.
Because this was the late 50s, all the names were available.
So he called his language a programming language, or APL.
APL, what?
So APL is a implementation of notation
as a tool for thought, by which he means math notation.
And Ken and his son went on to do many things,
but eventually they actually produced a new language that
was built on top of all the learnings of APL.
And that was called J. And J is the most expressive,
composable, beautifully designed language I've ever seen.
Does it have object-oriented components?
Does it have that kind of thing?
Or is it more like?
Not really.
It's an array-oriented language.
It's the third path.
Are you saying array?
Array-oriented, yeah.
Well, it needs to be array-oriented.
So array-oriented means that you generally
don't use any loops, but the whole thing
is done with kind of an extreme version of broadcasting,
if you're familiar with that NumPy slash Python concept.
So you do a lot with one line of code.
It looks a lot like math notation, highly compact.
And the idea is that because you can do so much
with one line of code, a single screen of code
is very unlikely to, you very rarely
need more than that to express your program.
And so you can kind of keep it all in your head,
and you can kind of clearly communicate it.
It's interesting that APL created two main branches, K and J.
J is this kind of like open source niche community
of crazy enthusiasts like me.
And then the other path, K, was fascinating.
It's an astonishingly expensive programming language, which
many of the world's most ludicrously rich hedge funds
use.
So the entire K machine is so small,
it sits inside level three cache on your CPU,
and it easily wins every benchmark I've ever seen
in terms of data processing speed.
But you don't come across it very much,
because it's like $100,000 per CPU to run it.
But it's like this path of programming languages
is just so much, I don't know, so much more powerful
in every way than the ones that almost anybody uses every day.
So it's all about computation.
It's really focusing on computation.
It's pretty heavily focused on computation.
I mean, so much of programming is data processing
by definition.
So there's a lot of things you can do with it.
But yeah, there's not much work being done
on making user interface toolkits or whatever.
I mean, there's some, but they're not great.
At the same time, you've done a lot of stuff
with Perl and Python.
So where does that fit into the picture of J and K and APL?
Well, it's just much more pragmatic.
In the end, you kind of have to end up where the libraries are.
Because to me, my focus is on productivity.
I just want to get stuff done and solve problems.
So Perl was great.
I created an email company called Fastmail.
And Perl was great because back in the late 90s, early 2000s,
it just had a lot of stuff it could do.
I still had to write my own monitoring system
and my own web framework and my own whatever.
Because none of that stuff existed.
But it was a super flexible language to do that in.
And you used Perl for Fastmail.
You used it as a back end.
So everything was written in Perl?
Yeah.
Yeah, everything was Perl.
Why do you think Perl hasn't succeeded or hasn't dominated
the market where Python really takes over a lot of this?
Well, I mean, Perl did dominate.
It was everything everywhere.
But then the guy that ran Perl, Larry Wool,
just didn't put the time in anymore.
And no project can be successful if there isn't.
Particularly one that started with a strong leader
that loses that strong leadership.
So then Python has kind of replaced it.
Python is a lot less elegant language in nearly every way.
But it has the data science libraries.
And a lot of them are pretty great.
So I kind of use it because it's the best we have.
But it's definitely not good enough.
What do you think the future of programming looks like?
What do you hope the future of programming looks like
if we zoom in on the computational fields
on data science and machine learning?
I hope Swift is successful because the goal of Swift,
the way Chris Latner describes it,
is to be infinitely hackable.
And that's what I want.
I want something where me and the people I do research with
and my students can look at and change everything
from top to bottom.
There's nothing mysterious and magical and inaccessible.
Unfortunately, with Python, it's the opposite of that
because Python is so slow.
It's extremely unhackable.
You get to a point where it's like, OK, from here on down at C.
So your debugger doesn't work in the same way.
Your profiler doesn't work in the same way.
Your build system doesn't work in the same way.
It's really not very hackable at all.
What's the part you like to be hackable?
Is it for the objective of optimizing training
of neural networks, inference of neural networks?
Is it performance of the system?
Or is there some non-performance related?
Just creative idea.
It's everything.
In the end, I want to be productive as a practitioner.
So at the moment, our understanding of deep learning
is incredibly primitive.
There's very little we understand.
Most things don't work very well, even
though it works better than anything else out there.
There's so many opportunities to make it better.
So you look at any domain area like, I don't know,
speech recognition with deep learning
or natural language processing classification
with deep learning or whatever.
Every time I look at an area with deep learning,
I always see like, oh, it's terrible.
There's lots and lots of obviously stupid ways
to do things that need to be fixed.
So then I want to be able to jump in there and quickly
experiment and make them better.
You think the programming language has a role in that?
Huge role.
Yeah, so currently, Python has a big gap
in terms of our ability to innovate particularly
around recurrent neural networks and natural language
processing because it's so slow.
The actual loop where we actually loop through words,
we have to do that whole thing in CUDA C.
So we actually can't innovate with the kernel, the heart,
of that most important algorithm.
And it's just a huge problem.
And this happens all over the place.
So we hit research limitations.
Another example, convolutional neural networks, which
are actually the most popular architecture for lots of things,
maybe most things in deep learning.
We almost certainly should be using
sparse convolutional neural networks.
But only like two people are because to do it,
we have to rewrite all of that CUDA C level stuff.
And yeah, just researchers and practitioners don't.
So there's just big gaps in what people actually research on,
what people actually implement because of the programming
language problem.
So you think it's just too difficult to write in CUDA C
that a higher level programming language like Swift
should enable the easier, fooling around,
creative stuff with RNNs, or with sparse convolutional
neural networks, who's at fault?
Who's at charge of making it easy for a researcher to play around?
I mean, no one's at fault.
Just nobody's got around to it yet.
Or it's just, it's hard, right?
And I mean, part of the fault is that we ignored that whole APL
kind of direction, or nearly everybody did for 60 years,
50 years.
But recently, people have been starting
to reinvent pieces of that and kind of create
some interesting new directions in the compiler technology.
So the place where that's particularly happening right now
is something called MLIR, which is something that, again,
Chris Lattner, the Swift guy, is leading.
And yeah, because it's actually not
going to be Swift on its own that solves this problem.
Because the problem is that currently writing
a acceptably fast GPU program is too complicated,
regardless of what language you use.
And that's just because if you have
to deal with the fact that I've got 10,000 threads
and I have to synchronize between them all
and I have to put my thing into grid blocks
and think about warps and all this stuff,
it's just so much boilerplate that to do that well,
you have to be a specialist at that,
and it's going to be a year's work to optimize that algorithm
in that way.
But with things like tensor comprehensions and tile
and MLIR and TVM, there's all these various projects which
are all about saying, let's let people
create domain-specific languages for tensor computations.
These are the kinds of things we do generally
on the GPU for deep learning and then have a compiler which
can optimize that tensor computation.
A lot of this work is actually sitting
on top of a project called Halide, which
is a mind-blowing project where they came up
with such a domain-specific language.
In fact, two, one domain-specific language
for expressing this is what my tensor computation is.
And another domain-specific language for expressing
this is the way I want you to structure
the compilation of that, like do it block by block
and do these bits in parallel.
And they were able to show how you can compress the amount
of code by 10x compared to optimized GPU code
and get the same performance.
So these are the things that are sitting on top
of that kind of research.
And MLIR is pulling a lot of those best practices together.
And now we're starting to see work done
on making all of that directly accessible through Swift
so that I could use Swift to write those domain-specific
languages.
And hopefully we'll get then Swift Kuda kernels written
in a very expressive and concise way that looks a bit like J
in APL and then Swift layers on top of that
and then a Swift UI on top of that.
And it'll be so nice if we can get to that point.
Now, does it all eventually boil down to Kuda and NVIDIA GPUs?
Unfortunately, at the moment, it does.
But one of the nice things about MLIR,
if AMD ever gets their act together, which they probably
want, is that they or others could write MLIR backends
for other GPUs or other tensor computation devices, of which
today there are increasing number like Graphcore
or Vertex AI or whatever.
So yeah, being able to target lots of backends
would be another benefit of this.
And the market really needs competition
because at the moment NVIDIA is massively
overcharging for their kind of enterprise class cards
because there is no serious competition
because nobody else is doing the software properly.
