search

J-Cuadrado / OF_EV_SNN

Optical Flow estimation from Event cameras and Spiking Neural Networks
GNU General Public License v3.0
25 stars 3 forks source link

Line 183 train 3dnet cat #3

Closed zig-zagreus closed 1 year ago

zig-zagreus commented 1 year ago

hi, for line 183 for the train 3dNet cat, the raise is being read as an error, saying that "No exception to reraise"

J-Cuadrado commented 1 year ago

Hello, that exception is raised when gradients disappear in training, usually due to not enough activation at a certain point within the network. We have found that a way to counteract this problem is by increasing the MultiplyFactor value, since it is equivalent to reducing the firing threshold and therefore encourages activity within the network. I hope this is of help to you, do not hesitate to contact us for further queries (:

zig-zagreus commented 1 year ago

Brilliant thank you so much for letting me know!