/sci/ - Science & Math

>>10056879
afaik tensorflow is typically used as the back-end and the user interface is mostly keras or pytorch... or TMVA lol

i'm sure you can find plenty of keras tutorials if you google it

>>10056879
>hipster faggot using a library-bound notebook

TensorFlow does a lot of nice hardware optimizations if you configure it right. I still like to use sklearn for tuning my initial models/doing feature selection.

Keras is also a neat project heavily streamlined for doing neural network stuff (network layers are primitives).

>>10056879
make a neural network in tensorflow

or, to start out, literally just make one equation with a variable and placeholder in tensorflow and go from there

basically, start doing it

>>10057251
thats hot.

>>10056879
I don't have anything to add to this discussion, but you have patrician taste in literature, OP.

>>10056879
>>>/g/ tfo

>>10057818
nah dude, /g/ is all about "post your screencaps" or "install gentoo"

their coding threads are all "fuck you, i implemented it in C so it's 3% more optimized" or "fucking FIZZBUZZ I CANT DO FIZZBUZZ"

ML General should be a /sci/ thread, not a /g/ thread

>>10057837
I admit /g/ is for brainlets, but nevertheless this is a /g/aylord thread because you are hacking on a library, not studying ML in the abstract.

>>10056879
>PhD in statistical physics
>Still needs to ask a question any CS undergrad could figure out in a Chinese anime board

>>10057848
I'm making conversation. I could also ask my work collegues who literally do ML 2 desks away from me but they always only talk about gpu optimigation and have a CV background, not python scripting 101

>>10059005
This board is not the place for conversation. It's a place for undergrads with inflated egos, miserable grad students, cranks, and retards who think the Jews are why they aren't winners.

Try IRC or mailing lists.

>>10059050

...b-but...the jews...

t. CAD monkey

>>10059050
Nah, there's good discussion here even with the shitpost.
/g/ is unbearable though

>>10056879
https://universalflowuniversity.com/Books/Computer%20Programming/Neural%20Networks%20and%20Deep%20Learning/Learning%20TensorFlow_%20A%20Guide%20to%20Building%20Deep%20Learning%20Systems.pdf

This, usually o reilly books begin bad in theorical topics but amazing for coding and baby step new fields.

https://developers.google.com/machine-learning/crash-course/first-steps-with-tensorflow/toolkit

>>10056879
Anon, if you are learning TensorFlow take a look at what eager mode is and use it to learn
There will be a major changes to the library by the end of the year and "eager execution" will be the default mode

>>10059050
>being this assblasted about people naming the Jew
Okay, Moshe.

>>10059931
I'm tired of seeing you retards post flat earth tier denials of environmental science and obvious slide threads about {{{SJWs}}} in science. Then, when someone calls out your 8th grade education and names the (You), you vomit /pol/ memes until the thread in unreadable. Go be retarded somewhere else.

The (You) is immunized against all dangers: one may call him a scoundrel, parasite, swindler, profiteer, it all runs off him like water off a raincoat. But call him a (You) and you will be astonished at how he recoils, how injured he is, how he suddenly shrinks back: “I’ve been found out.

wish me luck on my college entrance exam lads ;)

>>10056879
Are you *sincere*?

>>10061817
Good luck, nigger

>>10061863
thanks nigger

>>10061832
more sincere than DFW

>>10059204
Thanks. I'll take a look and them most likely start with Keras, as someone at the beginning suggested

>>10057251
>that 30 yo boomer writing his own library but its shit

>>10062254
By what metric would it be shit? Wouldn't the compiler optimize most of it, unless it was wildly inefficient in the first place? Or do you mean shit in the sense that it's not user friendly for others/bad documentation.

>>10059969
>The (You) is immunized against all dangers: one may call him a scoundrel, parasite, swindler, profiteer, it all runs off him like water off a raincoat. But call him a (You) and you will be astonished at how he recoils, how injured he is, how he suddenly shrinks back: “I’ve been found out.

>>10062261
because people who know alot more about optimizing for gpus than you have already done it