Name: Research with TensorFlow (TF Dev Summit '20)
Uploaded: 2020-03-25T08:28:31.000Z
Duration: 16 min 1 s
Description: Thousands of YouTube videos with English-Chinese subtitles! Now you can learn to understand native speakers, expand your vocabulary, and improve your pronunciation...

ALEXANDRE PASSOS: Hello, my name is Alex,

I am here today to tell you all a little bit about how

you can use TensorFlow to do deep learning research more

going to take a little tour of a few TensorFlow features that

show you how controllable, flexible, and composable

We'll take a quick look at those features, some old

And I want to start by helping you control how TensorFlow

has one kernel matrix and an optional bias

And these parameters are updated when you train your model.

And part of the whole point of training models

is so that we find out what value those parameters should

And if you're making your own layers library,

you can control absolutely everything about how

But you can also crack open the black box

even inside the libraries that we give you.

So for example, we're going to use this little running example

of what if I wanted to re-parametrize a Keras

layer so it does some computation to generate

The way to do this is to use tf.variable_creator_scrope.

It is a tool we have that lets you take control of the state

It's a context manager, and all variables created under it

And this function can choose to do nothing.

Or it can modify how variables are created.

Under the hood, this is what distributionstrategy.scope

to build TensorFlow that we make available to you,

And here, if I wanted to do this re-parametrization of the Keras

First, I define what type I want to use to store those things.

Here, I'm using this vectorize variable type,

objects of this type as a part of TensorFlow computations.

And what we do here is we do a matrix multiplication

of the left component and the right component.

And now that I know how to use this object, I can create it.

In this case, I want to peek at the shape.

Because now, you can take large libraries of code

that you did not write and do dependency injection

Probably if you're going to do this at scale,

you might want to implement your own layer

to be able to extend the ones that we provide you.

So use tf.variable_creator_scope to control the stage.

use these libraries to do research at all,

as opposed to just writing plain Python code,

is that deep learning is really dependent on very

Subtitles ListPlay Video

Research with TensorFlow (TF Dev Summit '20)

bunch

vocabulary

average

structure