draw_convnet
draw_convnet copied to clipboard
How to remove one fully connected layer?
I only have one fully connected layer in my conv net. However, when I try to remove it from list num_list = [1, 500]
I get add_layer(patches, colors, size=size_list[ind], num=num_show_list[ind], IndexError: list index out of range
How do I solve this problem?
size_list and num_list should have same size, change the line above to
size_list = [fc_unit_size, fc_unit_size]
would fix the error
Thanks! However, now I am getting a huge gap between conv and fully connected layers and no directed links between convolutional layers.
looks like u are using Python 3? or matplotlib 2.0 there're some problems there
python 2 and matplotlib 1.4 should have no problem.
sorry that i don't have time to look into details at this moment
On Mar 9, 2017 17:08, "Laurynas Tumosa" [email protected] wrote:
[image: convnet_fig] https://cloud.githubusercontent.com/assets/7099082/23777621/e79dcc38-052d-11e7-9ee7-df7995a37788.png
Thanks! However, now I am getting a huge gap between conv and fully connected layers and no directed links between convolutional layers.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/gwding/draw_convnet/issues/6#issuecomment-285542083, or mute the thread https://github.com/notifications/unsubscribe-auth/AG7ZuhACwFemZd2dUSiQgDBe6efBtN3qks5rkKKRgaJpZM4MYzCJ .
Thank you, downgrading the matplotlib fixed the connections. Could you tell me what parameter controls the distance between fully connected and hidden layers? That huge gap remained and doesn't look very good
@laur1s actually I'm not able to generate the huge gap you've shown , if i set
size_list = [fc_unit_size, fc_unit_size]
num_list = [500, 2]
if you still have the problem you might play with x_diff_list
to see if it fixes the problem