In the cloud, there is some competition, right?
But not really.
Other than TPUs, perhaps, but TPUs
are almost unprogrammable at the moment.
The TPUs have the same problem that you can't.
It's even worse.
So TPUs, Google actually made an explicit decision
to make them almost entirely unprogrammable
because they felt that there was too much IP in there.
And if they gave people direct access to program them,
people would learn their secrets.
So you can't actually directly program
the memory in a TPU.
You can't even directly create code that runs on
and that you look at on the machine that has the TPU.
It all goes through a virtual machine.
So all you can really do is this kind of cookie cutter
thing of like plug-in high-level stuff
together, which is just super tedious and annoying
and totally unnecessary.
So tell me if you could the origin story of fast AI.
What is the motivation, its mission, its dream?
So I guess the founding story is heavily
tied to my previous startup, which
is a company called Inletic, which
was the first company to focus on deep learning for medicine.
And I created that because I saw there
was a huge opportunity to, there's
about a 10x shortage of the number of doctors in the world
and the developing world that we need.
I expected it would take about 300 years to train enough doctors
to meet that gap.
But I guessed that maybe if we used deep learning
for some of the analytics, we could maybe make it
so you don't need as highly trained doctors.
For diagnosis.
For diagnosis and treatment planning.
Where's the biggest benefit just before we get to fast AI?
Where's the biggest benefit of AI in medicine that you see today?
Not much happening today in terms of stuff that's actually
out there.
It's very early.
But in terms of the opportunity, it's
to take markets like India and China and Indonesia, which
have big populations, Africa, small numbers of doctors,
and provide diagnostic, particularly treatment
planning and triage kind of on device
so that if you do a test for malaria or tuberculosis
or whatever, you immediately get something
that even a health care worker that's
had a month of training can get a very high quality assessment
of whether the patient might be at risk until OK,
we'll send them off to a hospital.
So for example, in Africa, outside of South Africa,
there's only five pediatric radiologists
for the entire continent.
So most countries don't have any.
So if your kid is sick and they need something
diagnosed with medical imaging, the person,
even if you're able to get medical imaging done,
the person that looks at it will be a nurse at best.
But actually, in India, for example, and China,
almost no x-rays are read by anybody,
by any trained professional because they don't have enough.
So if instead we had an algorithm that
could take the most likely high risk 5% and say triage,
basically say, OK, somebody needs to look at this,
it would massively change the kind of way
that what's possible with medicine in the developing world.
And remember, increasingly, they have money.
They're the developing world.
They're not the poor world.
They're the developing world.
So they have the money.
So they're building the hospitals.
They're getting the diagnostic equipment.
But there's no way for a very long time
will they be able to have the expertise.
Shortage of expertise.
OK, and that's where the deep learning systems can step in
and magnify the expertise they do have, essentially.
Exactly.
Yeah.
So you do see, just to linger a little bit longer,
the interaction, do you still see the human experts still
at the core of these systems?
Yeah, absolutely.
Is there something in medicine that can be automated
almost completely?
I don't see the point of even thinking about that,
because we have such a shortage of people.
Why would we not?
Why would we want to find a way not to use them?
Like, we have people.
So the idea of like, even from an economic point of view,
if you can make them 10x more productive,
getting rid of the person doesn't
impact your unit economics at all.
And it totally involves the fact that there
are things people do better than machines.
So it's just, to me, that's not a useful way
of framing the problem.
I guess, just to clarify, I guess
I meant there may be some problems
where you can avoid even going to the expert ever.
Sort of maybe preventative care or some basic stuff,
the low hanging fruit, allowing the expert to focus
on the things that are really that.
Well, that's what the triage would do, right?
So the triage would say, OK, 99% sure there's nothing here.
So that can be done on device.
And they can just say, OK, go home.
So the experts are being used to look at the stuff which
has some chance it's worth looking at, which most things
is not, you know, it's fine.
Why do you think we haven't quite made progress on that yet
in terms of the scale of how much AI is applied in the method?
There's a lot of reasons.
I mean, one is it's pretty new.
I only started in Liddick in 2014.
And before that, it's hard to express
to what degree the medical world was not
aware of the opportunities here.
So I went to RSNA, which is the world's largest radiology
conference.
And I told everybody I could, you know,
like, I'm doing this thing with deep learning.
Please come and check it out.
And no one had any idea what I was talking about.
And no one had any interest in it.
So like we've come from absolute zero, which is hard.
And then the whole regulatory framework, education system,
everything is just set up to think of doctoring
in a very different way.
So today, there is a small number of people
who are deep learning practitioners and doctors
at the same time.
And we started to see the first ones come out
of their PhD programs.
So Zach Cahane over in Boston, Cambridge
has a number of students now who are data science experts,
deep learning experts, and actual medical doctors.
Quite a few doctors have completed our fast AI course
now and are publishing papers and creating journal reading
groups in the American Council of Radiology.
And it's just starting to happen.
But it's going to be a long process.
The regulators have to learn how to regulate this.
They have to build guidelines.
And then the lawyers at hospitals
have to develop a new way of understanding
that sometimes it makes sense for data
to be looked at in raw form, in large quantities,
in order to create world-changing results.
Yeah, there's a regulation around data,
all that, it sounds probably the hardest problem,
but sounds reminiscent of autonomous vehicles as well.
Many of the same regulatory challenges,
many of the same data challenges.
Yeah, I mean, funnily enough, the problem
is less the regulation and more the interpretation
of that regulation by lawyers in hospitals.
So HIPAA was actually was designed.
The P in HIPAA does not stand for privacy.
It stands for portability.
It's actually meant to be a way that data can be used.
And it was created with lots of gray areas,
because the idea is that would be more practical
and it would help people to use this legislation
to actually share data in a more thoughtful way.
Unfortunately, it's done the opposite,
because when a lawyer sees a gray area,
they see, oh, if we don't know we won't get sued,
then we can't do it.
So HIPAA is not exactly the problem.
The problem is more that there's hospital lawyers
are not incented to make bold decisions
about data portability.
Or even to embrace technology that saves lives.
They more want to not get in trouble
for embracing that technology.
Also, it is also saves lives in a very abstract way,
which is like, oh, we've been able to release
these 100,000 anonymous records.
I can't point at the specific person whose life that's saved.
I can say, oh, we've ended up with this paper,
which found this result, which diagnosed 1,000 more people
than we would have otherwise, but which ones were helped.
It's very abstract.
And on the counter side of that,
you may be able to point to a life that was taken
because of something that was...
Yeah, or a person whose privacy was violated.
It's like, oh, this specific person,
there was de-identified, identified.
Just a fascinating topic.
We're jumping around.
We'll get back to fast AI.
But on the question of privacy, data
is the fuel for so much innovation in deep learning.
What's your sense on privacy?
Whether we're talking about Twitter, Facebook, YouTube,
just the technologies like in the medical field
that rely on people's data in order to create impact.
How do we get that right, respecting people's privacy
and yet creating technology that is learned from data?
One of my areas of focus is on doing more with less data,
which so most vendors, unfortunately, are strongly
and centered to find ways to require more data
and more computation.
So Google and IBM being the most obvious...
IBM.
Yeah, so Watson.
So Google and IBM both strongly push the idea
that they have more data and more computation
and more intelligent people than anybody else.
And so you have to trust them to do things
because nobody else can do it.
And Google's very upfront about this,
like Jeff Dain has gone out there and given talks
and said our goal is to require a thousand times
more computation, but less people.
Our goal is to use the people that you have better
and the data you have better and the computation you have better.
So one of the things that we've discovered is,
or at least highlighted, is that you very, very, very often
don't need much data at all.
And so the data you already have in your organization
will be enough to get state-of-the-art results.
So my starting point would be to say around privacy
is a lot of people are looking for ways
to share data and aggregate data,
but I think often that's unnecessary.
They assume that they need more data than they do
because they're not familiar with the basics of transfer
learning, which is this critical technique
for needing orders of magnitude less data.
Is your sense, one reason you might want to collect data
from everyone is, like in the recommender system context,
where your individual, Jeremy Howard's individual data
is the most useful for providing a product that's
impactful for you, so for giving you advertisements,
for recommending to you movies, for doing medical diagnosis.
