The problem is that DnnConv is not tagged properly for float16 support and
so its C code is disabled. It doesn't have a python implementation so it
just crashes.
Try this diff:
index 6e7bc11..28fd96b 100644
--- a/theano/gpuarray/dnn.py
+++ b/theano/gpuarray/dnn.py
@@ -405,7 +405,7 @@ class GpuDnnConv(DnnBase):
Default is the value of :attr:`config.dnn.conv.algo_fwd`.
"""
-
+ _f16_ok = True
__props__ = ('algo', 'inplace')
def __init__(self, algo=None, inplace=False):
2016-07-11 5:00 GMT-04:00 <[email protected]>:
> Hi Pascali,
> I tried what you suggested but nothing changed: id doesn't work with float
> 16.
> Did you asked to Arnaud Bergeron?
>
> Many Thanks
> Luca
>
>
>
>
> On Monday, July 4, 2016 at 10:19:49 AM UTC+2, [email protected] wrote:
>
>> Many thanks Pascal for your help.
>>
>>
>> On Saturday, July 2, 2016 at 4:34:28 AM UTC+2, Pascal Lamblin wrote:
>>
>> Thanks, it helps with formatting :)
>>
>> I am not sure what is happening with the Elemwise, Arnaud Bergeron would
>> be more qualified to answer, he should be back in a week or so.
>>
>> It may be possible that the cuDNN convolutions i or theyr gradient do
>> not support float16 yet.
>>
>> Two other remarks though:
>>
>> - Pooling and its gradient have limited GPU support when using
>> "ignore_border=False", which may explain why they are not transferred to
>> the GPU in your case
>>
>> - The default random sampling functions (available from
>> tensor.shared_randomstreams.RandomStreams) are executed in Python on
>> CPU, using NumPy. You can try theano.sandbox.MRG_RandomStreams instead,
>> which can actually sample on GPU.
>>
>> On Fri, Jul 01, 2016, [email protected] wrote:
>> > I attach the file with the result.
>> > Many Thanks
>> > Luca
>> >
>> > --
>> >
>> > ---
>> > 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.
>>
>> > Python 2.7.11 |Anaconda custom (64-bit)| (default, Dec 6 2015,
>> 18:08:32)
>> > [GCC 4.4.7 20120313 (Red Hat 4.4.7-1)] on linux2
>> > Type "help", "copyright", "credits" or "license" for more information.
>> > Anaconda is brought to you by Continuum Analytics.
>> > Please check out: http://continuum.io/thanks and https://anaconda.org
>> > >>> import run_multi_conv
>> > Mapped name None to device cuda: GeForce 840M
>> > Using cuDNN version 5005 on context None
>> > /home/luca/data/Theano-master/theano/tensor/signal/downsample.py:6:
>> UserWarning: downsample module has been moved to the
>> theano.tensor.signal.pool module.
>> > "downsample module has been moved to the theano.tensor.signal.pool
>> module.")
>> > >>> run_multi_conv.run_experiments()
>> > Disabling C code for Elemwise{mul,no_inplace} due to unsupported
>> float16
>> > Disabling C code for Elemwise{Cast{float32}} due to unsupported float16
>> > Disabling C code for Elemwise{Cast{float16}} due to unsupported float16
>> > Disabling C code for Elemwise{Cast{float16}} due to unsupported float16
>> > Disabling C code for Alloc due to unsupported float16
>> > Disabling C code for Cast{float16} due to unsupported float16
>> > Disabling C code for Cast{float16} due to unsupported float16
>> > Disabling C code for Cast{float16} due to unsupported float16
>> > Disabling C code for Cast{float16} due to unsupported float16
>> > Disabling C code for RandomFunction{binomial} due to unsupported
>> float16
>> > Disabling C code for RandomFunction{binomial} due to unsupported
>> float16
>> > Disabling C code for GpuDnnConv{algo='small', inplace=True} due to
>> unsupported float16
>> > Disabling C code for DiagonalSubtensor{inplace} due to unsupported
>> float16
>> > Disabling C code for Pool{ds=(4, 4), ignore_border=False, st=(4, 4),
>> padding=(0, 0), mode='max'} due to unsupported float16
>> > Disabling C code for Pool{ds=(1, 4), ignore_border=False, st=(1, 4),
>> padding=(0, 0), mode='max'} due to unsupported float16
>> > Disabling C code for MaxPoolGrad{ds=(1, 4), ignore_border=False, st=(1,
>> 4), padding=(0, 0), mode='max'} due to unsupported float16
>> > Disabling C code for MaxPoolGrad{ds=(4, 4), ignore_border=False, st=(4,
>> 4), padding=(0, 0), mode='max'} due to unsupported float16
>> > Disabling C code for IncDiagonalSubtensor due to unsupported float16
