Python Layer Error

[Jumabek Alikhanov]

Hi!
I need your help to find out what is the cause of error and how to solve it.

I created python layer for my siamese network(Prototxt included),
 when I train I am receiving following error:

I0918 13:13:56.165798 12587 net.cpp:434] fc8_p <- fc7_p
I0918 13:13:56.165804 12587 net.cpp:408] fc8_p -> fc8_p
I0918 13:13:56.267784 12587 net.cpp:150] Setting up fc8_p
I0918 13:13:56.267819 12587 net.cpp:157] Top shape: 256 1000 (256000)
I0918 13:13:56.267824 12587 net.cpp:165] Memory required for data: 2731155456
F0918 13:13:56.267841 12587 net.cpp:88] Check failed: layer_param.propagate_down_size() == layer_param.bottom_size() (1 vs. 2) propagate_down param must be specified either 0 or bottom_size times
*** Check failure stack trace: ***
    @     0x7f1531e73daa  (unknown)
    @     0x7f1531e73ce4  (unknown)
    @     0x7f1531e736e6  (unknown)
    @     0x7f1531e76687  (unknown)
    @     0x7f1532587acc  caffe::Net<>::Init()
    @     0x7f1532588ff5  caffe::Net<>::Net()
    @     0x7f153250763f  caffe::Solver<>::InitTrainNet()
    @     0x7f1532507b5b  caffe::Solver<>::Init()
    @     0x7f1532507fc8  caffe::Solver<>::Solver()
    @     0x7f153251e543  caffe::Creator_SGDSolver<>()
    @           0x416ce0  caffe::SolverRegistry<>::CreateSolver()
    @           0x40f859  train()
    @           0x40c384  main
    @     0x7f1530666f45  (unknown)
    @           0x40c455  (unknown)
    @              (nil)  (unknown)

I am also attaching my prototxt and PythonLayer files:
Thanks!

[Nathan Ing]

Hello, I am coming across a similar error. Have you solved it yet?

[Nathan Ing]

I've just overcome this error by adding a second line of "propagate_down: false" to the Python layer definition

[anis....@shopedia.fr]

Hi all
please can you  help to solve the same problem layer_param.propagate_down_size() == layer_param.bottom_size() (1 vs. 2) propagate_down param must be specified either 0 or bottom_size times
thank your per advance

ayer_param.propagate_down_size() == layer_param.bottom_size() (1 vs. 2) 

[anis....@shopedia.fr]

it solved by adding !!!!

propagate_down: false propagate_down: false at the place of a single propagate_down: false