Python is usually used as a frontend language for that. It's not running the actual NN. That's all C/C++.
As to why python was chosen. I'm thinking it's because big data people have a big interest in AI and big data people have been using python frontends (think anaconda/pandas) for a lot of their work already. It's not that strange.
python is just the front end. tensor flow is not written in python. it's a c++ project and python just provides an easy API so brainlet "data scientists" can use it without having to know when to make your d'tor virtual. it's literally training wheels for people too dumb to learn a proper language
but then again the whole AI/ML shit is a fucking meme
I wouldn't expect the haskell memory allocator to be that terrible.
And do you care about how much virtual address space you take?
>And do you care about how much virtual address space you take?
that's not how sbrk works
Linear types would have prevented this
besides the documentation should I read Think Python 2e or A Byte of Python to learn python? is there a web dev-oriented python book cuz I want to program web apps with python and use it as server-side language (planning to learn Flask then Django as soon as I finish one of those books)
*rustles*
/dpt/ is now a frog thread
dumb frog poster
Is there a way to rewrite the following python code without the extra indent for try catch?
try:
while True:
print(next(myiter))
except StopIteration:
return
The try is scoped with the while block, so ideally instead it would work something like this:
while True, try:
print(next(myiter))
except StopIteration
return