>> > Disabling C code for GpuDnnConvGradW{algo='none', inplace=True} due to
>> unsupported float16
>> > HostFromGpu(gpuarray) [id A] <TensorType(float16, scalar)> '' 150
>> > |GpuElemwise{Composite{((-Cast{float16}(((-i0) / i1))) + (i2 * i3) +
>> (i2 * i4))}}[(0, 0)]<gpuarray> [id B] <GpuArrayType<None>(float16, ())> ''
>> 147
>> > |GpuCAReduceCuda{add} [id C] <GpuArrayType<None>(float16, ())> ''
>> 135
>> > | |GpuCrossentropySoftmaxArgmax1HotWithBias.0 [id D]
>> <GpuArrayType<None>(float16, (False,))> '' 133
>> > | |Rebroadcast{0} [id E] <GpuArrayType<None>(float16, (False,
>> False))> '' 132
>> > | | |GpuGemm{inplace=True} [id F] <GpuArrayType<None>(float16,
>> (True, False))> '' 130
>> > | | |GpuAllocEmpty{dtype='float16', context_name=None} [id G]
>> <GpuArrayType<None>(float16, (True, False))> '' 20
>> > | | | |TensorConstant{1} [id H] <TensorType(int64, scalar)>
>> > | | | |Shape_i{1} [id I] <TensorType(int64, scalar)> '' 2
>> > | | | |DropoutLogisticRegression_W [id J]
>> <GpuArrayType<None>(float16, (False, False))>
>> > | | |TensorConstant{1.0} [id K] <TensorType(float16, scalar)>
>> > | | |GpuElemwise{mul,no_inplace} [id L]
>> <GpuArrayType<None>(float16, (False, False))> '' 129
>> > | | | |GpuElemwise{Cast{float16}}[]<gpuarray> [id M]
>> <GpuArrayType<None>(float16, (False, False))> '' 40
>> > | | | | |GpuFromHost<None> [id N] <GpuArrayType<None>(int64,
>> (False, False))> '' 35
>> > | | | | |RandomFunction{binomial}.1 [id O] <TensorType(int64,
>> matrix)> '' 30
>> > | | | | |<RandomStateType> [id P] <RandomStateType>
>> > | | | | |MakeVector{dtype='int64'} [id Q] <TensorType(int64,
>> vector)> '' 22
>> > | | | | | |TensorConstant{1} [id H] <TensorType(int64,
>> scalar)>
>> > | | | | | |Shape_i{1} [id R] <TensorType(int64, scalar)> ''
>> 6
>> > | | | | | |DropoutHiddenLayer_1_W [id S]
>> <GpuArrayType<None>(float16, (False, False))>
>> > | | | | |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | | |TensorConstant{0.85009765625} [id U]
>> <TensorType(float16, scalar)>
>> > | | | |GpuElemwise{Composite{scalar_sigmoid((i0 + i1))}}[(0,
>> 0)]<gpuarray> [id V] <GpuArrayType<None>(float16, (False, False))> '' 128
>> > | | | | |Rebroadcast{0} [id W] <GpuArrayType<None>(float16,
>> (False, False))> '' 127
>> > | | | | | |GpuGemm{inplace=True} [id X]
>> <GpuArrayType<None>(float16, (True, False))> '' 125
>> > | | | | | |GpuAllocEmpty{dtype='float16', context_name=None}
>> [id Y] <GpuArrayType<None>(float16, (True, False))> '' 23
>> > | | | | | | |TensorConstant{1} [id H] <TensorType(int64,
>> scalar)>
>> > | | | | | | |Shape_i{1} [id R] <TensorType(int64, scalar)> ''
>> 6
>> > | | | | | |TensorConstant{1.0} [id K] <TensorType(float16,
>> scalar)>
>> > | | | | | |Rebroadcast{0} [id Z] <GpuArrayType<None>(float16,
>> (False, False))> '' 124
>> > | | | | | | |GpuReshape{2} [id BA]
>> <GpuArrayType<None>(float16, (True, False))> '' 123
>> > | | | | | | |InplaceGpuDimShuffle{0,1,4,2,3} [id BB]
>> <GpuArrayType<None>(float16, (False, False, False, False, False))> '' 122
>> > | | | | | | | |Rebroadcast{0} [id BC]
>> <GpuArrayType<None>(float16, (False, False, False, False, False))> '' 121
>> > | | | | | | | |GpuReshape{5} [id BD]
>> <GpuArrayType<None>(float16, (True, False, False, False, False))> '' 120
>> > | | | | | | | |GpuFromHost<None> [id BE]
>> <GpuArrayType<None>(float16, (False, False, False, False))> '' 115
>> > | | | | | | | | |Pool{ds=(1, 4), ignore_border=False,
>> st=(1, 4), padding=(0, 0), mode='max'} [id BF] <TensorType(float16, 4D)> ''
>> 111
>> > | | | | | | | | |Reshape{4} [id BG]
>> <TensorType(float16, 4D)> '' 109
>> > | | | | | | | | |InplaceDimShuffle{0,1,3,4,2} [id
>> BH] <TensorType(float16, 5D)> '' 104
>> > | | | | | | | | | |Reshape{5} [id BI]
>> <TensorType(float16, 5D)> '' 102
>> > | | | | | | | | | |Pool{ds=(4, 4),
>> ignore_border=False, st=(4, 4), padding=(0, 0), mode='max'} [id BJ]
>> <TensorType(float16, 4D)> '' 100
>> > | | | | | | | | | | |HostFromGpu(gpuarray) [id BK]
>> <TensorType(float16, 4D)> '' 98
>> > | | | | | | | | | | |Gpusigmoid [id BL]
>> <GpuArrayType<None>(float16, (False, False, False, False))> '' 92
>> > | | | | | | | | | | |GpuReshape{4} [id BM]