Is your sense we can build with a small amount of data,
general models, that will have a huge impact for most people,
that we don't need to have data from each individual?
On the whole, I'd say yes.
I mean, there are things like recommender systems
have this cold start problem, where Jeremy is a new customer.
We haven't seen him before, so we can't recommend him things
based on what else he's bought and liked with us.
And there's various workarounds to that.
A lot of music programs will start out
by saying, which of these artists do you like?
Which of these albums do you like?
Which of these songs do you like?
Netflix used to do that.
Nowadays, people don't like that, because they think, oh,
we don't want to bother the user.
So you could work around that by having some kind of data
sharing, where you get my marketing record from Axiom
or whatever, and try to question that.
To me, the benefit to me and to society
of saving me five minutes on answering some questions
versus the negative externalities of the privacy issue
doesn't add up.
So I think a lot of the time, the places
where people are invading our privacy in order
to provide convenience is really about just trying
to make them more money.
And they move these negative externalities
into places that they don't have to pay for them.
So when you actually see regulations
appear that actually cause the companies that
create these negative externalities to have to pay for it
themselves, they say, well, we can't do it anymore.
So the cost is actually too high.
But for something like medicine, the hospital
has my medical imaging, my pathology studies,
my medical records.
And also, I own my medical data.
So I help a startup called DocAI.
One of the things DocAI does is that it has an app.
You can connect to Sutter Health and Labcore and Walgreens
and download your medical data to your phone
and then upload it, again, at your discretion
to share it as you wish.
So with that kind of approach, we
can share our medical information
with the people we want to.
Yeah, so control.
I mean, really being able to control
who you share it with and so on.
So that has a beautiful, interesting tangent
to return back to the origin story of FastAI.
Right, so before I started FastAI,
I spent a year researching where are the biggest
opportunities for deep learning.
Because I knew from my time at Kaggle in particular
that deep learning had kind of hit this threshold point where
it was rapidly becoming the state of the art
approach in every area that looked at it.
And I'd been working with neural nets for over 20 years.
I knew that from a theoretical point of view,
once it hit that point, it would do that in kind of just
about every domain.
And so I spent a year researching
what are the domains it's going to have the biggest low hanging
fruit in the shortest time period.
I picked medicine, but there were so many I could have picked.
And so there was a kind of level of frustration for me of like,
OK, I'm really glad we've opened up the medical deep learning
world and today it's huge, as you know.
But I can't do everything.
I don't even know like, like in medicine,
it took me a really long time to even get a sense of like,
what kind of problems do medical practitioners solve?
What kind of data do they have?
Who has that data?
So I kind of felt like I need to approach this differently
if I want to maximize the positive impact of deep learning.
Rather than me picking an area and trying
to become good at it and building something,
I should let people who are already domain experts
in those areas and who already have the data do it themselves.
So that was the reason for vast AI
is to basically try and figure out
how to get deep learning into the hands of people
who could benefit from it and help them to do so
in as quick and easy and effective a way as possible.
Got it.
So sort of empower the domain experts.
Yeah.
And like partly it's because like,
unlike most people in this field,
my background is very applied and industrial.
Like my first job was at McKinsey and Company.
I spent 10 years in management consulting.
I spend a lot of time with domain experts.
So I kind of respect them and appreciate them.
And though I know that's where the value generation in society
is.
And so I also know how most of them can't code.
And most of them don't have the time
to invest three years in a graduate degree or whatever.
So it's like, how do I upskill those domain experts?
I think that would be a super powerful thing.
Biggest societal impact I could have.
So yeah, that was the thinking.
So so much of vast AI students and researchers
and the things you teach are pragmatically minded,
practically minded, figuring out ways
how to solve real problems and fast.
So from your experience, what's the difference between theory
and practice of deep learning?
Well, most of the research in the deep mining world
is a total waste of time.
Right.
That's what I was getting at.
Yeah, it's a problem in science in general.
Scientists need to be published, which
means they need to work on things
that their peers are extremely familiar with
and can recognize in advance in that area.
So that means that they all need to work on the same thing.
And so it really ink.
And the thing they work on is nothing
to encourage them to work on things
that are practically useful.
So you get just a whole lot of research, which
is minor advances and stuff that's
been very highly studied and has no significant practical
impact, whereas the things that really make a difference,
like I mentioned, transfer learning.
If we can do better at transfer learning,
then it's this world-changing thing
where suddenly lots more people can do world-class work
with less resources and less data.
But almost nobody works on that.
Or another example, active learning,
which is the study of how do we get more out
of the human beings in the loop?
That's my favorite topic.
Yeah, so active learning is great,
but it's almost nobody working on it
because it's just not a trendy thing right now.
You know what somebody started to interrupt?
He was saying that nobody is publishing on active learning.
But there's people inside companies,
anybody who actually has to solve a problem,
they're going to innovate on active learning.
Yeah, everybody kind of reinvents active learning
when they actually have to work in practice because they
start labeling things and they think, gosh,
this is taking a long time and it's very expensive.
And then they start thinking, well,
why am I labeling everything?
I'm only the machine's only making mistakes
on those two classes.
They're the hard ones.
Maybe I'll just start labeling those two classes.
And then you start thinking, well, why did I do that manually?
Why can't I just get the system to tell me
which things are going to be hardest?
It's an obvious thing to do, but yeah,
it's just like transfer learning.
It's understudied and the academic world just
has no reason to care about practical results.
The funny thing is, I've only really ever written one paper.
I hate writing papers.
And I didn't even write it.
It was my colleague, Sebastian Ruder, who actually wrote it.
I just did the research for it.
But it was basically introducing successful transfer
learning to NLP for the first time, and the algorithm is
called ULMfit.
And I actually wrote it for the course.
For the first day of course, I wanted to teach people NLP.
And I thought, I only want to teach people practical stuff.
And I think the only practical stuff is transfer learning.
And I couldn't find any examples of transfer learning in NLP.
So I just did it.
And I was shocked to find that as soon as I did it,
it was the basic prototype took a couple of days.
It smashed the state of the art on one
of the most important data sets in a field
that I knew nothing about.
And I just thought, well, this is ridiculous.
And so I spoke to Sebastian about it,
and he kindly offered to write it up the results.
And so it ended up being published in ACL, which
is the top computational linguistics conference.
So people do actually care once you do it.
But I guess it's difficult for maybe junior researchers
or like, I don't care whether I get citations or papers
or whatever.
There's nothing in my life that makes
that important, which is why I've never actually
bothered to write a paper myself.
But for people who do, I guess they
have to pick the kind of safe option, which
is like, yeah, make a slight improvement on something
that everybody's already working on.
Yeah, nobody does anything interesting or succeeds in life
with the safe option.
Although, I mean, the nice thing is nowadays,
everybody is now working on NLP transfer learning.
Because since that time, we've had GPT and GPT2 and BERT.
And it's like, so yeah, once you show that something's
possible, everybody jumps in, I guess.
I hope to be a part of it.
I hope to see more innovation and active learning
in the same way.
I think transfer learning and active learning
are fascinating, public, open work.
I actually helped start a startup called Platform AI, which
is really all about active learning.
And yeah, it's been interesting trying
to kind of see what research is out there
and make the most of it.
And there's basically none, so we've
had to do all our own research.
Once again, and just as you described,
can you tell the story of the Stanford competition,
Dawn Bench, and fast AI's achievement on it?
Sure.
So something which I really enjoy is that I basically
teach two courses a year, the practical deep learning
for coders, which is kind of the introductory course,
and then cutting edge deep learning for coders, which
is the kind of research level course.
And while I teach those courses, I basically
have a big office at the University of San Francisco.
It'd be enough for like 30 people.
And I invite any student who wants to come and hang out
with me while I build the course.
And so generally, it's full.
And so we have 20 or 30 people in a big office
with nothing to do but study deep learning.
So it was during one of these times
that somebody in the group said, oh, there's
a thing called Dawn Bench that looks interesting.
And I say, what the hell is that?
I'm going to set out some competition
to see how quickly you can train a model.
