CS231n-2016 assginment implementation
See more information in here
I didn't fully implement assignment1 (knn, image feature) so i don't upload assignment1.
And no answers in inline questions too. :(
Note
Currently working assginment3
Notice in assignment2
cnn.py
and cnn_ap.py
files are for test convolution in locally.
cnn.py, classifier/conv.py
is [conv - relu - conv - relu] - pool - fc1 - fc2 - softmax architecture, and cnn_ap.py, classifier/conv_ap.py
use average pooling technique instead of fc1
layer in conv.py
And use xavier init instead of gaussian normalization in all files.
But there are no correct answers, maybe xavier init and average pooling are wrong. If you find something wrong in my assginment, please tell me something.