>> <GpuArrayType<None>(float16, (False, False, False, False))> '' 89
>> > | | | | | | | | | | |GpuElemwise{Add}[(0,
>> 0)]<gpuarray> [id BN] <GpuArrayType<None>(float16, (True, False, False,
>> False, False))> '' 85
>> > | | | | | | | | | | |
>> |InplaceGpuDimShuffle{0,2,4,3,1} [id BO] <GpuArrayType<None>(float16,
>> (True, False, False, False, False))> '' 83
>> > | | | | | | | | | | | | |GpuReshape{5} [id
>> BP] <GpuArrayType<None>(float16, (True, False, False, False, False))> ''
>> 78
>> > | | | | | | | | | | | |
>> |GpuCAReduceCuda{add}{1} [id BQ] <GpuArrayType<None>(float16, (False,
>> False))> '' 76
>> > | | | | | | | | | | | | | |GpuReshape{3}
>> [id BR] <GpuArrayType<None>(float16, (False, False, False))> '' 74
>> > | | | | | | | | | | | | |
>> |GpuFromHost<None> [id BS] <GpuArrayType<None>(float16, (True, False,
>> False, False, False, False))> '' 64
>> > | | | | | | | | | | | | | |
>> |DiagonalSubtensor{inplace} [id BT] <TensorType(float16, (True, False,
>> False, False, False, False))> '' 58
>> > | | | | | | | | | | | | | |
>> |HostFromGpu(gpuarray) [id BU] <TensorType(float16, (True, False, False,
>> False, False, False))> '' 57
>> > | | | | | | | | | | | | | | |
>> |GpuReshape{6} [id BV] <GpuArrayType<None>(float16, (True, False, False,
>> False, False, False))> '' 56
>> > | | | | | | | | | | | | | | |
>> |GpuDnnConv{algo='small', inplace=True} [id BW]
>> <GpuArrayType<None>(float16, (False, False, False, False))> '' 55
>> > | | | | | | | | | | | | | | | |
>> |GpuContiguous [id BX] <GpuArrayType<None>(float16, (False, True, False,
>> False))> '' 45
>> > | | | | | | | | | | | | | | | | |
>> |GpuReshape{4} [id BY] <GpuArrayType<None>(float16, (False, True, False,
>> False))> '' 37
>> > | | | | | | | | | | | | | | | | |
>> |InplaceGpuDimShuffle{0,4,1,3,2} [id BZ] <GpuArrayType<None>(float16,
>> (True, False, True, False, False))> '' 32
>> > | | | | | | | | | | | | | | | | |
>> | |GpuReshape{5} [id CA] <GpuArrayType<None>(float16, (True, True, False,
>> False, False))> '' 27
>> > | | | | | | | | | | | | | | | | |
>> | |GpuFromHost<None> [id CB] <GpuArrayType<None>(float16, (False, False,
>> False, False))> '' 15
>> > | | | | | | | | | | | | | | | | |
>> | | |x [id CC] <TensorType(float16, 4D)>
>> > | | | | | | | | | | | | | | | | |
>> | |TensorConstant{[ 1 1 20 20 20]} [id CD] <TensorType(int64, vector)>
>> > | | | | | | | | | | | | | | | | |
>> |TensorConstant{[20 1 20 20]} [id CE] <TensorType(int64, vector)>
>> > | | | | | | | | | | | | | | | |
>> |GpuContiguous [id CF] <GpuArrayType<None>(float16, (False, True, False,
>> False))> '' 31
>> > | | | | | | | | | | | | | | | | |
>> |GpuReshape{4} [id CG] <GpuArrayType<None>(float16, (False, True, False,
>> False))> '' 25
>> > | | | | | | | | | | | | | | | | |
>> |InplaceGpuDimShuffle{0,4,1,3,2} [id CH] <GpuArrayType<None>(float16,
>> (False, False, False, False, False))> '' 9
>> > | | | | | | | | | | | | | | | | |
>> | |Conv_layer_1_W [id CI] <GpuArrayType<None>(float16, (False, False,
>> False, False, False))>
>> > | | | | | | | | | | | | | | | | |
>> |TensorConstant{[100 1 5 5]} [id CJ] <TensorType(int64, vector)>
>> > | | | | | | | | | | | | | | | |
>> |GpuAllocEmpty{dtype='float16', context_name=None} [id CK]
>> <GpuArrayType<None>(float16, (False, False, False, False))> '' 54
>> > | | | | | | | | | | | | | | | | |
>> |Subtensor{int64} [id CL] <TensorType(int64, scalar)> '' 51
>> > | | | | | | | | | | | | | | | | | |
>> |Shape [id CM] <TensorType(int64, vector)> '' 47
>> > | | | | | | | | | | | | | | | | | |
>> | |GpuContiguous [id BX] <GpuArrayType<None>(float16, (False, True, False,
>> False))> '' 45
>> > | | | | | | | | | | | | | | | | | |
>> |Constant{0} [id CN] <int64>
>> > | | | | | | | | | | | | | | | | |
>> |Subtensor{int64} [id CO] <TensorType(int64, scalar)> '' 44
>> > | | | | | | | | | | | | | | | | | |
>> |Shape [id CP] <TensorType(int64, vector)> '' 36
>> > | | | | | | | | | | | | | | | | | |
>> | |GpuContiguous [id CF] <GpuArrayType<None>(float16, (False, True, False,
>> False))> '' 31
>> > | | | | | | | | | | | | | | | | | |
>> |Constant{0} [id CN] <int64>
>> > | | | | | | | | | | | | | | | | |
>> |Elemwise{Composite{((((i0 + i1) - i2) // i3) + i4)}} [id CQ]
