I am trying to integrate a part of the code (configuration/convroll4.py) from the link: https://github.com/benanne/kaggle-ndsb
There is a dimension-mismatch issue at the first layer (cyclic slice layer) of convroll4.py. The line of code is:
l0 = nn.layers.InputLayer((batch_size, 1, patch_size[0], patch_size[1]))
l0c = dihedral.CyclicSliceLayer(l0)
I am getting errors during training at the following line:
line in train vl, va = val_fn(Xb, yb):
Note that this is part of the training loop of the code, the snippet is given below:
for Xb, yb in L_val:
yb = yb.flatten().astype(np.int32)
vl, va = val_fn(Xb, yb)
The errors are as follows:
File "/anaconda/lib/python2.7/site-packages/theano/compile/function_module.py", line 871, in __call__
storage_map=getattr(self.fn, 'storage_map', None))
File "/anaconda/lib/python2.7/site-packages/theano/gof/link.py", line 314, in raise_with_op
reraise(exc_type, exc_value, exc_trace)
File "/anaconda/lib/python2.7/site-packages/theano/compile/function_module.py", line 859, in __call__
outputs = self.fn()
ValueError: dimension mismatch in x,y_idx arguments
Apply node that caused the error: GpuCrossentropySoftmaxArgmax1HotWithBias(GpuDot22.0, b, GpuFromHost.0)
Toposort index: 419
Inputs types: [CudaNdarrayType(float32, matrix), CudaNdarrayType(float32, vector), CudaNdarrayType(float32, vector)]
Inputs shapes: [(32, 2), (2,), (128,)]
Inputs strides: [(2, 1), (1,), (1,)]
Inputs values: ['not shown', CudaNdarray([ 0.05 0.05]), 'not shown']
Outputs clients: [[GpuCAReduce{add}{1}(GpuCrossentropySoftmaxArgmax1HotWithBias.0)], [], []]
HINT: Re-running with most Theano optimization disabled could give you a back-trace of when this node was created.
This can be done with by setting the Theo flag 'optimizer=fast_compile'.
If that does not work, Theano optimizations can be disabled with 'optimizer=None'.
HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint and storage map footprint of this apply node.
This is causing errors within the python package (Note: I am using Theano 0.8.0) rather than the code in the link. Please help give an indication as to what might be the problem.
