chiphuyen / python-is-cool Goto Github PK

Hi! Super useful compilation - some of these took me years to discover. I'll certainly be directing people here, so thanks for taking the time.

I believe there may be a small issue with the text in section 2.2, this bit:

The syntax [x:y:z] means "take every zth element of a list from index x to index y", with 0 being the default value of x and -1 being the default value of y. When z is negative, it means going backwards. So if we want to take every 2th element of a list, we use [::2], which is the same as [0:-1:2].

My understanding is that -1 isn't the default value of y, and [::2] is different to [0:-1:2]. Specifically, [x:y] means 'start at index x, stop before index y'. The default value of y is kind-of -0 (although that would be bad syntax), and using [:-1] will drop the last element.

Concretely (I tried it with python 3.6):

k = [0, 1, 2, 3, 4, 5, 6, 7, 8] # odd number of elements
a = k[::2]    # a <- [0, 2, 4, 6, 8]
b = k[0:-1:2] # b <- [0, 2, 4, 6]

In the second case the indexing starts at 0, ends before 8, and strides by 2.

Hope this is a useful contribution!

Will it be cool to add forced-named arguments to this list?

def foo(pos, *, forcenamed):
     print(pos, forcenamed)

>>> foo(pos=10, forcenamed=20)
10 20

>>> foo(10, forcenamed=20)
10 20

Thank you for this guide.

In section 5 I think:

If we intend that only Endoer, Decoder, and Loss are ever to be imported and used in another module, we should specify that in parts.py using the all keyword.

was intended to be:

If we intend that only Encoder, Decoder, and Loss are ever to be imported and used in another module, we should specify that in parts.py using the all keyword.

In block [5] of the notebook, you accidentally took the square root of the total square error, converting it into mean absolute error.