Name: TensorFlow的發展，增加了對JS和Swift的支持（TensorFlow Meets）。 (The growth of TensorFlow with added support for JS & Swift (TensorFlow Meets))
Uploaded: 2021-01-14T10:02:40.000Z
Duration: 7 min 4 s
Description: 【看影片學英語】數萬部 YouTube 影片，搭配英漢字典即點即查，輕鬆掌握單字發音與用法，長久累積看電影不必再看字幕。

祈使句型

And welcome to the Tensorflow developer Summit.

現在進行式

We're right here in the Tensorflow Cafe and I'm chatting with Sandeep, who's a product manager on Tensorflow itself.

Really, really excited to be Are you enjoying the summit so far?

I mean, I think, you know, to see all the people of the users that we have here turned out here in person to listen to their stories.

See the interest Onda following of tensorflow and not just here in the room, back with the other for life stream viewers and the global viewing parties and all that.

It's being really, really good to see all this on.

Like this is a product that you're working on.

Just the sheer excitement be the ideas, the creativity.

I think it just just amazing where we are in the field of machine learning as a whole and rolled a tense up look floor is playing in something.

Yes, yes, absolutely So one of the things we saw in the keynote was just how quickly tensorflow is growing and the trajectory that it's on.

No, I don't think I don't think any of us expected that.

It's, you know, the way machine learning is sort of transitioning in the society right now from not just playing a role in research and sort of very advanced scientific discovery type of things, but beginning to make a difference in so many different areas.

And I'm so I think they seem that user community.

It's just the the explosive growth on the number of people.

And I think it's a testament to this incredibly powerful platform that our engineering teams have built, which is, you know, so flexible on DDE can address these really, really fundamental problems and then our focus and making it easier to use and opening up a lot of new applications, then how the community has also gotten involved, like just check so much stuff in and contributed so much that's being a big part of it, and I think that is that has really helped with our partners with our collaborators and how they have contributed and help sort of open up doors to a lot of applications that we wouldn't have thought about ourselves.

So, yeah, I think it's It's just a great place to be in the best time on some of the scenarios that we've seen.

Yeah, yeah, yeah, the cassava story we saw earlier today, right?

I mean, I think just all these all these applications that touching upon all aspects of human life, I think it's just that that's the best part of It's a beautiful thing.

So you know, when we're seeing this trajectory of growth and like, you know, But now just to pivot to the future a little bit like from a developer perspective, it's, I find it's just really beautiful to see new programming languages.

Yes, yes, yeah, and that has been a big sort of focus for us.

Don't make the black form be supported on in languages and in environments that uses are at.

And so today we announced support for Swift.

In fact, the talk backless slackness is later today on javascript.

I'm super excited about that because you know, javascript, as we know, is the number one language programming language with a huge Web developer community that's out there traditionally, you know what we have seen is that machine learning has Bean sort of little bit more in the fight on world and so on.

I think this opens up the doors for all of the web of lovers to try to show just easily incorporate machine learning into their applications.

They can use all kinds off sensors that are very well connected into the browser world and develop some really nice interactive things.

And we have some amazing demos here that uses are seen.

I can't wait to get home and try it and beat my son with Yes, exactly.

So I think what jealous Kip will allow us to do, really is that one side of it works in the browser, and it's sort of there's really nothing no heavy duty installations or anything like that.

It's very Quicken to get something up and running.

We can take the C train models on with the type of converters that we're making available to this library.

It's going to be very easy to bring in models that have been previously trained another data and then retrained them and then applied them to your applications.

In addition to that, you can also train in the browser.

It is a full fledged training environment.

And you can build very cool interactive applications by doing this.

My very first driver Square program was like to put in a lockbox when somebody used the right mouse button.

Yes, yes, I think we've all written one of them.

Now to think that that language can be used not just for training, but also yes, yes, yes.

And the other thing you can do is I mean, there's this deep learning playground which is available on the loan with Google site.

And that's another way that javascript allows you to sort of visualize these things in a such an interactive way that I think it will drive on ability to understand these models and really sort of a trade faster and come up with looking solutions.

So if you don't wanna learn Pai phone and you're already a javascript developer, you know, just thinking about it from a career perspective that your Java script skills have taken on a whole new meaning.

So if I am a JavaScript developer and I want to take advantage of this, yes.

So I think, you know, there are a lot of resources available specifically for JavaScript.

What we're launching today is a new website, JSTOR tensorflow Dark JST on that website has a very nice getting started section, which has some step by step tutorial is that it walks you through and some quick examples off some girl fitting and regression on dhe misclassification that you can get up and running in your browser s.

I think as we go forward, we'll sort of continuously make more.

The more of these types of applications that people want to do and make them available in this in this platform is Pac Man gonna be there.

Backman demo is already there, actually, so you can.

You can clone that, and you can sort of mess around with it.

Maybe get two people to simultaneously driving.

I'm thinking of an April Fool's prank where I trained the left arrow to be right in the er pyro to be down and see if I could get people to be back.

Sandy learned so much from this as I always do when I chat with you.

So really, really appreciate it and enjoy the rest of the summit.

Thanks, everybody, for watching this episode.