>> <TensorType(int64, scalar)> '' 53
>> > | | | | | | | | | | | | | | | | | |
>> |Subtensor{int64} [id CR] <TensorType(int64, scalar)> '' 50
>> > | | | | | | | | | | | | | | | | | |
>> | |Shape [id CM] <TensorType(int64, vector)> '' 47
>> > | | | | | | | | | | | | | | | | | |
>> | |Constant{2} [id CS] <int64>
>> > | | | | | | | | | | | | | | | | | |
>> |TensorConstant{0} [id CT] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | | | | | | |
>> |Subtensor{int64} [id CU] <TensorType(int64, scalar)> '' 43
>> > | | | | | | | | | | | | | | | | | |
>> | |Shape [id CP] <TensorType(int64, vector)> '' 36
>> > | | | | | | | | | | | | | | | | | |
>> | |Constant{2} [id CS] <int64>
>> > | | | | | | | | | | | | | | | | | |
>> |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | | | | | | |
>> |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | | | | | |
>> |Elemwise{Composite{((((i0 + i1) - i2) // i3) + i4)}} [id CV]
>> <TensorType(int64, scalar)> '' 52
>> > | | | | | | | | | | | | | | | | |
>> |Subtensor{int64} [id CW] <TensorType(int64, scalar)> '' 49
>> > | | | | | | | | | | | | | | | | |
>> | |Shape [id CM] <TensorType(int64, vector)> '' 47
>> > | | | | | | | | | | | | | | | | |
>> | |Constant{3} [id CX] <int64>
>> > | | | | | | | | | | | | | | | | |
>> |TensorConstant{0} [id CT] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | | | | | |
>> |Subtensor{int64} [id CY] <TensorType(int64, scalar)> '' 42
>> > | | | | | | | | | | | | | | | | |
>> | |Shape [id CP] <TensorType(int64, vector)> '' 36
>> > | | | | | | | | | | | | | | | | |
>> | |Constant{3} [id CX] <int64>
>> > | | | | | | | | | | | | | | | | |
>> |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | | | | | |
>> |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | | | | |
>> |GpuDnnConvDesc{border_mode='valid', subsample=(1, 1), conv_mode='conv',
>> precision='float16'} [id CZ] <CDataType{cudnnConvolutionDescriptor_t}> ''
>> 41
>> > | | | | | | | | | | | | | | | | |
>> |Shape [id CP] <TensorType(int64, vector)> '' 36
>> > | | | | | | | | | | | | | | | |
>> |Constant{1.0} [id DA] <float16>
>> > | | | | | | | | | | | | | | | |
>> |Constant{0.0} [id DB] <float16>
>> > | | | | | | | | | | | | | | |
>> |TensorConstant{[ 1 20 20 5 16 16]} [id DC] <TensorType(int64, vector)>
>> > | | | | | | | | | | | | | |
>> |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | | |
>> |TensorConstant{3} [id DD] <TensorType(int8, scalar)>
>> > | | | | | | | | | | | | |
>> |MakeVector{dtype='int64'} [id DE] <TensorType(int64, vector)> '' 72
>> > | | | | | | | | | | | | |
>> |Elemwise{mul,no_inplace} [id DF] <TensorType(int64, scalar)> '' 69
>> > | | | | | | | | | | | | | |
>> |Shape_i{2} [id DG] <TensorType(int64, scalar)> '' 62
>> > | | | | | | | | | | | | | | |
>> |DiagonalSubtensor{inplace} [id BT] <TensorType(float16, (True, False,
>> False, False, False, False))> '' 58
>> > | | | | | | | | | | | | | |
>> |TensorConstant{1} [id H] <TensorType(int64, scalar)>
>> > | | | | | | | | | | | | | |
>> |Shape_i{1} [id DH] <TensorType(int64, scalar)> '' 60
>> > | | | | | | | | | | | | | |
>> |DiagonalSubtensor{inplace} [id BT] <TensorType(float16, (True, False,
>> False, False, False, False))> '' 58
>> > | | | | | | | | | | | | | |Shape_i{3}
>> [id DI] <TensorType(int64, scalar)> '' 63
>> > | | | | | | | | | | | | | |
>> |DiagonalSubtensor{inplace} [id BT] <TensorType(float16, (True, False,
>> False, False, False, False))> '' 58
>> > | | | | | | | | | | | | |
>> |Elemwise{mul,no_inplace} [id DJ] <TensorType(int64, scalar)> '' 66
>> > | | | | | | | | | | | | |
>> |Shape_i{4} [id DK] <TensorType(int64, scalar)> '' 61
>> > | | | | | | | | | | | | | |
>> |DiagonalSubtensor{inplace} [id BT] <TensorType(float16, (True, False,
>> False, False, False, False))> '' 58
>> > | | | | | | | | | | | | |
>> |Shape_i{5} [id DL] <TensorType(int64, scalar)> '' 59
>> > | | | | | | | | | | | | |
>> |DiagonalSubtensor{inplace} [id BT] <TensorType(float16, (True, False,
>> False, False, False, False))> '' 58
>> > | | | | | | | | | | | |
>> |MakeVector{dtype='int64'} [id DM] <TensorType(int64, vector)> '' 68
>> > | | | | | | | | | | | |
>> |TensorConstant{1} [id H] <TensorType(int64, scalar)>
