array indexing

andrej karphy 의 makemore 강의에서

C = torch.randn(27,2)
# X.shape # 228146, 3
emb = C[X] 
# emb.shape = 228146, 3, 2

가 나옴.
X의 실제 범위는 [0,27)
X 의 index 를 2d 로 embedding 함

y = np.arange(35).reshape(5, 7)
>>> y[np.array([0,2,4], np.array([0,1,2])]
array([ 0, 15, 30])

# 이제 이해 됨 @ 2022.11.11
>>> y[np.array([0,2,4])]
array([[ *0,  1,  2,  3,  4,  5,  6],
       [14, *15, 16, 17, 18, 19, 20],
       [28, 29, *30, 31, 32, 33, 34]])
>>> y[0]
array([[ 0,  1,  2,  3,  4,  5,  6]])

• https://numpy.org/doc/1.18/user/basics.indexing.html#indexing-multi-dimensional-arrays
• https://towardsdatascience.com/advanced-numpy-array-indexing-made-easy-fc49fdaef367