Seems kind of not exactly relevant to what we're doing,
but it sounds like the kind of thing
which you might be interested in.
And I checked it out and I was like, oh, crap,
there's only 10 days till it's over.
It's pretty much too late.
And we're kind of busy trying to teach this course.
But we're like, oh, it would make an interesting case study
for the course like it's all the stuff we're already doing.
Why don't we just put together our current best practices
and ideas.
So me and I guess about four students just decided
to give it a go.
And we focused on this small one called
Syfar 10, which is little 32 by 32 pixel images.
Can you say what Dawn Bench is?
Yeah, so it's a competition to train a model as fast as possible.
It was run by Stanford.
And as cheap as possible, too.
That's also another one for as cheap as possible.
And there's a couple of categories, ImageNet and Syfar 10.
So ImageNet's this big 1.3 million image
thing that took a couple of days to train.
I remember a friend of mine, Pete Warden, who's now at Google.
I remember he told me how he trained ImageNet a few years
ago when he basically had this little granny flat out
the back that he turned into was ImageNet training center.
And after a year of work, he figured out
how to train it in 10 days or something.
That was a big job.
Well, Syfar 10, at that time, you could train in a few hours.
It's much smaller and easier.
So we thought we'd try Syfar 10.
And yeah, I've really never done that before.
Like, things like using more than one GPU at a time
was something I tried to avoid.
Because to me, it's very against the whole idea of accessibility.
Is she better do things with one GPU?
I mean, have you asked in the past
before, after having accomplished something,
how do I do this faster, much faster?
Oh, always.
But it's always, for me, it's always how do I make it much faster
on a single GPU that a normal person could
afford in their day-to-day life?
It's not how could I do it faster by having a huge data center?
Because to me, it's all about, like,
as many people should be able to use something as possible
without fussing around with infrastructure.
So anyway, so in this case, it's like, well,
we can use 8GPUs just by renting a AWS machine.
So we thought we'd try that.
And yeah, basically, using the stuff we were already doing,
we were able to get the speed.
Within a few days, we had the speed down to a very small number
of minutes.
I can't remember exactly how many minutes it was,
but it might have been like 10 minutes or something.
And so yeah, we found ourselves at the top of the leaderboard
easily for both time and money, which really shocked me.
Because the other people competing in this
were like Google and Intel and stuff,
where like, no, we're not more about this stuff
than I think we do.
So then we were emboldened.
We thought, let's try the ImageNet one, too.
I mean, it seemed way out of our league.
But our goal was to get under 12 hours.
And we did, which was really exciting.
And but we didn't put anything up on the leaderboard,
but we were down to like 10 hours.
But then Google put in like five hours or something,
and we're just like, oh, we're so screwed.
But we kind of thought, we'll keep trying,
if Google can do it.
I mean, Google did on five hours on like a TPU pod
or something, like a lot of hardware.
But we kind of like had a bunch of ideas to try.
Like a really simple thing was, why
are we using these big images?
They're like 224, 256 by 256 pixels.
Why don't we try smaller ones?
And just to elaborate, there's a constraint on the accuracy
that your train model is supposed to achieve.
Yeah, you've got to achieve 93%.
I think it was for ImageNet.
Exactly.
Which is very tough.
So you have to.
Yeah, 93%.
Like, they picked a good threshold.
It was a little bit higher than what the most commonly used
ResNet 50 model could achieve at that time.
So yeah, so it's quite a difficult problem to solve.
But yeah, we realized if we actually just use
64 by 64 images, it trained a pretty good model.
And then we could take that same model
and just give it a couple of epochs
to learn 224 by 224 images.
And it was basically already trained.
It makes a lot of sense.
If you teach somebody, like, here's
what a dog looks like and you show them low res versions.
And then you say, here's a really clear picture of a dog.
They already know what a dog looks like.
So that, like, just we jumped to the front
and we ended up winning parts of that competition.
We actually ended up doing a distributed version
over multiple machines a couple of months later
and ended up at the top of the leaderboard.
We had 18 minutes.
ImageNet.
Yeah, and people have just kept on blasting through again
and again since then.
So what's your view on multi-GPU or multiple machine
training in general as a way to speed code up?
I think it's largely a waste of time.
Both multi-GPU on a single machine and?
Yeah, particularly multi-machines,
because it's just plunky.
Multi-GPUs is less plunky than it used to be.
But to me, anything that slows down your iteration speed
is a waste of time.
So you could maybe do your very last perfecting of the model
on multi-GPUs if you need to.
But so for example, I think doing stuff on ImageNet
is generally a waste of time.
Why test things on 1.3 million images?
Most of us don't use 1.3 million images.
And we've also done research that shows that doing things
on a smaller subset of images gives you
the same relative answers anyway.
So from a research point of view, why waste that time?
So actually, I released a couple of new data sets recently.
One is called ImageNet, the French ImageNet,
which is a small subset of ImageNet, which
is designed to be easy to classify.
What's how do you spell ImageNet?
It's got an extra T and E at the end,
because it's very French.
Image, OK.
And then another one called ImageWolf,
which is a subset of ImageNet that only contains dog breeds.
But that's a hard one, right?
That's a hard one.
And I've discovered that if you just look at these two
subsets, you can train things on a single GPU in 10 minutes.
And the results you get are directly transferrable
to ImageNet nearly all the time.
And so now I'm starting to see some researchers start
to use these smaller data sets.
I so deeply love the way you think,
because I think you might have written a blog post saying
that going with these big data sets
is encouraging people to not think creatively.
Absolutely.
So you're too, it sort of constrains you
to train on large resources, and because you
have these resources, you think more resources will be better,
and then you start to like somehow you kill the creativity.
Yeah.
And even worse than that, Lex, I keep hearing from people
who say, I decided not to get into deep learning,
because I don't believe it's accessible to people
outside of Google to do useful work.
So like I see, a lot of people make an explicit decision
to not learn this incredibly valuable tool,
because they've drunk the Google Kool-Aid, which is that only
Google's big enough and smart enough to do it.
And I just find that so disappointing and it's so wrong.
And I think all of the major breakthroughs in AI
in the next 20 years will be doable on a single GPU.
Like I would say, my sense is all the big sort of.
Well, let's put it this way.
None of the big breakthroughs of the last 20 years
have required multiple GPUs.
So like batch norm, value, dropout,
to demonstrate that there's something to them.
Every one of them, none of them has required multiple GPUs.
GANs, the original GANs, didn't require multiple GPUs.
Well, and we've actually recently shown
that you don't even need GANs.
So we've developed GAN level outcomes without needing GANs.
And we can now do it with, again,
by using transfer learning, we can do it in a couple of hours
on a single GPU.
So you're using a generator model
without the adversarial part?
Yeah.
So we've found loss functions that work super well
without the adversarial part.
And then one of our students, a guy called Jason Antich,
has created a system called Dealtify,
which uses this technique to colorize
old black and white movies.
You can do it on a single GPU, colorize a whole movie
in a couple of hours.
And one of the things that Jason and I did together
was we figured out how to add a little bit of GAN
at the very end, which it turns out for colorization,
makes it just a bit brighter and nicer.
And then Jason did masses of experiments
to figure out exactly how much to do.
But it's still all done on his home machine,
on a single GPU in his lounge room.
And if you think about colorizing Hollywood movies,
that sounds like something a huge studio would have to do.
But he has the world's best results on this.
There's this problem of microphones.
We're just talking to microphones now.
Yeah.
It's such a pain in the ass to have these microphones
to get good quality audio.
And I tried to see if it's possible to plop down
a bunch of cheap sensors and reconstruct higher quality
audio from multiple sources.
Because right now, I haven't seen work from, OK,
we can save inexpensive mics, automatically combining
audio from multiple sources to improve the combined audio.
People haven't done that.
And that feels like a learning problem.
So hopefully somebody can.
Well, I mean, it's evidently doable.
And it should have been done by now.
I felt the same way about computational photography
four years ago.
That's right.
Why are we investing in big lenses when
three cheap lenses plus actually a little bit of intentional
movement, so like take a few frames,
gives you enough information to get excellent subpixel
resolution, which particularly with deep learning,
you would know exactly what you're meant to be looking at.
