Differentiable Programming Manifesto

https://github.com/apple/swift/blob/master/docs/DifferentiableProgramming.md

by kristianp • 7 years ago

180 75 7 years ago

kragen3• 7 years ago

This looks like it's going to be a really big deal unless they totally fuck it up. Autodiff greatly simplifies all kinds of continuous mathematical optimization algorithms, and using mathematical optimization means you only have to write code that can recognize things that look similar to a solution instead of code that can find the solution — in a sense, mathematical optimization derives your code from tests.

Autodiff is useful for other kinds of things too, not just mathematical optimization, as they mention in the manifesto. This weekend I wrote a real-time SDF raymarcher in a page of Lua (https://gitlab.com/kragen/bubbleos/blob/master/yeso/sdf.png https://gitlab.com/kragen/bubbleos/blob/master/yeso/sdf.lua), but the surface shading is pretty lame because without sampling the SDF multiple times it doesn't yet have any way to calculate surface normals. Guess what a surface normal is? It's the gradient of the SDF, and reverse-mode autodiff can sample the SDF and its gradient in only double the time of just sampling the SDF.

Pretty much anything you do in continuous domains can get simpler and/or better-performing with autodiff. This is why I've written so much about it in Dercuano.

It's worth pointing out that you don't need to redesign your language to get first-class differentiable programming. Source-to-source transformation has been used to do autodiff in FORTRAN for at least ten years, including on pretty massive applications. (This is mentioned in the manifesto.)

seanmcdirmid2• 7 years ago

Conal Elliott (the guy who co-invented FRP) did a nice project on shading computational surfaces by extracting derivatives from code more than a decade ago: http://conal.net/Vertigo/

He is now very active in the differentiable programming community, but he got started on the problem in the graphics field.

noblethrasher• 7 years ago

He also gave nice presentation on AD: https://www.youtube.com/watch?v=ne99laPUxN4 (surprised that you didn't already mention it, considering the venue ;-))

kragen3• 7 years ago

The manifesto links one of his papers, "The simple essence", which I've started but haven't finished.

sitkack2• 7 years ago

AD also has a nice synergy with DNN, the whole formalization allows for moving the runtime into a DNN and thus onto a GPU.

I have been looking for a technique for representing algorithms that are amenable to conversion to anytime [1] I believe AD could be that technique.

[1] https://en.wikipedia.org/wiki/Anytime_algorithm

kragen3• 7 years ago

Well, successive-approximation mathematical optimization algorithms are sort of inherently anytime, in a way: you can stop after any integer number of iterations and use the approximate solution. Some of them fail to be anytime in practice, like SOR (“successive over-relaxation”) where you usually quit after like six iterations but each iteration takes a long time. And autodiff makes it a lot easier to apply mathematical optimization, in particular variants of gradient descent, one popular use of which is training deep ANNs.

Is that what you meant, or are you talking about something else?

dual_basis8• 7 years ago

I like differentiable programming a lot, but can anyone explain the motivation for Swift as their language of choice for this? I assume it was simply because they wanted to get Chris Lattner, and Swift was his baby. Julia seems like a much better target for this, I wish they had just dedicated some money or resources to that. I don't have a problem with Swift's language design per-se (although Julia's dynamic nature really lends itself more naturally to the sort of problems you might be trying to solve using differential programming, in my opinion) but its cross-platform support is terrible. Why would you choose Mac's boutique language as the foundation of your new paradigm?

nnq2• 7 years ago

Julia seems like a much better target for this

Exactly, at least from an outside perspective, why tf?! It seems like people love so much bolting stuff to "classic, conservative" languages that they do it even if it's 10x more work than doing it in a language already having full fledged mature real macros. And this is not Common Lisp we're talking about... Julia is a pretty "normal" language that wouldn't scare anyone off!

(EDIT+: though, as a software engineer primarily, I do have a slight personal preference for more classic single-dispatchy languages like Swift... I imagine that in general it's some king of Stockholm-syndrome we developers experience :P)

MiroF• 7 years ago

"classic, conservative"

Very odd to see these words used in conjunction with a modern language like Swift.

dnautics2• 7 years ago

classic single-dispatchy languages like Swift

Is C++ not classic enough these days?

cbkeller• 7 years ago

Julia seems to be making good progress on this front anyways :) https://github.com/FluxML/Zygote.jl

asplake• 7 years ago

Most likely explanation: they were going to do it in Swift anyway, and they’re doing out in the open, for others to scrutinise, copy and potentially improve upon. That’s cool, right?

