The growth of TensorFlow with added support for JS & Swift (TensorFlow Meets)

Key Takeaways

[Music] hi everybody and welcome to the tensorflow developer summit we're right here in the tensorflow cafe and I'm chatting with Sandeep who's a product manager on tensorflow itself and Sandee thanks for coming Thank You Don and really really excited to be here are you enjoying the summit so far oh this has been incredibly exciting I mean I think you know to see all the people the users that we have here we turned out here in person to listen to their stories what they're doing see all the cool work see the interest and and the following of tensorflow and not just here in the room but with all of our you know livestream viewers and the global viewing parties and all that it's it's been really really good to see all this and it's very motivating it's like this is a product that you're working on and just the sheer excitement that's out there about it it's like yeah maybe the ideas the creativity I think it's just just amazing where we are in the field of machine learning as a whole and the roller tensorflow is playing in that sometimes I just have to pinch myself yes yes absolutely yeah so one of the things we saw in the keynote was just how quickly tensorflow is growing and the trajectory that it's not on there was that what you expected no I don't think I don't think any of us expected that or anticipated that right I think it's mean 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 think we've seen that in our in our user community right it's the just the the explosive growth and the number of people and I think it's a testament to the this incredibly powerful platform that our engineering teams have built which is you know so flexible and 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 and then how the community has also gotten involved and like just checks so much stuff in and contributed so much that's been 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 helped 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 it's just a great place to be at the and some of the scenarios that we've seen right it's just like it's it's all inspiring like finding planets yeah yeah yeah and 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's that's the best part about this it's a beautiful thing so you know when we're seeing this trajectory of growth in like you know but now just a pivot to the future a little bit like you know from a developer perspective it's I find it's just really beautiful to see new programming language is it yes yes yeah and that has been a big sort of focus for us to make the platform be supported on in languages and in environments that that our users are at and so today we announced support for Swift in fact the talkback listener is later today and JavaScript and 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 been sort of a little bit more in the in the Python world and so on I think this opens up the doors for all of the web developers to try to zoom just easily incorporate machine learning into their applications they can use all kinds of 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 that users are seen pac-man is fun I can't wait to get home and try it and I beat my son with it yes I could probably go faster like that yeah and this is all done in JavaScript that's right exactly so I think what javascript will allow us to do really is that one sort 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 quick and to get something up and running we can take previously train models and with the type of converters that we are making a little available through this library it's going to be very easy to bring in models that have been previously trained another data and then retrain them and then apply them to your applications and in addition to that you can also train in the browser that's yes exactly yes which is just mind-blowing yes so it is it is yeah it is a full fledge training environment and you can build very cool interactive applications by doing this yes I remember my very first JavaScript program was like to put an alert box when somebody used the right mouse button yes yes I think we've all written one of them yeah yeah and now to think that that language can be used not just for training but also for inferences then yes yes yes and the other thing you can do is I mean there is this deep learning playground which is available on the learn 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 an ability to understand these models and really sort of iterate faster and come up with working solutions so if you don't want to learn Python and you're already a JavaScript developer you know just thinking about it from a career perspective that your JavaScript skills have taken on a whole new meaning yes yes yes start so if I am a JavaScript developer and I want to take advantage of this yes how what would you recommend to get started you mentioned that playground yes yeah yeah so I think you know there are a lot of resources available so specifically for JavaScript what we are launching today is a new website JSON support and that website has a very nice getting started section which has some step-by-step tutorials that it walks you through and some quick examples of some curve fitting and regression and image classification that you can get up and running in your browser so that's an excellent way to start and I think as we go forward we will sort of continuously make more and 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 pac-man 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 drive it but you can do some very fun things I'm thinking of an April Fool's prank where I train the left arrow to be right and the up arrow to be down and see if I can get people to be back yeah that'll be fun don't tell anybody it's just everybody here you're the only ones who know so thank you so much sandy learn so much from this as I always do when I chat with you so really really appreciate it and enjoy the rest of the summer Thank You Lawrence Thank You Don so awesome try to do and thanks everybody for watching this episode I'm Laurence Moroney I've had Sandeep Gupta with me if you have any questions for me or if you have any questions for Sandeep please leave them in the comments below and we'll put links to what we spoke about in the description for this video so thank you everybody [Music] [Music]