We can totally do the same thing with audio.
I think the madness that it hasn't been done yet.
Has there been progress on photography companies?
Yeah, photography is basically a standard now.
So the Google Pixel Nightlight, I
don't know if you've ever tried it, but it's astonishing.
You take a picture and almost pitch black
and you get back a very high quality image.
And it's not because of the lens.
Same stuff with adding the bokeh to the background blurring.
It's done computationally.
Just the pics over here.
Yeah, basically, everybody now is doing most of the fanciest
stuff on their phones with computational photography
and also increasingly, people are putting more than one lens
on the back of the camera.
So the same will happen for audio, for sure.
And there's applications in the audio side.
If you look at an Alexa-type device,
most people I've seen, especially I worked at Google before,
when you look at noise background removal,
you don't think of multiple sources of audio.
You don't play with that as much as I would hope people would.
But I mean, you can still do it even with one.
Again, not much work's been done in this area.
So we're actually going to be releasing an audio library soon,
which hopefully will encourage development of this
because it's so underused.
The basic approach we used for our super resolution,
in which Jason uses for de-oldify of generating
high quality images, the exact same approach
would work for audio.
No one's done it yet, but it would be a couple of months work.
OK, also learning rate in terms of Dawn Bench.
There's some magic on learning rate that you played around
with that's been interesting.
Yeah, so this is all work that came from a guy called Leslie
Smith.
Leslie is a researcher who, like us,
cares a lot about just the practicalities of training
neural networks quickly and accurately,
which you would think is what everybody should care about,
but almost nobody does.
And he discovered something very interesting,
which he calls super convergence, which
is there are certain networks that with certain settings
of high parameters could suddenly be trained 10 times faster
by using a 10 times higher learning rate.
Now, no one published that paper because it's not
an area of active research in the academic world.
No academics recognize this is important.
And also, deep learning in academia
is not considered an experimental science.
So unlike in physics, where you could say,
I just saw a subatomic particle do something
which the theory doesn't explain,
you could publish that without an explanation.
And then in the next 60 years, people
can try to work out how to explain it.
We don't allow this in the deep learning world.
So it's literally impossible for Leslie to publish a paper that
says, I've just seen something amazing happen.
This thing trained 10 times faster than it should have.
I don't know why.
And so the reviewers were like, well,
you can't publish that because you don't know why.
So anyway.
That's important to pause on because there's
so many discoveries that would need to start like that.
Every other scientific field I know of works of that way.
I don't know why ours is uniquely
disinterested in publishing unexplained
experimental results.
But there it is.
So it wasn't published.
Having said that, I read a lot more unpublished papers
than published papers because that's where
you find the interesting insights.
So I absolutely read this paper.
And I was just like, this is astonishingly mind-blowing
and weird and awesome.
And why isn't everybody only talking about this?
Because if you can train these things 10 times faster,
they also generalize better because you're doing less epochs,
which means you look at the data less,
so you get better accuracy.
So I've been kind of studying that ever since.
And eventually Leslie kind of figured out
a lot of how to get this done.
And we added minor tweaks.
And a big part of the trick is starting
at a very low learning rate, very gradually increasing it.
So as you're training your model,
you take very small steps at the start.
And you gradually make them bigger and bigger
until eventually you're taking much bigger steps
than anybody thought was possible.
There's a few other little tricks to make it work.
But basically, we can reliably get super convergence.
And so for the dorm bench thing,
we were using just much higher learning rates
than people expected to work.
What do you think the future of, I mean,
it makes so much sense for that to be a critical hyperparameter
learning rate that you vary.
What do you think the future of learning rate magic looks like?
Well, there's been a lot of great work
in the last 12 months in this area.
And people are increasingly realizing that we just
have no idea really how optimizers work.
And the combination of weight decay,
which is how we regularize optimizers,
and the learning rate, and then other things
like the epsilon we use in the atom optimizer,
they all work together in weird ways.
And different parts of the model,
this is another thing we've done a lot of work on,
is research into how different parts of the model
should be trained at different rates in different ways.
So we do something we call discriminative learning rates,
which is really important, particularly
for transfer learning.
So really, I think in the last 12 months,
a lot of people have realized that all this stuff is important.
There's been a lot of great work coming out.
And we're starting to see algorithms appear, which
have very, very few dials, if any, that you have to touch.
So I think what's going to happen
is the idea of a learning rate, well, it almost already
has disappeared in the latest research.
And instead, it's just like, we know enough
about how to interpret the gradients
and the change of gradients we see
to know how to set every parameter in our way.
There you can automate it.
So you see the future of deep learning, where really,
where's the input of a human expert needed?
Well, hopefully, the input of a human expert
will be almost entirely unneeded from the deep learning
point of view.
So again, Google's approach to this
is to try and use thousands of times more compute
to run lots and lots of models at the same time
and hope that one of them is good.
A lot of mal-kind of stuff.
Yeah, a lot of mal-kind of stuff, which I think is insane.
When you better understand the mechanics of how models learn,
you don't have to try 1,000 different models
to find which one happens to work the best.
You can just jump straight to the best one, which
means that it's more accessible in terms of compute, cheaper,
and also with less hyperparameters to set.
It means you don't need deep learning experts
to train your deep learning model for you, which
means that domain experts can do more of the work, which
means that now you can focus the human time
on the kind of interpretation, the data gathering,
identifying model errors, and stuff like that.
Yeah, the data side.
How often do you work with data these days
in terms of the cleaning, looking at it like Darwin looked
at different species while traveling about?
Do you look at data?
Have you in your roots in Kaggle just look at data?
Yeah, I mean, it's a key part of our course.
It's like before we train a model in the course,
we see how to look at the data.
And then the first thing we do after we train our first model,
which we fine-tune an ImageNet model for five minutes.
And then the thing we immediately do after that
is we learn how to analyze the results of the model
by looking at examples of misclassified images
and looking at a classification matrix
and then doing research on Google
to learn about the kinds of things that it's misclassifying.
So to me, one of the really cool things
about machine learning models in general
is that when you interpret them, they
tell you about things like what are the most important features,
which groups you're misclassifying,
and they help you become a domain expert more quickly
because you can focus your time on the bits that the model is
telling you is important.
So it lets you deal with things like data leakage.
For example, if it says, oh, the main feature I'm looking at
is customer ID.
And you're like, oh, customer ID should be predictive.
And then you can talk to the people that manage customer IDs.
And they'll tell you, oh, yes, as soon as a customer's
application is accepted, we add a one on the end
of their customer ID or something.
So yeah, looking at data, particularly
from the lens of which parts of the data the model says
is important, is super important.
Yeah, and using the model to almost debug the data
to learn more about the data.
Exactly.
What are the different cloud options
for training your networks?
Last question related to Don Bench.
Well, it's part of a lot of the work you do.
But from a perspective of performance,
I think you've written this in a blog post.
There's AWS, there's a TPU from Google.
What's your sense?
What the future holds?
What would you recommend now in terms of training them out?
So from a hardware point of view,
Google's TPUs and the best Nvidia GPUs are similar.
And maybe the TPUs are like 30% faster,
but they're also much harder to program.
There isn't a clear leader in terms of hardware right now.
Although much more importantly, the Nvidia's GPUs
are much more programmable.
They've got much more written problems.
That's the clear leader for me and where
I would spend my time as a researcher and practitioner.
But in terms of the platform, we're
super lucky now with stuff like Google, GCP, Google Cloud,
and AWS that you can access a GPU pretty quickly and easily.
But for AWS, it's still too hard.
You have to find an AMI and get the instance running
and then install the software you want and blah, blah, blah.
GCP is currently the best way to get started
on a full server environment because they
have a fantastic fast AI in PyTorch,
ready to go instance, which has all the courses pre-installed.
It has Jupyter Notebook pre-running.
Jupyter Notebook is this wonderful interactive computing
system which everybody basically
should be using for any kind of data-driven research.
But then even better than that, there
are platforms like Salamander, which we own,
and Paperspace, where literally you click a single button
and it pops up a Jupyter Notebook straight away
without any kind of installation or anything.
And all the course notebooks are all pre-installed.
