Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Swift for TensorFlow – A system for deep learning and differentiable computing

tensorflow.org

Post reply on HN
Print view View original

81–90 of 142 posts

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#81

Post by pjmlp »

MiroF wrote: post #56
pjmlp wrote: post #35

Earlier quoted context omitted.

Interesting given some of the renowned names using it, probably with more revenue than plenty of Rust unicorns.

Apple, Microsoft, Google? This website isn't counting "Julia only" stacks, it's just companies that have used Julia for one project or another. If you really want to compare that to rust, julia is again going to fall short.

Those were not the companies I had in mind with my Rust remark, but if you so wish, Swift, Kotlin/Native, Go, C++, .NET Native, Verona, Checked C, Objective-C.

It remains to be seen how much Rust they will actually make into tier 1 OS SDKs for userspace applications.

In fact, currently it looks more they are bringing their experience with Rust into their platform languages than anything else.

Swift memory ownership, Verona, C++ Core Guidelines checker, Kotlin/Native ownership rules.

Beware wishing for Julia's downfall with glass ceilings.

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#82

Post by mkolodny »

MiroF wrote: post #55
throw6606 wrote: post #13

Careful, folks. S4TF is pretty much dead on arrival. It was pushed aggressively by Chris Lattner (for obvious reasons) but he left Google a while ago and since then most internal users lost interest. There's nothing in Swift that's inherently suitable for ML and building the ecosystem is a ton of work; without all the political pushing, it went nowhere and is close to a "semi-abandoned research project" phase.

> There's nothing in Swift that's inherently suitable for ML The type system?

Speed, too. For PyTorch to train models and run inference quickly, your Python code gets translated to C++/CUDA. Part of the idea with S4TF is to be able to write ML code in a single, fast language.

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#83

Post by kgwgk »

rtorr wrote: post #15
throw6606 wrote: post #13

Careful, folks. S4TF is pretty much dead on arrival. It was pushed aggressively by Chris Lattner (for obvious reasons) but he left Google a while ago and since then most internal users lost interest. There's nothing in Swift that's inherently suitable for ML and building the ecosystem is a ton of work; without all the political pushing, it went nowhere and is close to a "semi-abandoned research project" phase.

“Stillborn” is a pretty awful term to use for software.

Now that I think of it, does anyone know if "abort" is among the computing terms to be avoided (like master, black list or sanity check)?

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#84

Post by p1esk »

kgwgk wrote: post #83
rtorr wrote: post #15

Earlier quoted context omitted.

“Stillborn” is a pretty awful term to use for software.

Now that I think of it, does anyone know if "abort" is among the computing terms to be avoided (like master, black list or sanity check)?

None of these should be avoided, as long as they are accurate terms.

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#85

Post by pjmlp »

dsabanin wrote: post #68
threatofrain wrote: post #46

Earlier quoted context omitted.

If Apple has intrinsic care for Swift on Linux then let it be evident in their story or direction.

I can't speak for Apple, of course, but some indications of their seriousness are there. SwiftNIO is a cross-platform asynchronous event-driven network application framework for rapid development of maintainable high performance protocol servers & clients. ( https://github.com/apple/swift-nio ) Distributed Membership Protocol implementations in Swift: https://github.com/apple/swift-cluster-membership Docker Official…

Still waiting for the day when import Glibc isn't a thing on Swift examples.

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#86

Post by MiroF »

mkolodny wrote: post #82
MiroF wrote: post #55

Earlier quoted context omitted.

> There's nothing in Swift that's inherently suitable for ML The type system?

Speed, too. For PyTorch to train models and run inference quickly, your Python code gets translated to C++/CUDA. Part of the idea with S4TF is to be able to write ML code in a single, fast language.

I am skeptical that there are no calls to underlying C or CUDA libraries occurring. Swift doesn’t naively beat BLAS

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#87

Post by MiroF »

Razengan wrote: post #76
ddbb33 wrote: post #17

Earlier quoted context omitted.

Can you backup your claims?

Newly created account claims something without proof and comments asking for such seem to be getting buried.

Yes, or a Googler who doesn’t want to be seen publicly attacking a Google project

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#88

Post by iflp »

bodono wrote: post #39
gas9S9zw3P9c wrote: post #37

Earlier quoted context omitted.

I'm skeptical of JAX. It feels good right now, but when the first TF beta version came out it was very much like that too - clean, simple, minimal, and just a better version of Theano. Then the "crossing the chasm" effort started and everyone at Google wanted to be part of it, making TF the big complex mess it is today. It's a great example of Conway's Law. I'm not convinced the same won't happen to JAX as it catches…

I know where you're coming from, but TF in my opinion was very user-hostile even on arrival. I can't tell you how much hair-pulling I did over tf.conds, tf.while_loops and the whole gather / scatter paradigm for simple indexing into arrays. I really think the people working on it wanted users to write TF code in a certain, particular way and made it really difficult to use it in other ways. Just thinking back on that…

You can't blame the TF people for things like while_loop. Those are inherited from Theano, and back then the dynamic graph idea wasn't obvious.

JAX is indeed a different situation as it has a more original design (although TF1 came with a huge improvement in compilation speed, so maybe there were innovations under the hood). But I don't know if I like it. The framework itself is quite neat, but last time I checked, the accompanying NN libraries had horrifying designs.

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#89

Post by iflp »

gas9S9zw3P9c wrote: post #37
bodono wrote: post #23

I'm not sure this is really going to take off, it seems that most people who are abandoning TF are moving to Jax or pytorch. My own experience with Jax is that it is much easier to use then TF, just an all round more pleasant experience. It would be interesting to try this, but at this point I'm not really willing to learn 'yet another deep learning framework' and the extreme anti-user problems that TF had make me lo…

I'm skeptical of JAX. It feels good right now, but when the first TF beta version came out it was very much like that too - clean, simple, minimal, and just a better version of Theano. Then the "crossing the chasm" effort started and everyone at Google wanted to be part of it, making TF the big complex mess it is today. It's a great example of Conway's Law. I'm not convinced the same won't happen to JAX as it catches…

> I'm not convinced the same won't happen to JAX

And now there are already multiple NN libraries for JAX from Google...

Re: Swift for TensorFlow – A system for deep learning and differentiable computing

#90

Post by yyyk »

marcinzm wrote: post #20
xvilka wrote: post #11

Should have used Julia instead. The choice was only because of the Chris Lattner who already left.

They presumably wanted a semi-popular statically typed language as the gains of Julia over Python aren't enough to be worth it (and Julia isn't popular enough).

Semi-popular where? The typical iOS programmer isn't going to touch ML. In the relevant demography, Julia had (still has) more users. Swift4TF was a poor choice, and now that Lattner is gone I doubt it has a future.
Post reply on HN