Issues
- 0
Is `grad_req='add'` possible to replace explicit grad saving and adding?
#10 opened by kaleidoscopical - 0
How to implement DCGAN for 3d input
#9 opened by limberc - 0
Why the fix_gamma = True in the batch_norm
#8 opened by 1292765944 - 2
about training ferr
#7 opened by tornadomeet - 1
dcgan32x32 problem
#6 opened by lucasjinreal - 0
sym_dec = encoder.dcgan(ngf=ngf / 2) File "../mxgan/encoder.py", line 31, in dcgan net = mx.sym.Convolution(data, kernel=(4, 4), stride=(2, 2), pad=(1, 1), num_filter=ngf, name="e1_conv") File "mxnet/cython/symbol.pyx", line 151, in symbol._make_atomic_symbol_function.creator (mxnet/cython/symbol.cpp:3591) File "mxnet/cython/base.pyi", line 36, in symbol.CALL (mxnet/cython/symbol.cpp:1624) mxnet.base.MXNetError: b'Invalid Parameter format for num_filter expect int (non-negative) but value=\'32.0\', in operator Convolution(name="", num_filter="32.0", pad="(1, 1)", kernel="(4, 4)", stride="(2, 2)")'
#5 opened by lucasjinreal - 0
- 2
cudnnv5 support?
#1 opened by SJTUzhanglj