RossBencina2• 7 years ago

If I remember correctly, there is an FAQ somewhere that says that they chose Swift because they were already familiar with it, and didn't know anything about Julia.

nnq2• 7 years ago

...practically makes sense, but it's also kind of the very definition of NIH!

tonynguyen1• 7 years ago

https://github.com/tensorflow/swift/blob/master/docs/WhySwif.... Maybe this one

dual_basis8• 7 years ago

That might explain why they chose it, but it doesn't give me (or anyone else who isn't familiar with Swift already) any reason to consider it.

enos_feedler2• 7 years ago

While this is true, the prospect seems exciting to me because best in class ML might be a thing that could drive Swift towards better cross platform support.

kkarakk3• 7 years ago

Cross platform is never going to be a target for current gen Apple execs. It would require a seismic level shift in their tech focus

They don't have a real presence in traditional servers, they don't have any real presence in cloud, they have minimal presence in non consumer focused services area. Cross platform at this point would just obsolete their platform in favor of more widely present OS'es and mobile tech

uoaei• 7 years ago

It's already been done by another group of researchers.[1] I'm not sure if that was a factor in their choice of Swift.

[1] https://arxiv.org/abs/1907.07587

ninjin2• 7 years ago

I think it is a bit unfair to say “I assume it was simply because they wanted to get Chris Lattner, and Swift was his baby”. Swift is an interesting programming language in its own right – and I am saying that as someone coding pretty much exclusively in Julia – and as I have stated before “If you are sitting on a team deeply familiar and passionate about a language – Swift – what kind of managerial fool would not let them take a stab at it? Especially with Lattner’s excellent track record” [1].

[1] https://news.ycombinator.com/item?id=19717815

Only yesterday I gave a lecture to my cohort of MSc students on precisely this topic; there is history going back to the 60s, implementations alive since the 80s, and so much development over the last five years. As someone that cares deeply about the science (and as an engineer at heart), I simply can not be too sad that Googled picked up Lattner et al., poured money over them, and asked them to push the envelope of what is possible with differential programming. After all, what will stop us from lifting over advances to Julia et al. later down the line? Sure, I can wish that it would have been Bezanson et al. and not Lattner et al. that got the money poured over them, but that feels very petty. Let the “best” language win, and I still feel the same way that I felt in 2018 when Swift first planted their flag: “Swift has a non-existing scientific computing community […] they will have to build it entirely from scratch and community building is difficult. […] My decision to side with Julia is partially to stay my own course, partially a preference for ‘the bazaar’ development model, and partially because I have a hunch that Julia has a better chance to capture the scientific computing community as a whole which is likely to yield benefits down the line”.

[2] https://news.ycombinator.com/item?id=16939525

high_derivative• 7 years ago

I disagree, I don't think it's unfair to say at all. It is simply mismanagement to input resources into a language rework that is dead on arrival, community wise. This is of course par for the course for big tech research orgs where big names get a lot of free rein, but that does not mean it's not a strategic mistake here.

This is simply about focus as an org, and this is the reason why PyTorch is getting so popular.

There seems to be a massive lack of focus and direction in the TF org, too many egos wanting to put their stamp all over the APIs and subsystems (tf.keras anyone?).

TensorFlow eager with autograph or Pytorch solve all differentiation problems as far as researchers and practitioners are concerned.

gwd• 7 years ago

I like differentiable programming a lot, but can anyone explain the motivation for Swift as their language of choice for this?

Back in January, Jeremy Howard (of fast.ai fame) wrote up a blog post talking about why he was exploring using Swift for this purpose:

https://www.fast.ai/2019/01/10/swift-numerics/

The post is quite in-depth; but just to give a brief take on other potential languages:

8<---

Here’s my personal view of some languages that I’ve used and enjoyed, but all of which have limitations I’ve found frustrating at times:

Python: Slow at runtime, poor support for parallel processing (but very easy to use)

C, C++: hard to use (and C++ is slow at compile time), but fast and (for C++) expressive

Javascript: Unsafe (unless you use Typescript); somewhat slow (but easy to use and flexible)

Julia: Poor support for general purpose programming, but fast and expressive for numeric programming. ( Edit: this may be a bit unfair to Julia; it’s come a long way since I’ve last looked at it!)