>> > | | | | | | | | | | | | |Shape_i{1} [id
>> DH] <TensorType(int64, scalar)> '' 60
>> > | | | | | | | | | | | | |Shape_i{2} [id
>> DG] <TensorType(int64, scalar)> '' 62
>> > | | | | | | | | | | | | |Shape_i{4} [id
>> DK] <TensorType(int64, scalar)> '' 61
>> > | | | | | | | | | | | | |Shape_i{5} [id
>> DL] <TensorType(int64, scalar)> '' 59
>> > | | | | | | | | | | |
>> |InplaceGpuDimShuffle{x,0,x,x,x} [id DN] <GpuArrayType<None>(float16,
>> (True, False, True, True, True))> '' 16
>> > | | | | | | | | | | | |Conv_layer_1_b [id
>> DO] <GpuArrayType<None>(float16, (False,))>
>> > | | | | | | | | | | |Elemwise{Cast{int32}}
>> [id DP] <TensorType(int32, vector)> '' 77
>> > | | | | | | | | | | |Join [id DQ]
>> <TensorType(int64, vector)> '' 75
>> > | | | | | | | | | | |TensorConstant{0}
>> [id CT] <TensorType(int8, scalar)>
>> > | | | | | | | | | |
>> |InplaceDimShuffle{x} [id DR] <TensorType(int64, (True,))> '' 73
>> > | | | | | | | | | | |
>> |Prod{acc_dtype=int64} [id DS] <TensorType(int64, scalar)> '' 71
>> > | | | | | | | | | | |
>> |MakeVector{dtype='int64'} [id DT] <TensorType(int64, vector)> '' 67
>> > | | | | | | | | | | |
>> |TensorConstant{1} [id H] <TensorType(int64, scalar)>
>> > | | | | | | | | | | | |Shape_i{2}
>> [id DG] <TensorType(int64, scalar)> '' 62
>> > | | | | | | | | | | | |Shape_i{5}
>> [id DL] <TensorType(int64, scalar)> '' 59
>> > | | | | | | | | | | |TensorConstant{(1,)
>> of 1} [id DU] <TensorType(int64, (True,))>
>> > | | | | | | | | | |
>> |MakeVector{dtype='int64'} [id DV] <TensorType(int64, vector)> '' 65
>> > | | | | | | | | | | |Shape_i{4} [id
>> DK] <TensorType(int64, scalar)> '' 61
>> > | | | | | | | | | | |Shape_i{1} [id
>> DH] <TensorType(int64, scalar)> '' 60
>> > | | | | | | | | | |MakeVector{dtype='int64'} [id
>> DW] <TensorType(int64, vector)> '' 95
>> > | | | | | | | | | |TensorConstant{1} [id H]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | | |Shape_i{2} [id DG]
>> <TensorType(int64, scalar)> '' 62
>> > | | | | | | | | | |Shape_i{5} [id DL]
>> <TensorType(int64, scalar)> '' 59
>> > | | | | | | | | | |Elemwise{Composite{(i0 + ((i1
>> + Switch(EQ(i2, i3), i4, i2)) // i5))}} [id DX] <TensorType(int64, scalar)>
>> '' 88
>> > | | | | | | | | | | |TensorConstant{1} [id H]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | | | |TensorConstant{-1} [id DY]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | | | |Subtensor{int64} [id DZ]
>> <TensorType(int32, scalar)> '' 80
>> > | | | | | | | | | | | |Elemwise{Cast{int32}} [id
>> DP] <TensorType(int32, vector)> '' 77
>> > | | | | | | | | | | | |Constant{2} [id CS]
>> <int64>
>> > | | | | | | | | | | |TensorConstant{-1} [id EA]
>> <TensorType(int8, scalar)>
>> > | | | | | | | | | | |Elemwise{Composite{((i0 *
>> i1 * i2 * i3) // (-(i4 * i5 * i6 * i7)))}} [id EB] <TensorType(int64,
>> scalar)> '' 84
>> > | | | | | | | | | | | |Shape_i{2} [id DG]
>> <TensorType(int64, scalar)> '' 62
>> > | | | | | | | | | | | |Shape_i{5} [id DL]
>> <TensorType(int64, scalar)> '' 59
>> > | | | | | | | | | | | |Shape_i{4} [id DK]
>> <TensorType(int64, scalar)> '' 61
>> > | | | | | | | | | | | |Shape_i{1} [id DH]
>> <TensorType(int64, scalar)> '' 60
>> > | | | | | | | | | | | |Subtensor{int64} [id EC]
>> <TensorType(int32, scalar)> '' 82
>> > | | | | | | | | | | | | |Elemwise{Cast{int32}}
>> [id DP] <TensorType(int32, vector)> '' 77
>> > | | | | | | | | | | | | |Constant{0} [id CN]
>> <int64>
>> > | | | | | | | | | | | |Subtensor{int64} [id ED]
>> <TensorType(int32, scalar)> '' 81
>> > | | | | | | | | | | | | |Elemwise{Cast{int32}}
>> [id DP] <TensorType(int32, vector)> '' 77
>> > | | | | | | | | | | | | |Constant{1} [id EE]
>> <int64>
>> > | | | | | | | | | | | |Subtensor{int64} [id DZ]
>> <TensorType(int32, scalar)> '' 80
>> > | | | | | | | | | | | |Subtensor{int64} [id EF]
>> <TensorType(int32, scalar)> '' 79
>> > | | | | | | | | | | | |Elemwise{Cast{int32}}
>> [id DP] <TensorType(int32, vector)> '' 77
>> > | | | | | | | | | | | |Constant{3} [id CX]
>> <int64>
>> > | | | | | | | | | | |TensorConstant{4} [id EG]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | | |Elemwise{Composite{(i0 + ((i1
>> + Switch(EQ(i2, i3), i4, i2)) // i5))}}[(0, 4)] [id EH] <TensorType(int64,