So for me, this is one of the things
we spent a lot of time curating and working on.
Because when we first started our courses,
the biggest problem was people dropped out of lesson one
because they couldn't get an AWS instance running.
So things are so much better now.
And we actually have, if you go to course.fast.ai,
the first thing it says is, here's
how to get started with your GPU.
And it's like, you just click on the link, and you click Start,
and you're going.
And you will go to GCP.
I have to confess, I've never used the Google GCP.
Yeah, GCP gives you $300 of compute for free,
which is really nice.
But as I say, Salamander and Paperspace are even easier
still.
So from the perspective of deep learning frameworks,
you work with fast AI, if you think of it as framework,
and PyTorch, and TensorFlow, what
are the strengths of each platform in your perspective?
So in terms of what we've done our research on and taught
in our course, we started with Theano and Keras.
And then we switched to TensorFlow and Keras.
And then we switched to PyTorch, and then we
switched to PyTorch and fast AI.
And that kind of reflects a growth and development
of the ecosystem of deep learning libraries.
Theano and TensorFlow were great,
but were much harder to teach and to do research and development
on because they define what's called a computational graph
upfront, a static graph, where you basically
have to say, here are all the things
that I'm going to eventually do in my model.
And then later on, you say, OK, do those things with this data.
And you can't debug them.
You can't do them step by step.
You can't program them interactively
in a Jupyter notebook and so forth.
PyTorch was not the first, but PyTorch
was certainly the strongest entrant to come along and say,
let's not do it that way.
Let's just use normal Python.
And everything you know about in Python
is just going to work, and we'll figure out
how to make that run on the GPU as and when necessary.
That turned out to be a huge leap in terms of what
we could do with our research and what we could do
with our teaching.
Because it wasn't limiting.
Yeah, I mean, it was critical for us
for something like Dawn Bench to be able to rapidly try things.
It's just so much harder to be a researcher and practitioner
when you have to do everything upfront
and you can't inspect it.
Problem with PyTorch is it's not at all
accessible to newcomers because you have to write
your own training loop and manage the gradients
and all this stuff.
And it's also not great for researchers
because you're spending your time dealing with all this boiler
plate and overhead rather than thinking about your algorithm.
So we ended up writing this very multi-layered API
that at the top level, you can train a state-of-the-art neural
network in three lines of code, which
kind of talks to an API, which talks to an API, which
talks to an API, which you can dive into at any level
and get progressively closer to the machine kind of levels
of control.
And this is the fast AI library.
That's been critical for us and for our students
and for lots of people that have won big learning competitions
with it and written academic papers with it.
It's made a big difference.
We're still limited, though, by Python.
And particularly this problem with things
like our current neural nets, where you just can't change
things unless you accept it going so slowly
that it's impractical.
So in the latest incarnation of the course
and with some of the research we're now
starting to do, we're starting to do some stuff in Swift.
I think we're three years away from that being super practical,
but I'm in no hurry.
I'm very happy to invest the time to get there.
But with that, we actually already
have a nascent version of the fast AI library for vision
running on Swift for TensorFlow.
Because Python for TensorFlow is not going to cut it.
It's just a disaster.
What they did was they tried to replicate the bits
that people were saying they like about PyTorch,
the interactive computation.
But they didn't actually change their foundational runtime
components.
So they kind of added this like syntax sugar.
They call TF Eager TensorFlow Eager, which
makes it look a lot like PyTorch.
But it's 10 times slower than PyTorch
to actually do a step.
So because they didn't invest the time
in retooling the foundations because their code base
is so horribly complex.
Yeah, I think it's probably very difficult to do that kind
of retooling.
Yeah, well, particularly the way TensorFlow was written,
it was written by a lot of people very quickly
in a very disorganized way.
So when you actually look in the code, as I do often,
I'm always just like, oh, god, what were they thinking?
It's just it's pretty awful.
So I'm really extremely negative about the potential future
for Python TensorFlow that Swift for TensorFlow
can be a different beast altogether.
It can basically be a layer on top of MLIR
that takes advantage of all the great compiler stuff
that Swift builds on with LLVM.
And yeah, I think it will be absolutely fantastic.
Well, you're inspiring me to try.
Evan truly felt the pain of TensorFlow 2.0 Python.
It's fine by me.
But of course.
Yeah, I mean, it does the job if you're
using predefined things that somebody's already written.
But if you actually compare, like I've
had to do a lot of stuff with TensorFlow recently,
you actually compare, like, I want to write something
from scratch, and you're like, I just keep finding it's like,
oh, it's running 10 times slower than PyTorch.
So is the biggest cost.
Let's throw running time out the window.
How long it takes you to program?
That's not too different now, thanks to TensorFlow Eager.
That's not too different.
But because so many things take so long to run,
you wouldn't run it at 10 times slower.
Like, you just go like, oh, this is taking too long.
And also, there's a lot of things which are just less programmable,
like tf.data, which is the way data processing works in TensorFlow
is just this big mess.
It's incredibly inefficient.
And they kind of had to write it that way because of the TPU
problems I described earlier.
So I just feel like they've got this huge technical debt, which
they're not going to solve without starting from scratch.
So here's an interesting question, then.
If there's a new student starting today,
what would you recommend they use?
Well, I mean, we obviously recommend FastAI and PyTorch
because we teach new students, and that's what we teach with.
So we would very strongly recommend
that because it will let you get on top of the concepts
much more quickly.
So then you'll become an action.
And you'll also learn the actual state-of-the-art techniques.
So you'll actually get world-class results.
Honestly, it doesn't much matter what library you learn
because switching from Shaina to MXNet to TensorFlow to PyTorch
is going to be a couple of days' work if you long as you
understand the foundation as well.
But you think we'll Swift creep in there as a thing
that people start using?
Not for a few years, particularly
because Swift has no data science community,
libraries, schooling.
And the Swift community has a total lack of appreciation
and understanding of numeric computing.
So they keep on making stupid decisions.
For years, they've just done dumb things
around performance and prioritization.
That's clearly changing now because the developer of Swift,
Chris Lattner, is working at Google on Swift for TensorFlow.
So that's a priority.
It'll be interesting to see what happens with Apple
because Apple hasn't shown any sign of caring
about numeric programming in Swift.
So hopefully they'll get off there us
and start appreciating this because currently all
of their low-level libraries are not written in Swift.
They're not particularly Swift-y at all, stuff like Core ML.
They're really pretty rubbish.
So there's a long way to go.
But at least one nice thing is that Swift for TensorFlow
can actually directly use Python code and Python
libraries in literally the entire lesson one notebook
of fast AI runs in Swift right now in Python mode.
So that's a nice intermediate thing.
How long does it take if you look at the two fast AI courses,
how long does it take to get from 0.0
to completing both courses?
It varies a lot.
Somewhere between two months and two years, generally.
So for two months, how many hours a day on average?
So somebody who is a very competent coder
can do 70 hours per course and pick up.
70, 70, that's it.
But a lot of people, I know, take a year off
to study fast AI full time and say at the end of the year,
they feel pretty competent because generally there's
a lot of other things you do.
Generally, they'll be entering Kaggle competitions.
They might be reading Ian Goodfellow's book.
They'll be doing a bunch of stuff.
And often, particularly if they are a domain expert,
their coding skills might be a little on the pedestrian side.
So part of it's just like doing a lot more writing.
What do you find is the bottleneck for people usually,
except getting started and setting stuff up?
I would say coding.
Yeah, I would say the people who are strong coders
pick it up the best.
Although another bottleneck is people
who have a lot of experience of classic statistics
can really struggle because the intuition is so
the opposite of what they're used to.
They're very used to trying to reduce
the number of parameters in their model
and looking at individual coefficients and stuff like that.
So I find people who have a lot of coding background
and know nothing about statistics
generally going to be the best stuff.
So you taught several courses on deep learning.
And as Feynman says, best way to understand something
is to teach it.
What have you learned about deep learning from teaching it?
A lot.
It's a key reason for me to teach the courses.
I mean, obviously, it's going to be
necessary to achieve our goal of getting domain experts
to be familiar with deep learning.
But it was also necessary for me to achieve
my goal of being really familiar with deep learning.