Java: verbose (but getting better, particularly if you use Kotlin), less flexible (due to JVM issues), somewhat slow (but overall a language that has many useful application areas)

C# and F#: perhaps the fewest compromises of any major programming language, but still requires installation of a runtime, limited flexibility due to garbage collection, and difficulties making code really fast (except on Windows, where you can interface via C++/CLI)

C1sc0cat• 7 years ago

Not sure why Julia is ruled out as not a general purpose language here - I got the impression that Julia was for numerical ie Technical programming

ByThyGrace• 7 years ago

OT but I like your "snip" ASCII art, though it took me a while too long to parse it. :)

thanatropism• 7 years ago

Rust?

I’m still crawling like a baby with Rust but I’m quickly getting familiarized with low-level language notions while finding the struct/trait system notebook intuitive than Python.

jsinai• 7 years ago

Edit: this May be unfair to Julia

Can’t understate this enough. Ultimately there is a trade-off being made between end user applications and numerical computing. Julia is surely superior when it comes to numerical computing. It has a super scientific stack and multi-dispatch means it is easy to mix and match these.

Of course there is no reason why Julia can’t evolve a similarly good ecosystem for webapps and other end user front ends. It is after all a _general purpose programming language_ which happens to have native numerical support.

Nelkins2• 7 years ago

C# and F# don't require the installation of a runtime per se . With the most recent versions of .NET Core, C# and F# applications can be published as completely self-contained applications.

(Yes .NET languages have a runtime, but I don't think that's what the author was referring to).

dual_basis8• 7 years ago

Reading the link, it seems that Jeremy Howard only tried out Swift because Swift for Tensorflow existed. I got the impression from your comment that using Swift for this purpose was somehow "in-the-air" at the time, and many people independently considered it, but I don't think this is true. My impression is quite the contrary - when they announced Tensorflow support for Swift, I think many people were surprised.

ddragon4• 7 years ago

Julia and Swift are different enough that there is benefit in having both with that functionality. Julia is a dynamic and heavily interactive language (especially when it improves more JIT latency and "time to first plot") for exploratory analysis (like what Python provides) and development without compromising performance in any step, with strong focus on HPC/Distributed computing.

Swift has some interactivity, but it's core execution strategy is to create static binaries to be deployed in mobile and desktop, with much better support for high quality end user interfaces (especially in apple devices).

While as a researcher and mostly linux developer I will probably not use Swift, I can easily see situations in which Swift would be the clear choice given both Zygote and Swift for Tensorflow equally mature, even when Julia's scientific computing support is much broader.

dual_basis8• 7 years ago

s/especially/exclusively/

taharvey6• 7 years ago

Or... as Jeremy Howard said it, having a real compiler with a strongly typed language is "mind blowing" after working in Python.

A compiler isn't in your way, with modern JIT compilation and a language with an actual type system, its actually an productivity assistant. Shouldn't AI researchers have automation too?

taharvey6• 7 years ago

Swift is:

  1) A fast general purpose systems language (like C++)

  2) With high level semantics and usability (like Python)

  3) With type system from functional programming (like Haskel)

  4) With concurrent memory model (like Rust)

  5) That uses compile-time analysis for static guarantees

  6) That runs everywhere from CLI scripting to ML to App devlopment

  7) With a huge user base (millions of devs)

  8) And two large corporate backers (Apple & Google)

  9) The developers include the originators of Rust and the developers of LLVM & Clang.

Thats a killer list. And Julia only addresses 1 & 2.

It is further believed by the core group at Google and Fast.ai (and history) that the future must be a static type checked language to scale beyond the script-level experimentation of python to large application development where the lines between ML, application, and system development will blur. Its will also be critical to next-gen performance breakthroughs and heterogeneous computing. And the type system provides a fully hackable language for fast experimentation.

Also Richard Wei, not Lattner, I believe initiated the project with his DLVM thesis... specifically because Swift was such a good fit and is so hackable.

dual_basis8• 7 years ago

Except it basically only runs on MacOS, which means that it doesn't have strong datacenter support, which means that it won't be used in any serious applications.

It's also not dynamic, which is almost a must-have for exploratory data analysis.

I will avoid a point-by-point rebuttal, mostly to avoid seeming too antagonistic, but Julia supports many of the other features on the list you provided, and (more importantly, for the case at hand) already has proven it's capability in the differentiable programming paradigm with Zygote.

targonca• 7 years ago

So now everyone's pet project gets a language extension in Swift? Apple extended the language with @_functionBuilder for SwiftUI. Now Google wants to extend it with this differentiable stuff for TensorFlow.

