How to start working on TensorFlow?

When you and I download Tensorflow too and start working on it we work on a new way of TensorFlow to not the old one Okay has by any chance anybody here worked on tools Such a spigot HaIf in big data begin Hi You worked on which one I’m talking to you Hi Um what do I do Correct It translates all the qualities that have given high into my pretties jobs Right on Do you know that I’m sure you know that high is a lazy execution engine, Yeah Similarly there’s another tool called pig which is a lazy execution engine We use these two on big data technologies to make creator do not produce jobs.

What it means is you give the set off instructions Those instructions are not immediately executed You declare a number A declared a number B then declare another number C is equal to a plus B These two engines they keep noting Don’t I have to do this at this point out to create a variable called See at this point after added this variable in this value to create the season.

It just’s a to-do list They don’t immediately execute your commands and give the results That concept is called lazy execution So in lazy execution what these tools do is the tools generate a graph from the starting point The tools generate a graph and the grafts could have branches so and so forth This is scored ah directed basically graph dags Dogg Yeah Veggies The advantage of doing graphs ISS The tool itself will scan the graph on the Optimize your execution steps if you’re done the same oppression in two places Multiplication between say two variables C and B.

If you’re done it in two places will know down on little copy that result off scene today into memory and reuse it wherever it’s acquired It will not do it, again and again, Looking at the branches it will see which branches can be run in Berlin which branches and which are patients can be running parallel which operations need to be run in sequence Those things which can be running parallel residents those things that that cannot be pearlescent running in sequence so that graph-based approach vertical lazy execution had its own advantage Where that wanted was when you optimize the operations the results of us that your application will run faster than other ways Okay.

When you’re researching on you’re doing some statements or their operations that there and you want immediate results to see what is the impact of the step that was not possible in this graft basement hood So what did it is intensively or 2.0 They gave you what it’s called eager execution In eager execution There is no graph every commander to give the commanders immediately executed on to get the results on the screen However they also have given your facility that if we want to continue to use graph.

It has its own advantages then you can still generate graphs in TensorFlow two points It also but the fourth behavior of tens of 2.0 is eager execution Eager executions means no use of graphs You’ll get immediate results Where’s intensive for one point Little You’ll notice that after every step you’re not getting results You’re only getting meta information There’s a big difference between the two and the other differences both Kira’s and transit flew actually originated in Who will I think they were done by different means on their running properly evolving independently Cara’s used to support various other engines, not just TensorFlow But recently in 2019 when the release washing 2.0.

They have taken a decision that Sarah’s in Tensorflow will become tighter and tighter integrated Eventually Kira’s will stop supporting the other company engines But for the foreseeable future both care us on Tensorflow caters So when you don’t know Tensorflow now along with the Tensorflow Kira’s comes bundled with it Other than that we also have separate one So we have tens of Lucas Kato’s on We also have independent killers available today Both of them will continue to evil but eventually this will I think fear doubt they’ll remove They’ll start supper Stop supporting it.

What they’re advising today in the marketers they never use a kid Us trying to use TensorFlow Kira’s as much as possible because eventually this will fade away Okay Venice that eventuality We don’t know What we’re going to do today is we’re going to use TensorFlow 2.0 on We’re going to use the chaos that comes bundled with TensorFlow 2.0 between the TensorFlow Kira’s and this killer’s as we saw it was not in this class it of the class There are some differences in syntax There are some differences in the Norman creature off the various functions that views So there are some differences Looks like the differences eventually start growing bigger and bigger.

To be on the safe side we’re going to use TensorFlow to Pancho on the chaos that comes bundled with TensorFlow 2.0 Okay all the 50 All right many of us go get please distribute worked at this time Mr. Actually what does care us in intensive Okay Curious You can think of this as a utility using which I’ll build my deep learning models as we’re going to do Know Okay What it will do is it will transform all those things into TensorFlow language for us We can directly going work on the TensorFlow core itself We can directly work itself But the learning curve for TensorFlow is much higher than the learning or for Kira’s He didn’t see me get comfortable with this in a matter of a few days.

This is not so easy Sibley Based to caresses it’s ah no it’s on You based We will use it No it has its own commands Will import those things functions utilities and on will be But it’s very simple Some control Any of the questions Tell him one Okay Ah yeah The Google column which you’re using right now by the former Google collaborates still supporting TensorFlow one points total However upgrading to TensorFlow two points of us A simple commander here to give the commander here to give us a tip Install But don’t give the command right now Don’t give it Don’t use it, Okay All If you are in Gurukul, No, Okay Please go to Bullwinkle I want you to you open both the files dif learning wind or deduce.

It also did not go don’t I p wind be on deep learning m in ist daughter Be Wendy but use both of them will use book on the wine’s data set is what I will start with first, Okay That reminds me of another point while I’m waiting for you people when you install TensorFlow TensorFlow can make use of GP use Graphical processing units on that makes things very fast Okay but to be able to use the Jeep use you have to have another library for NVIDIA called Cued up C U D e this c uta earlier had a full form which subsequently they dropped Okay Andi I’ve been trying to memorize this It just doesn’t fit into my brain every time it slips out Uh it’s last a something to do with architecture.

Development architectural C stands for I think composite Uh I don’t know what you stand for Please find out NVIDIA uses this word cura c u d a To represent a bunch of libraries which they have designed They have the design Toby used along with their chips When you use them When you install the computer library then TensorFlow when your download and install you can install TensorFlow for deep use.

Leave a Comment