theano - Adaptive learning rate Lasagne -
i using lasagne , theano library build own deep learning model following mnist example. can please tell me how adaptively change learning rate?
i recommend having @ https://github.com/lasagne/lasagne/blob/master/lasagne/updates.py.
if using sgd, can use momentum term (e.g. https://github.com/lasagne/lasagne/blob/master/lasagne/updates.py#l156) adaptively change learning rate. if want make non-standard, momentum implementation give enough hints how create similar on own.
Comments
Post a Comment