Name: Swift for TensorFlow (TensorFlow @ O'Reilly AI Conference, San Francisco '18) (Swift for TensorFlow (TensorFlow @ O’Reilly AI Conference, San Francisco '18))
Uploaded: 2021-01-14T10:00:14.000Z
Duration: 12 min 35 s
Description: VoiceTubeの動画で発音を聞きながら英語表現を覚えよう！学べる英語：

My name's Richard Wei, and I'm from the Google Brain Team.

The users are required to explicitly build a graph

and run it and pass it to a session at runtime.

Eager execution allows you to write everything

in the source code, and you interpret it line by line.

But there's a tension between graph execution

but it's not very easy to use, especially when

but can be challenging to achieve high performance,

Well, can we combine usability and performance?

that combines the ease of use with high performance

But to do that, we have to do something more than just

We had to innovate in the programming language.

that we're willing to adopt a programming language

not only do we need the ease of use and high performance,

driven by an established community for the language.

This brings us to the Swift programming language.

Swift is a fast, modern programming language

It's cross-platform, has a clean syntax like Python,

and it has all the powerful capabilities,

like type inference, optionals, and supports

functional programming and all the great features.

Well, the key components of Swift for TensorFlow

As you would expect, you can initialize a tensor like this,

Or you can use a full power generics to power

your models, libraries, and applications.

Well, math operators, like sigmoid and matmul

are type saved generic functions in Swift,

without the need of name spacing when you use them.

When you combine things together, it's nice and clean.

The programming style looks just like Python's eager execution.

and is part of the TensorFlow library, written in Swift.

However, it works very differently from other language

bindings because TensorFlow, while tensor

is supported as a first-class citizen in the programming

For this, we built a technology called graph program extraction

In addition to the ability to treat tensors

It's graph program extraction, using automatic differentiation

and do interesting things even with Python

Well, all of these components are integrated natively

providing great productivity for a developer.

So let's start with a TensorFlow library.

The TensorFlow library has all the common APIs like the Tensor

In data API, you can use dataset initializer.

You can use functional zip and batch operations just

Well in addition to nice-looking TensorFlow APIs defined

in Swift, you also have access to all the raw TensorFlow

Tensor code sits alongside the rest of the Swift program,

interacts with arbitrary user-defined data structures

The TensorFlow library works quite differently

Swift for TensorFlow is a full-fledged compiler,

and it makes our code run fast without sacrificing

The technology is called graph program extraction.

Well, we want users to write code like eager execution,

but we also want it to graph level performance.

And we want to support native language control flow

Swift for TensorFlow (TensorFlow @ O'Reilly AI Conference, San Francisco '18) (Swift for TensorFlow (TensorFlow @ O’Reilly AI Conference, San Francisco '18))

字幕リスト動画再生

Swift for TensorFlow (TensorFlow @ O'Reilly AI Conference, San Francisco '18) (Swift for TensorFlow (TensorFlow @ O’Reilly AI Conference, San Francisco '18))

arbitrary

advantage

script

integrate