>> scalar)> '' 90
>> > | | | | | | | | | |TensorConstant{1} [id H]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | | |TensorConstant{-1} [id DY]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | | |Subtensor{int64} [id EF]
>> <TensorType(int32, scalar)> '' 79
>> > | | | | | | | | | |TensorConstant{-1} [id EA]
>> <TensorType(int8, scalar)>
>> > | | | | | | | | | |Elemwise{Composite{((i0 *
>> i1 * i2 * i3) // (-(i4 * i5 * i6 * i7)))}} [id EB] <TensorType(int64,
>> scalar)> '' 84
>> > | | | | | | | | | |TensorConstant{4} [id EG]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | |Elemwise{Cast{int32}} [id EI]
>> <TensorType(int32, vector)> '' 103
>> > | | | | | | | | |Join [id EJ] <TensorType(int64,
>> vector)> '' 101
>> > | | | | | | | | |TensorConstant{0} [id CT]
>> <TensorType(int8, scalar)>
>> > | | | | | | | | |InplaceDimShuffle{x} [id EK]
>> <TensorType(int64, (True,))> '' 99
>> > | | | | | | | | | |Prod{acc_dtype=int64} [id EL]
>> <TensorType(int64, scalar)> '' 97
>> > | | | | | | | | | |MakeVector{dtype='int64'}
>> [id EM] <TensorType(int64, vector)> '' 91
>> > | | | | | | | | | |TensorConstant{1} [id H]
>> <TensorType(int64, scalar)>
>> > | | | | | | | | | |Shape_i{2} [id DG]
>> <TensorType(int64, scalar)> '' 62
>> > | | | | | | | | | |Elemwise{Composite{(i0 +
>> ((i1 + Switch(EQ(i2, i3), i4, i2)) // i5))}} [id DX] <TensorType(int64,
>> scalar)> '' 88
>> > | | | | | | | | |TensorConstant{(1,) of 1} [id
>> DU] <TensorType(int64, (True,))>
>> > | | | | | | | | |MakeVector{dtype='int64'} [id
>> EN] <TensorType(int64, vector)> '' 93
>> > | | | | | | | | |Elemwise{Composite{(i0 + ((i1
>> + Switch(EQ(i2, i3), i4, i2)) // i5))}}[(0, 4)] [id EH] <TensorType(int64,
>> scalar)> '' 90
>> > | | | | | | | | |Shape_i{5} [id DL]
>> <TensorType(int64, scalar)> '' 59
>> > | | | | | | | |MakeVector{dtype='int64'} [id EO]
>> <TensorType(int64, vector)> '' 117
>> > | | | | | | | |TensorConstant{1} [id H]
>> <TensorType(int64, scalar)>
>> > | | | | | | | |Shape_i{2} [id DG] <TensorType(int64,
>> scalar)> '' 62
>> > | | | | | | | |Elemwise{Composite{(i0 + ((i1 +
>> Switch(EQ(i2, i3), i4, i2)) // i5))}} [id DX] <TensorType(int64, scalar)>
>> '' 88
>> > | | | | | | | |Elemwise{Composite{Switch(EQ(i0, i1),
>> i2, i0)}} [id EP] <TensorType(int64, scalar)> '' 112
>> > | | | | | | | | |Subtensor{int64} [id EQ]
>> <TensorType(int32, scalar)> '' 106
>> > | | | | | | | | | |Elemwise{Cast{int32}} [id EI]
>> <TensorType(int32, vector)> '' 103
>> > | | | | | | | | | |Constant{2} [id CS] <int64>
>> > | | | | | | | | |TensorConstant{-1} [id EA]
>> <TensorType(int8, scalar)>
>> > | | | | | | | | |Elemwise{Composite{((i0 * i1 * i2 *
>> i3) // (-(i4 * i5 * i6 * i7)))}}[(0, 2)] [id ER] <TensorType(int64,
>> scalar)> '' 110
>> > | | | | | | | | |Shape_i{2} [id DG]
>> <TensorType(int64, scalar)> '' 62
>> > | | | | | | | | |Elemwise{Composite{(i0 + ((i1 +
>> Switch(EQ(i2, i3), i4, i2)) // i5))}} [id DX] <TensorType(int64, scalar)>
>> '' 88
>> > | | | | | | | | |Elemwise{Composite{(i0 + ((i1 +
>> Switch(EQ(i2, i3), i4, i2)) // i5))}}[(0, 4)] [id EH] <TensorType(int64,
>> scalar)> '' 90
>> > | | | | | | | | |Shape_i{5} [id DL]
>> <TensorType(int64, scalar)> '' 59
>> > | | | | | | | | |Subtensor{int64} [id ES]
>> <TensorType(int32, scalar)> '' 108
>> > | | | | | | | | | |Elemwise{Cast{int32}} [id EI]
>> <TensorType(int32, vector)> '' 103
>> > | | | | | | | | | |Constant{0} [id CN] <int64>
>> > | | | | | | | | |Subtensor{int64} [id ET]
>> <TensorType(int32, scalar)> '' 107
>> > | | | | | | | | | |Elemwise{Cast{int32}} [id EI]
>> <TensorType(int32, vector)> '' 103
>> > | | | | | | | | | |Constant{1} [id EE] <int64>
>> > | | | | | | | | |Subtensor{int64} [id EQ]
>> <TensorType(int32, scalar)> '' 106
>> > | | | | | | | | |Subtensor{int64} [id EU]
>> <TensorType(int32, scalar)> '' 105
>> > | | | | | | | | |Elemwise{Cast{int32}} [id EI]
>> <TensorType(int32, vector)> '' 103
>> > | | | | | | | | |Constant{3} [id CX] <int64>
>> > | | | | | | | |Elemwise{Composite{(i0 + ((i1 +
>> Switch(EQ(i2, i3), i4, i2)) // i5))}}[(0, 4)] [id EV] <TensorType(int64,
>> scalar)> '' 116
>> > | | | | | | | |TensorConstant{1} [id H]
>> <TensorType(int64, scalar)>