I mean, to see so many domain experts
from so many different backgrounds,
it's definitely, I wouldn't say taught me,
but convinced me something that I
liked to believe was true, which was anyone can do it.
So there's a lot of kind of snobbishness out there
about only certain people can learn to code,
only certain people are going to be smart enough to do AI.
That's definitely bullshit.
I've seen so many people from so many different backgrounds
get state-of-the-art results in their domain areas now.
It's definitely taught me that the key differentiator
between people that succeed and people that fail is tenacity.
That seems to be basically the only thing that matters.
The people, a lot of people give up.
But if the ones who don't give up pretty much everybody
succeeds, even if at first I'm just kind of thinking,
wow, they really aren't quite getting it yet, are they?
But eventually people get it and they succeed.
So I think that's been, I think they're both things
I've liked to believe was true, but I don't feel like I really
had strong evidence for them to be true.
But now I can say I've seen it again and again.
So what advice do you have for someone
who wants to get started in deep learning?
Train lots of models.
That's how you learn it.
So I think, it's not just me, I think our course is very good.
But also lots of people independently have said it's very good.
It recently won the COGX Award for AI courses.
It's being the best in the world.
I'd say come to our course, course.fast.ai.
And the thing I keep on harping on in my lessons
is train models, print out the inputs to the models,
print out to the outputs to the models,
like study, change the inputs a bit,
look at how the outputs vary, just run lots of experiments
to get an intuitive understanding of what's going on.
To get hooked, do you think, you mentioned training,
do you think just running the models inference?
Like, if we talk about getting started.
No, you've got to fine-tune the models.
So that's the critical thing, because at that point,
you now have a model that's in your domain area.
So there's no point running somebody else's model,
because it's not your model.
So it only takes five minutes to fine-tune a model
for the data you care about.
And in lesson two of the course, we
teach you how to create your own data set from scratch
by scripting Google Image Search.
So we show you how to actually create a web application
running online.
So I create one in the course that differentiates
between a teddy bear, a grizzly bear, and a brown bear.
And it does it with basically 100% accuracy.
It took me about four minutes to scrape the images from Google
Search and the script.
There's little graphical widgets we
have in the notebook that help you clean up the data set.
There's other widgets that help you study the results
to see where the errors are happening.
And so now we've got over 1,000 replies in our Share Your Work
Here thread of students saying, here's a thing I built.
And so there's people who are like,
and a lot of them are state-of-the-art.
Like somebody said, oh, I tried looking at Dev and Gary
characters, and I couldn't believe it.
The thing that came out was more accurate
than the best academic paper after lesson one.
And then there's others which are just more fun,
like somebody who's doing Trinidad and Tobago hummingbirds.
She said, that's kind of their national bird.
And she's got something that can now classify Trinidad
and Tobago hummingbirds.
So yeah, train models, fine-tune models with your data set
and then study their inputs and outputs.
How much is Fast.AI courses?
Free.
Everything we do is free.
We have no revenue sources of any kind.
It's just a service to the community.
Year of Saint.
OK, once a person understands the basics, trains
a bunch of models, if we look at the scale of years,
what advice do you have for someone wanting to eventually
become an expert?
Train lots of models.
But specifically, train lots of models in your domain area.
So an expert, what?
We don't need more expert like create slightly evolutionary
research in areas that everybody's studying.
We need experts at using deep learning
to diagnose malaria.
Or we need experts at using deep learning
to analyze language to study media bias.
So we need experts in analyzing fisheries
to identify problem areas and the ocean.
That's what we need.
So become the expert in your passion area.
And this is a tool which you can use just about anything.
And you'll be able to do that thing better than other people,
particularly by combining it with your passion
and domain expertise.
So that's really interesting.
Even if you do want to innovate on transfer learning
or active learning, your thought is,
that means one I certainly share,
is you also need to find a domain or a data set that you
actually really care for.
If you're not working on a real problem that you understand,
how do you know if you're doing it any good?
How do you know if your results are good?
How do you know if you're getting bad results?
Why are you getting bad results?
Is it a problem with the data?
Like, how do you know you're doing anything useful?
Yeah, to me, the only really interesting research
is not the only, but the vast majority of interesting research
is try and solve an actual problem
and solve it really well.
So both understanding sufficient tools on the deep learning
side and becoming a domain expert in a particular domain
are really things within reach for anybody.
Yeah, I mean, to me, I would compare it
to studying self-driving cars, having never looked at a car
or been in a car or turned a car on,
which is like the way it is for a lot of people.
They'll study some academic data set
where they literally have no idea about that.
By the way, I'm not sure how familiar
with autonomous vehicles, but that is literally,
you describe a large percentage of robotics folks
working in self-driving cars as they actually
haven't considered driving.
They haven't actually looked at what
driving looks like.
They haven't driven.
And it applies.
Because you know when you've actually driven,
these are the things that happened to me when I was driving.
There's nothing that beats the real world examples
of just experiencing them.
You've created many successful startups.
What does it take to create a successful startup?
Same thing as becoming a successful deep learning
practitioner, which is not getting up.
So you can run out of money or run out of time
or run out of something.
But if you keep costs super low and try and save up
some money beforehand so you can afford to have some time,
then just sticking with it is one important thing.
Doing something you understand and care about is important.
By something, I don't mean the biggest problem I see
with deep learning people is they do a PhD in deep learning
and then they try and commercialize their PhD.
It is a waste of time because that doesn't solve an actual
problem.
You picked your PhD topic because it
was an interesting kind of engineering or math
or research exercise.
But yeah, if you've actually spent time as a recruiter
and you know that most of your time
was spent sifting through resumes
and you know that most of the time you're just
looking for certain kinds of things
and you can try doing that with a model for a few minutes
and see whether that's something which a model is going
to be able to do as well as you could,
then you're on the right track to creating a startup.
And then I think just being just be pragmatic and try and stay
away from venture capital money as long as possible,
preferably forever.
So yeah, on that point, do you venture capital?
So were you able to successfully run startups
with self-funded for quite a while?
Yeah, so my first two were self-funded
and that was the right way to do it.
Is that scary?
No, species startups are much more scary
because you have these people on your back who do this
all the time and who have done it for years telling you,
grow, grow, grow, grow.
They don't care if you fail.
They only care if you don't grow fast enough.
So that's scary.
Whereas doing the ones myself with partners who were friends
is nice because we just went along
at a pace that made sense and we were able to build it
to something which was big enough that we never
had to work again but was not big enough that any VC would
think it was impressive.
And that was enough for us to be excited.
So I thought that's a much better way
to do things than most people.
And generally speaking now for yourself,
but how do you make money during that process?
Do you cut into savings?
So yeah, so I started Fast Mail and Optimal Decisions
at the same time in 1999 with two different friends.
And for Fast Mail, I guess I spent $70 a month on the server.
And when the server ran out of space,
I put a payments button on the front page
and said, if you want more than 10 meg of space,
you have to pay $10 a year.
So run low like I keep your cost down.
Yeah, so I kept my cost down.
And once I needed to spend more money,
I asked people to spend the money for me.
And that was that basically from then on.
We were making money, and I was profitable from then.
For Optimal Decisions, it was a bit harder
because we were trying to sell something that was more like
a $1 million sale.
But what we did was we would sell scoping projects,
so kind of like prototypy projects.
But rather than doing it for free,
we would sell them $50,000 to $100,000.
So again, we were covering our costs.
And also making the client feel like we were doing something
valuable.
So in both cases, we were profitable from six months
in.
Ah, nevertheless, it's scary.
I mean, yeah, sure.
I mean, it's scary before you jump in.
And I guess I was comparing it to the scaredy-ness of VC.
I felt like with VC stuff, it was more scary.
You're kind of much more in somebody else's hands.
Will they fund you or not?
And what do they think of what you're doing?
I also found it very difficult with VC's back startups
to actually do the thing which I thought
was important for the company, rather than doing the thing which
I thought would make the VC happy.
Now, VC's always tell you not to do the thing that
makes them happy, but then if you don't do the thing that
makes them happy, they get sad.
And do you think optimizing for the whatever they call it,
the exit, is a good thing to optimize for?
