TensorFlow Dev Summit 2018
March 30, 2018, California, USA
TensorFlow Dev Summit 2018
Video
Using the tf.data API to build input pipelines
Available
In cart
Free
Free
Free
Free
Free
Free
Add to favorites
7.28 K
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speaker

Laurence Moroney
Staff Developer Advocate at Google

Laurence is a developer advocate at Google working on machine learning and artificial intelligence. He's the author of dozens of programming books, and hundreds of articles. When not Googling, he's author of a best-selling Science Fiction book series, and a produced screenwriter.

View the profile

About the talk

In this episode of TensorFlow Meets, Laurence Moroney sits down with Derek Murray to discuss the latest on tf.data. Together, they discuss the impact of building fast, flexible, & efficient input pipelines with the tf.data API. Derek also touches on tf.data achieving performance benchmarks, the addition of convenient functions to make it easier for the TensorFlow community to get started, and datasets auto-tuning. Let us know what you think in the comments below!

Share

Hi everybody and welcome to the tense of glow developer summertime Laurence Maroney, and I'm here at the tensorflow cafe. And it's my honor to chat with Derrick Murray and Erick just did a talk on tfdata and lots of great stuff and see if they didn't let the new stuff coming to Performance and we make some announcements about the the new performance numbers were cheating with second number that we're really proud of what we also know that we can totally be timer still working so I can make a making further advances going to be

bigger and a lot of the library and to do with different kinds of data formats. One of the things he spoke about a lot and I really like that country was like Fast flexible and easy-to-use, you know, so I guess is that your catchphrase talk to you think that's going to have so I can people building models before we had a library with these kind of been to either of those feeding route, which was very quiet. So we should buy any Cody lights and pipe them and it was working to get into Penn Circle rank the first little

held up by the keep up with more than three years ago. You could do that for pipeline. Deborah lot faster one more efficient blocks very sharp edges forget to put a line. What we want to do is take the good parts of both of those and put them together. So they would need to be able to view my case. Like I'm a relatively new developer. Can I get a lot of this works? Cuz I got my data and I throw it at I never thought about how I refine my data to be able to get in to make it more efficient and I'm sure if I was doing it be a hundred images II of the

fountain but how do I get from that to the 13,000? There are a lot of tips and tricks to play with Easy to use and then performance was reconsideration. But as we discovered that we were becoming more and more than I do in is the clarity of thought that was coming out here and the best way to learn how to do it yourself if you look at the performance and it's Elementary queuing Theory, right? So hopefully it stays that's what we're also planning to do and Brandon States today is also training your pipeline a bit more.

All the parts you need so that someone skilled in the art can put something together. It's fast, but we want that speed today accessible to everyone as well because it's only by Nicki things fast enough so we can I purchase informative you use of machine learning. So we're going to apply these changes automatically going to be sitting at the end. Of course people should watch or talk later this afternoon. He is the steps of using our ultimate editions adults are monitoring tools to like work out

how to make your pipeline more efficient sounds good. So if I want to get start Where should I go? Well, I think the best place to get started. I would like a single command you to toggle data sets download and the name of the days that you seen on their website and said one come out of this all the files from your local directory and then intend to blow you can use this you make CSV to decide if you are a nuclear Rachel Lindsay just join the team did in our first couple weeks. It's really simplifies the task of working right

one line of code and you're off to the races for newly doing Google can do that, but that's that's the community contribution. The library offer different formats in four different data analysis approaches that people want to take and we really appreciate it and you'll invite people to spend as many as possible. Thanks to you. So, thank you, and thanks everybody for watching. If you got any questions for me, if you got any questions with Derek, please leave them in the comments below and don't forget to hit that subscribe button. Thanks.

Cackle comments for the website

Buy this talk

Access to the talk “Using the tf.data API to build input pipelines”
Available
In cart
Free
Free
Free
Free
Free
Free

Access to all the recordings of the event

Get access to all videos “TensorFlow Dev Summit 2018”
Available
In cart
Free
Free
Free
Free
Free
Free
Ticket

Interested in topic “AI and Machine learning”?

You might be interested in videos from this event

March 11, 2020
Sunnyvale
30
204.76 K
dev, google, js, machine learning, ml, scaling, software , tensorflow, web

Similar talks

Megan Kacholia
Engineering Director at Google
Available
In cart
Free
Free
Free
Free
Free
Free
Alex Passos
Software Engineer at TensorFlow team
Available
In cart
Free
Free
Free
Free
Free
Free
Alex Irpan
Software Engineer at Google
Available
In cart
Free
Free
Free
Free
Free
Free

Buy this video

Video

Access to the talk “Using the tf.data API to build input pipelines”
Available
In cart
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
558 conferences
22053 speakers
8194 hours of content