Why won't they implement proper metaprogramming support instead of proliferating the core language?

classified2• 7 years ago

Why won't they implement proper metaprogramming support instead of proliferating the core language?

Proper metaprogramming support is hard. Polluting the language with special case hocus pocus is relatively easy.

marmaduke2• 7 years ago

Polluting the language with special case hocus pocus is relatively easy

or if it's well done is it still hocus pocus pollution?

fuklief• 7 years ago

Meanwhile, Google researchers (Plotkin and Abadi) are going to present "A Simple Differentiable Programming Language" at POPL20[1] in January.

I can't find a preprint though :'(

[1] https://popl20.sigplan.org/track/POPL-2020-Research-Papers#e...

ml_thoughts• 7 years ago

Went up this morning on arXiv: https://arxiv.org/abs/1911.04523

jpz• 7 years ago

Does anyone else dislike the usage of the word "manifesto" in computing? It always seems to me to indicate a faith-based rather than evidence-based commitment to some train of thought.

Seems strange to see manifestos discussed in the ML space, which is nominally a measure-oriented world. Mind you, when has statistics not been polemical.

new2628• 7 years ago

My experience is that evidence-based commitments are overrated and are usually a post-hoc rationalization for faith- (or gut-feeling)-based ones.

nuclx• 7 years ago

Striving for concise syntax for mathematical operations is a noble intention, but it's hardly a good idea achieve that by shoehorning changes into an existing language.

A language framework designed for extensibility has to be the basis allowing for the implementation of concepts such as differential or probabilistic programming, and it's not Haskell, i.e. we're not there yet.

byt143• 7 years ago

Julia,'s source to source autodiff is implemented as a custom compiler pass in a third party package. Similar approached are bring taken by prob programming packages.

mratsim• 7 years ago

So the software is improving by leaps and bounds but where is the hardware?

OpenCL --> deprecated

Cuda --> No Nvidia GPU

ROCm/HIP: does it work on Mac or are you strong-armed into using Apple Metal?

Metal --> I couldn't find an AXPY example

keithalewis• 7 years ago

There is a simpler way to do this; https://github.com/keithalewis/epsilon/blob/master/README.md.

bcheung• 7 years ago

Pardon my ignorance but I have a few questions...

1) This seems like it would only work for varying the values passed to variables in a function. Can you differential a function other than just changing some values -- like using a genetic algorithm. Seems like this is just finding the right values to pass to a function not finding a different function itself.

2) Is this limited to only continuous / ordered values? How can you differentiate logic, branching, and what other functions to call? Seems like to do so would require mapping a function to some kind of n-dimensional euclidian space / manifold.

3) Why does this need to be an extension to a language? Can't ordinary interfaces in the OOP sense or monads in the FP sense be used to wrap functions and give them this functionality for free?

ddragon4• 7 years ago

1) It's not restricted to the variables being passed, for example Zygote [1] has an example in the README with IO gradient(x -> fs[readline()](x), 1), and it's not using numerical differentiation (varying inputs to check the output), but finding a formula (approximate, not a closed form like a symbolic differentiation) of how changing the output affects the input. Genetic algorithms is a black box metaheuristic, so people would favor using gradient descent given that the code is differentiable, but how the model is optimized is technically open to any method.

2) Yes, it can differentiate all control flow. What happens is that each forward/backward pass will have a different graph based on what branches it would follow (effectively a subderivative). Each one of these graphs is independently differentiable and do not contain the control flow by themselves.

3) It doesn't need a language extension (Julia doesn't, but that's because you can access Julia IR using the language itself, not many language support this level of metaprogramming). The OOP strategy (like Pytorch) overload methods so every time you call the overloaded method it builds the graph. It depends on customized types and methods and does not support code that isn't using those custom types that can hold the graph (native/custom types), native control flow (only implicitly by changing what graph is being created), reduces the possible optimizations (you only have one subderivative, not the complete graph to optimize) and end up creating a DSL with it's own error messages and quirks that is less natural than just using the host language. I can't say how it would look using monads though.

[1] https://github.com/FluxML/Zygote.jl

Nelkins2• 7 years ago

I was recently looking at a library in F# that does something similar: https://diffsharp.github.io/DiffSharp/

Differentiable Programming Manifesto

Related Stories