>> > | | | | | | | |TensorConstant{-1} [id DY]
>> <TensorType(int64, scalar)>
>> > | | | | | | | |Subtensor{int64} [id EU]
>> <TensorType(int32, scalar)> '' 105
>> > | | | | | | | |TensorConstant{-1} [id EA]
>> <TensorType(int8, scalar)>
>> > | | | | | | | |Elemwise{Composite{((i0 * i1 * i2 *
>> i3) // (-(i4 * i5 * i6 * i7)))}}[(0, 2)] [id ER] <TensorType(int64,
>> scalar)> '' 110
>> > | | | | | | | |TensorConstant{4} [id EG]
>> <TensorType(int64, scalar)>
>> > | | | | | | |TensorConstant{[ 1 -1]} [id EW]
>> <TensorType(int64, vector)>
>> > | | | | | |DropoutHiddenLayer_1_W [id S]
>> <GpuArrayType<None>(float16, (False, False))>
>> > | | | | | |TensorConstant{0.0} [id EX] <TensorType(float16,
>> scalar)>
>> > | | | | |InplaceGpuDimShuffle{x,0} [id EY]
>> <GpuArrayType<None>(float16, (True, False))> '' 17
>> > | | | | |DropoutHiddenLayer_1_b [id EZ]
>> <GpuArrayType<None>(float16, (False,))>
>> > | | | |GpuElemwise{Cast{float16}}[]<gpuarray> [id FA]
>> <GpuArrayType<None>(float16, (False, False))> '' 39
>> > | | | |GpuFromHost<None> [id FB] <GpuArrayType<None>(int64,
>> (False, False))> '' 34
>> > | | | |RandomFunction{binomial}.1 [id FC] <TensorType(int64,
>> matrix)> '' 29
>> > | | | |<RandomStateType> [id FD] <RandomStateType>
>> > | | | |MakeVector{dtype='int64'} [id Q] <TensorType(int64,
>> vector)> '' 22
>> > | | | |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | |TensorConstant{0.85009765625} [id U]
>> <TensorType(float16, scalar)>
>> > | | |DropoutLogisticRegression_W [id J]
>> <GpuArrayType<None>(float16, (False, False))>
>> > | | |TensorConstant{0.0} [id EX] <TensorType(float16, scalar)>
>> > | |DropoutLogisticRegression_b [id FE]
>> <GpuArrayType<None>(float16, (False,))>
>> > | |GpuFromHost<None> [id FF] <GpuArrayType<None>(int32, (False,))>
>> '' 0
>> > | |y [id FG] <TensorType(int32, vector)>
>> > |GpuElemwise{Cast{float32}}[]<gpuarray> [id FH]
>> <GpuArrayType<None>(float32, ())> '' 28
>> > | |GpuFromHost<None> [id FI] <GpuArrayType<None>(int64, ())> '' 19
>> > | |Shape_i{0} [id FJ] <TensorType(int64, scalar)> '' 1
>> > | |y [id FG] <TensorType(int32, vector)>
>> > |GpuArrayConstant{0.000100016593933} [id FK]
>> <GpuArrayType<None>(float16, ())>
>> > |GpuCAReduceCuda{add} [id FL] <GpuArrayType<None>(float16, ())> ''
>> 144
>> > | |GpuElemwise{Abs}[(0, 0)]<gpuarray> [id FM]
>> <GpuArrayType<None>(float16, (False, False))> '' 142
>> > | |DropoutLogisticRegression_W [id J] <GpuArrayType<None>(float16,
>> (False, False))>
>> > |GpuCAReduceCuda{pre=sqr,red=add} [id FN]
>> <GpuArrayType<None>(float16, ())> '' 5
>> > |DropoutLogisticRegression_W [id J] <GpuArrayType<None>(float16,
>> (False, False))>
>> > GpuElemwise{Composite{(i0 - (i1 * i2))}}[(0, 0)]<gpuarray> [id FO]
>> <GpuArrayType<None>(float16, (False, False, False, False, False))> '' 190
>> > |Conv_layer_1_W [id CI] <GpuArrayType<None>(float16, (False, False,
>> False, False, False))>
>> > |GpuArrayConstant{[[[[[ 0.0010004]]]]]} [id FP]
>> <GpuArrayType<None>(float16, (True, True, True, True, True))>
>> > |InplaceGpuDimShuffle{0,2,4,3,1} [id FQ] <GpuArrayType<None>(float16,
>> (False, False, False, False, False))> '' 189
>> > |GpuReshape{5} [id FR] <GpuArrayType<None>(float16, (False, False,
>> False, False, False))> '' 188
>> > |GpuDnnConvGradW{algo='none', inplace=True} [id FS]
>> <GpuArrayType<None>(float16, (False, True, False, False))> '' 187
>> > | |GpuContiguous [id BX] <GpuArrayType<None>(float16, (False,
>> True, False, False))> '' 45
>> > | |GpuContiguous [id FT] <GpuArrayType<None>(float16, (False,
>> False, False, False))> '' 178
>> > | | |GpuReshape{4} [id FU] <GpuArrayType<None>(float16, (False,
>> False, False, False))> '' 176
>> > | | |GpuFromHost<None> [id FV] <GpuArrayType<None>(float16,
>> (True, False, False, False, False, False))> '' 174
>> > | | | |IncDiagonalSubtensor [id FW] <TensorType(float16, (True,
>> False, False, False, False, False))> '' 172
>> > | | | |TensorConstant{(1, 20, 20..16) of 0.0} [id FX]
>> <TensorType(float16, (True, False, False, False, False, False))>
>> > | | | |TensorConstant{1} [id T] <TensorType(int8, scalar)>
>> > | | | |TensorConstant{3} [id DD] <TensorType(int8, scalar)>