I mean, it can be, but not at the VC level,
because the VC exit needs to be 1,000x.
So where else the lifestyle exit,
if you can sell something for $10 million,
then you've made it, right?
So it depends.
If you want to build something that's going to,
you're kind of happy to do forever, then fine.
If you want to build something you want to sell in three years'
time, that's fine too.
I mean, they're both perfectly good outcomes.
So you're learning Swift now?
In a way.
I mean, you've already tried to.
And I read that you use, at least in some cases,
space repetition as a mechanism for learning new things.
I use Anki quite a lot myself.
Me too.
I actually don't never talk to anybody about it.
Don't know how many people do it,
but it works incredibly well for me.
Can you talk to your experience?
Like, how did you, what do you, first of all, OK.
Let's back it up.
What is space repetition?
So space repetition is an idea created
by a psychologist named Ebbinghaus.
I don't know.
Must be a couple of hundred years ago or something,
150 years ago.
He did something which sounds pretty damn tedious.
He wrote down random sequences of letters on cards
and tested how well he would remember those random sequences
a day later, a week later, whatever.
He discovered that there was this kind of a curve
where his probability of remembering one of them
would be dramatically smaller the next day
and then a little bit smaller the next day,
a little bit smaller the next day.
What he discovered is that if he revised those cards
after a day, the probabilities would decrease at a smaller
rate.
And then if he revised them again a week later,
they would decrease at a smaller rate again.
And so he basically figured out a roughly optimal equation
for when you should revise something you want to remember.
So space repetition learning is using this simple algorithm,
just something like revise something after a day
and then three days and then a week and then three weeks
and so forth.
And so if you use a program like Anki, as you know,
it will just do that for you.
And it will say, did you remember this?
And if you say no, it will reschedule it back
to appear again like 10 times faster than it otherwise
would have.
It's a kind of a way of being guaranteed to learn something
because by definition, if you're not learning it,
it will be rescheduled to be revised more quickly.
Unfortunately, though, it doesn't let you for yourself.
If you're not learning something,
your revisions will just get more and more.
So you have to find ways to learn things productively
and effectively treat your brain well.
So using mnemonics and stories and context
and stuff like that.
So yeah, it's a super great technique.
It's like learning how to learn is something which everybody
should learn before they actually learn anything,
but almost nobody does.
So what have you, so it certainly works well
for learning new languages, for learning small projects,
almost.
But I started using it for, I forget who wrote a blog post
about this inspired me.
It might have been you, I'm not sure.
I started when I read papers, concepts and ideas,
I'll put them.
Was it Michael Nielsen?
It was Michael Nielsen.
Yeah, so Michael started doing this recently
and has been writing about it.
So the kind of today's ebbing house
is a guy called Peter Wozniak, who
developed a system called SuperMemo.
And he's been basically trying to become the world's greatest
renaissance man over the last few decades.
He's basically lived his life with spaced repetition
learning for everything.
And sort of like Michael's only very recently got into this,
but he started really getting excited about doing it
for a lot of different things.
For me personally, I actually don't
use it for anything except Chinese.
And the reason for that is that Chinese is specifically
a thing I made a conscious decision
that I want to continue to remember,
even if I don't get much of a chance to exercise it,
because I'm not often in China, so I don't.
Or else something like programming languages or papers,
I have a very different approach, which
is I try not to learn anything from them,
but instead I try to identify the important concepts
and actually ingest them.
So really understand that concept deeply and study it carefully.
Well, decide if it really is important,
if it is incorporate it into our library,
incorporate it into how I do things,
or decide it's not worth it.
So I find I then remember the things
that I care about because I'm using it all the time.
So for the last 25 years, I've committed
to spending at least half of every day learning
or practicing something new, which is all my colleagues
have always hated because it always looks like I'm not
working on what I'm meant to be working on,
but it always means I do everything faster
because I've been practicing a lot of stuff.
So I give myself a lot of opportunity
to practice new things.
And so I find now I don't often find myself wishing
I could remember something because if it's something
that's useful, then I've been using it a lot.
It's easy enough to look it up on Google.
But speaking Chinese, you can't look it up on Google.
Do you have advice for people learning new things?
What have you learned as a process?
I mean, it all starts as just making the hours
and the day available.
Yeah, you've got to stick with it, which is, again,
the one thing that 99% of people don't do.
So the people I started learning Chinese with,
none of them were still doing it 12 months later.
I'm still doing it 10 years later.
I tried to stay in touch with them,
but they just, no one did it.
For something like Chinese, study how human learning works.
So every one of my Chinese flashcards
is associated with a story.
And that story is specifically designed to be memorable.
And we find things memorable, which are funny or disgusting
or sexy or related to people that we know or care about.
So I try to make sure all the stories that are in my head
have those characteristics.
Yeah, so you won't remember things well
if they don't have some context.
And yeah, you won't remember them well if you don't regularly
practice them, whether it be just part of your day-to-day life
or the Chinese semi flashcards.
I mean, the other thing is, let yourself fail sometimes.
So I've had various medical problems
over the last few years, and basically, my flashcards
just stopped for about three years.
And then there have been other times
I've stopped for a few months, and it's so hard
because you get back to it, and it's like,
you have 18,000 cards due.
It's like, and so you just have to go, all right,
well, I can either stop and give up everything
or just decide to do this every day for the next two years
until I get back to it.
The amazing thing has been that even after three years,
I, you know, the Chinese were still in there.
Like, it was so much faster to relearn
than it was to mine the first time.
Yeah, absolutely.
It's in there.
I have the same with guitar, with music, and so on.
It's sad because work sometimes takes away,
and then you won't play for a year.
But really, if you then just get back to it every day,
you're right there again.
What do you think is the next big breakthrough
in artificial intelligence?
What are your hopes in deep learning or beyond
that people should be working on,
or you hope there'll be breakthroughs?
I don't think it's possible to predict.
I think what we already have
is an incredibly powerful platform
to solve lots of societally important problems
that are currently unsolved.
So I just hope that people will,
lots of people will learn this toolkit
and try to use it.
I don't think we need a lot of new technological breakthroughs
to do a lot of great work right now.
And when do you think we're going
to create a human-level intelligence system?
Do you think how hard is it?
How far away are we?
Don't know.
Have no way to know.
I don't know why people make predictions about this,
because there's no data and nothing to go on.
And it's just like there's so many
societally important problems to solve right now,
I just don't find it a really interesting question
to even answer.
So in terms of societally important problems,
what's the problem that is within reach?
Well, I mean, for example, there are problems
that AI creates, right?
So more specifically, labor force displacement
is going to be huge,
and people keep making this frivolous econometric argument
of being like, oh, there's been other things
that aren't AI that have come along before
and haven't created massive labor force displacement.
Therefore, AI won't.
So that's a serious concern for you?
Oh, yeah.
Andrew Yang is running on it.
Yeah, it's desperately concerned.
And you see already that the changing workplace
has lived to a hollowing out of the middle class.
You're seeing that students coming out of school today
have a less rosy financial future
ahead of them than the parents did,
which has never happened in the last few hundred years.
We've always had progress before.
And you see this turning into anxiety and despair
and even violence.
So I very much worry about that.
You've written quite a bit about ethics, too.
I do think that every data scientist
working with deep learning needs to recognize
they have an incredibly high leverage tool
that they're using that can influence society in lots of ways.
And if they're doing research, that research
is going to be used by people doing this kind of work.
And they have a responsibility to consider the consequences
and to think about things like, how will humans be
in the loop here?
How do we avoid runaway feedback loops?
How do we ensure an appeals process for humans
that are impacted by my algorithm?
How do I ensure that the constraints of my algorithm
are adequately explained to the people that end up using them?
There's all kinds of human issues,
which only data scientists are actually
in the right place to educate people are about.
But data scientists tend to think of themselves as just
engineers and that they don't need
to be part of that process, which is wrong.
Well, you're in the perfect position to educate them better,
to read literature, to read history, to learn from history.
Well, Jeremy, thank you so much for everything
you do for inspiring a huge amount of people,
getting them into deep learning, and having
the ripple effects, the flap of a butterfly's wings,
they'll probably change the world.
So thank you very much.
Cheers.