MindSKKU
commented
6 years ago Naive uses iteration loop, while vectorized doesn't as written in your assignment. This is just an implementation issue, so it is not treated in the lecture. For example, dot product can be performed for 1-d vector by np.matmul, but naive uses for loops.
for i in range(x.shape[0])
result += x[i]*x[i]This also help you to understand. https://www.youtube.com/watch?v=yl0LTcDIhxc
두개의 function 을 채워야 하는데 naive softmax 와 vectorized softmax 차이점을 "정확히" 좀 알려주시면 좋겠습니다. 수업시간에는 그런 용어로 따로 배우지는 않았던것 같아서요.