>> > | | | |HostFromGpu(gpuarray) [id FY] <TensorType(float16,
>> (True, False, False, False, False, False))> '' 169
>> > | | | |GpuAlloc<None> [id FZ] <GpuArrayType<None>(float16,
>> (True, False, False, False, False, False))> '' 166
>> > | | | |InplaceGpuDimShuffle{x,4,1,x,3,2} [id GA]
>> <GpuArrayType<None>(float16, (True, False, False, True, False, False))> ''
>> 164
>> > | | | | |GpuElemwise{Composite{Cast{float16}(((i0 -
>> scalar_sigmoid(i1)) * i2 * scalar_sigmoid(i1)))}}[(0, 1)]<gpuarray> [id GB]
>> <GpuArrayType<None>(float16, (True, False, False, False, False))> '' 163
>> > | | | | |GpuArrayConstant{[[[[[ 1.]]]]]} [id GC]
>> <GpuArrayType<None>(float32, (True, True, True, True, True))>
>> > | | | | |GpuElemwise{Add}[(0, 0)]<gpuarray> [id BN]
>> <GpuArrayType<None>(float16, (True, False, False, False, False))> '' 85
>> > | | | | |GpuReshape{5} [id GD]
>> <GpuArrayType<None>(float16, (True, False, False, False, False))> '' 162
>> > | | | | |GpuFromHost<None> [id GE]
>> <GpuArrayType<None>(float16, (False, False, False, False))> '' 161
>> > | | | | | |MaxPoolGrad{ds=(4, 4), ignore_border=False,
>> st=(4, 4), padding=(0, 0), mode='max'} [id GF] <TensorType(float16, 4D)> ''
>> 160
>> > | | | | | |HostFromGpu(gpuarray) [id BK]
>> <TensorType(float16, 4D)> '' 98
>> > | | | | | |Pool{ds=(4, 4), ignore_border=False,
>> st=(4, 4), padding=(0, 0), mode='max'} [id BJ] <TensorType(float16, 4D)> ''
>> 100
>> > | | | | | |Reshape{4} [id GG] <TensorType(float16,
>> 4D)> '' 159
>> > | | | | | |InplaceDimShuffle{0,1,4,2,3} [id GH]
>> <TensorType(float16, 5D)> '' 158
>> > | | | | | | |Reshape{5} [id GI]
>> <TensorType(float16, 5D)> '' 157
>> > | | | | | | |MaxPoolGrad{ds=(1, 4),
>> ignore_border=False, st=(1, 4), padding=(0, 0), mode='max'} [id GJ]
>> <TensorType(float16, 4D)> '' 156
>> > | | | | | | | |Reshape{4} [id BG]
>> <TensorType(float16, 4D)> '' 109
>> > | | | | | | | |Pool{ds=(1, 4),
>> ignore_border=False, st=(1, 4), padding=(0, 0), mode='max'} [id BF]
>> <TensorType(float16, 4D)> '' 111
>> > | | | | | | | |HostFromGpu(gpuarray) [id GK]
>> <TensorType(float16, 4D)> '' 155
>> > | | | | | | | |GpuReshape{4} [id GL]
>> <GpuArrayType<None>(float16, (False, False, False, False))> '' 154
>> > | | | | | | |
>> |InplaceGpuDimShuffle{0,1,3,4,2} [id GM] <GpuArrayType<None>(float16,
>> (False, False, False, False, False))> '' 153
>> > | | | | | | | | |Rebroadcast{0} [id GN]
>> <GpuArrayType<None>(float16, (False, False, False, False, False))> '' 152
>> > | | | | | | | | |GpuReshape{5} [id GO]
>> <GpuArrayType<None>(float16, (True, False, False, False, False))> '' 151
>> > | | | | | | | | |Rebroadcast{0} [id GP]
>> <GpuArrayType<None>(float16, (False, False))> '' 149
>> > | | | | | | | | | |GpuGemm{inplace=True}
>> [id GQ] <GpuArrayType<None>(float16, (True, False))> '' 146
>> > | | | | | | | | |
>> |GpuAllocEmpty{dtype='float16', context_name=None} [id GR]
>> <GpuArrayType<None>(float16, (True, False))> '' 24
>> > | | | | | | | | | | |TensorConstant{1}
>> [id H] <TensorType(int64, scalar)>
>> > | | | | | | | | | | |Shape_i{0} [id
>> GS] <TensorType(int64, scalar)> '' 8
>> > | | | | | | | | | |
>> |DropoutHiddenLayer_1_W [id S] <GpuArrayType<None>(float16, (False,
>> False))>
>> > | | | | | | | | | |TensorConstant{1.0}
>> [id K] <TensorType(float16, scalar)>
>> > | | | | | | | | |
>> |GpuElemwise{Composite{Cast{float16}(((i0 - i1) * i1 * i2 * i3 * i4))}}[(0,
>> 1)]<gpuarray> [id GT] <GpuArrayType<None>(float16, (False, False))> ''
>> 143
>> > | | | | | | | | | |
>> |GpuArrayConstant{[[ 1.]]} [id GU] <GpuArrayType<None>(float32, (True,
>> True))>
>> > | | | | | | | | | |
>> |GpuElemwise{Composite{scalar_sigmoid((i0 + i1))}}[(0, 0)]<gpuarray> [id V]
>> <GpuArrayType<None>(float16, (False, False))> '' 128
>> > | | | | | | | | | |
>> |GpuElemwise{Cast{float16}}[]<gpuarray> [id FA]
>> <GpuArrayType<None>(float16, (False, False))> '' 39
>> > | | | | | | | | | | |Rebroadcast{0}
>> [id GV] <GpuArrayType<None>(float16, (False, False))> '' 141
>> > | | | | | | | | | | |
>> |GpuGemm{inplace=True} [id GW] <GpuArrayType<None>(float16, (True, False))>
>> '' 138
>> > | | | |
>>
>> ...
>
> --
>
> ---
> 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.
