you can use get_value() and set_value() on the shared variable to save and restaure the weights:
http://deeplearning.net/software/theano/tutorial/examples.html#using-shared-variables Fred On Wed, Jul 20, 2016 at 4:56 AM, Beatriz G. <[email protected]> wrote: > do you solved your problem? > > I am trying to modify Lenet in order to separate trian and test, so I need > to train the CNN, save de parameters(w, and b of each layer) and then, in > the test, create the CNN with those parameters and test the images. and I > do not know how to crete a new cnn with the parameters loadesd after saving > it. > > El viernes, 12 de junio de 2015, 19:01:53 (UTC+2), André Lopes escribió: >> >> Hi Group! >> Im a bit new to python so im having difficulties to save my ConvNet. >> I know i have to pickle it but i dont know how. >> >> Can anyone help me? >> >> >> https://bitbucket.org/andrelopes1705/theanoconv/src/7128ed2d95d7a432f9a01542f9404d20f59ad65f/main/convolutional_mlp.py?at=default >> >> Im following the tutorial on the deeplearning.net. >> And i tried my own pack of 20 images just to test it. >> >> I saw an old question here about saving and loading but according to the >> replies, it was wrong. So im questioning this again and i will post my >> solution to help others with the same problem. >> >> Thanks a lot! >> >> -- > > --- > You received this message because you are subscribed to the Google Groups > "theano-users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. > -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
