cli/jobs/parallel/3a_mnist_batch_identification/mnist

�� : Add x"T y"T z"T" Ttype: 2 � ApplyGradientDescent var"T� alpha"T delta"T out"T�" Ttype: 2 " use_lockingbool( x Assign ref"T� value"T output_ref"T�" Ttype" validate_shapebool(" use_lockingbool(� ~ BiasAdd value"T bias"T output"T" Ttype: 2 "- data_formatstringNHWC: NHWCNCHW ~ BiasAddGrad out_backprop"T output"T" Ttype: 2 "- data_formatstringNHWC: NHWCNCHW N Cast x"SrcT y"DstT" SrcTtype" DstTtype" Truncatebool( 8 Const output"dtype" valuetensor" dtypetype W ExpandDims input"T dim"Tdim output"T" Ttype" Tdimtype0: 2 ^ Fill dims" index_type value"T output"T" Ttype" index_typetype0: 2 ? FloorDiv x"T y"T z"T" Ttype: 2 . Identity input"T output"T" Ttype W InTopKV2 predictions targets"T k"T precision " Ttype0: 2 q MatMul a"T b"T product"T" transpose_abool(" transpose_bbool(" Ttype: 2 ; Maximum x"T y"T z"T" Ttype: 2 � � Mean input"T reduction_indices"Tidx output"T" keep_dimsbool(" Ttype: 2 " Tidxtype0: 2 = Mul x"T y"T z"T" Ttype: 2 � NoOp C Placeholder output"dtype" dtypetype" shapeshape: X PlaceholderWithDefault input"dtype output"dtype" dtypetype" shapeshape L PreventGradient input"T output"T" Ttype" messagestring � Prod input"T reduction_indices"Tidx output"T" keep_dimsbool(" Ttype: 2 " Tidxtype0: 2 ~ RandomUniform shape"T output"dtype" seedint" seed2int" dtypetype: 2" Ttype: 2 � > RealDiv x"T y"T z"T" Ttype: 2 E Relu features"T activations"T" Ttype: 2 V ReluGrad gradients"T features"T backprops"T" Ttype: 2 [ Reshape tensor"T shape"Tshape output"T" Ttype" Tshapetype0: 2 o RestoreV2 prefix tensor_names shape_and_slices tensors2dtypes" dtypes list(type)(0� l SaveV2 prefix tensor_names shape_and_slices tensors2dtypes" dtypes list(type)(0� P Shape input"T output"out_type" Ttype" out_typetype0: 2 � #SparseSoftmaxCrossEntropyWithLogits features"T labels"Tlabels loss"T backprop"T" Ttype: 2" Tlabelstype0 : 2 : Sub x"T y"T z"T" Ttype: 2 c Tile input"T multiples" Tmultiples output"T" Ttype" Tmultiplestype0: 2 s VariableV2 ref"dtype�" shapeshape" dtypetype" containerstring" shared_namestring� & ZerosLike x"T y"T" Ttype*1.13.12b'v1.13.1-0-g6612da8951'�� n network/XPlaceholder*( _output_shapes :��* shape:��* dtype0 N network/yPlaceholder* _output_shapes :* shape:* dtype0 � *h1/kernel/Initializer/random_uniform/shapeConst* _class loc:@h1/kernel* valueB",* dtype0* _output_shapes : � (h1/kernel/Initializer/random_uniform/minConst* _class loc:@h1/kernel* valueB *�]��* dtype0* _output_shapes : � (h1/kernel/Initializer/random_uniform/maxConst* dtype0* _output_shapes :* _class loc:@h1/kernel* valueB *�]�= � 2h1/kernel/Initializer/random_uniform/RandomUniform RandomUniform*h1/kernel/Initializer/random_uniform/shape* T0* _class loc:@h1/kernel* seed2* dtype0* _output_shapes : ��* seed � (h1/kernel/Initializer/random_uniform/subSub(h1/kernel/Initializer/random_uniform/max(h1/kernel/Initializer/random_uniform/min* _output_shapes :* T0* _class loc:@h1/kernel � (h1/kernel/Initializer/random_uniform/mulMul2h1/kernel/Initializer/random_uniform/RandomUniform(h1/kernel/Initializer/random_uniform/sub* T0* _class loc:@h1/kernel* _output_shapes : �� $h1/kernel/Initializer/random_uniformAdd(h1/kernel/Initializer/random_uniform/mul(h1/kernel/Initializer/random_uniform/min* _output_shapes : ��* T0* _class loc:@h1/kernel � h1/kernel VariableV2* _class loc:@h1/kernel* container* shape: ��* dtype0* _output_shapes : ��* shared_name � h1/kernel/AssignAssign h1/kernel$h1/kernel/Initializer/random_uniform* use_locking(* T0* _class loc:@h1/kernel* validate_shape(* _output_shapes : �� n h1/kernel/readIdentity h1/kernel* _output_shapes : ��* T0* _class loc:@h1/kernel � h1/bias/Initializer/zerosConst* _class loc:@h1/bias* valueB�** dtype0* _output_shapes :� � h1/bias VariableV2* container* shape:�* dtype0* _output_shapes :�* shared_name* _class loc:@h1/bias � h1/bias/AssignAssignh1/biash1/bias/Initializer/zeros* use_locking(* T0* _class loc:@h1/bias* validate_shape(* _output_shapes :� c h1/bias/readIdentityh1/bias* _output_shapes :�* T0* _class loc:@h1/bias � network/h1/MatMulMatMul network/Xh1/kernel/read* transpose_a(*( _output_shapes :��* transpose_b(* T0 � network/h1/BiasAddBiasAddnetwork/h1/MatMulh1/bias/read* T0* data_formatNHWC*( _output_shapes :�� ^ network/h1/ReluRelunetwork/h1/BiasAdd* T0*( _output_shapes :�� *h2/kernel/Initializer/random_uniform/shapeConst* _class loc:@h2/kernel* valueB",d* dtype0* _output_shapes : � (h2/kernel/Initializer/random_uniform/minConst* _output_shapes :* _class loc:@h2/kernel* valueB *��* dtype0 � (h2/kernel/Initializer/random_uniform/maxConst* _class loc:@h2/kernel* valueB *��=* dtype0* _output_shapes : � 2h2/kernel/Initializer/random_uniform/RandomUniform RandomUniform*h2/kernel/Initializer/random_uniform/shape* dtype0* _output_shapes : �d* seed* T0* _class loc:@h2/kernel* seed2 � (h2/kernel/Initializer/random_uniform/subSub(h2/kernel/Initializer/random_uniform/max(h2/kernel/Initializer/random_uniform/min* _output_shapes :* T0* _class loc:@h2/kernel � (h2/kernel/Initializer/random_uniform/mulMul2h2/kernel/Initializer/random_uniform/RandomUniform(h2/kernel/Initializer/random_uniform/sub* _class loc:@h2/kernel* _output_shapes : �d* T0 � $h2/kernel/Initializer/random_uniformAdd(h2/kernel/Initializer/random_uniform/mul(h2/kernel/Initializer/random_uniform/min* T0* _class loc:@h2/kernel* _output_shapes : �d � h2/kernel VariableV2* shape: �d* dtype0* _output_shapes : �d* shared_name* _class loc:@h2/kernel* container � h2/kernel/AssignAssign h2/kernel$h2/kernel/Initializer/random_uniform* validate_shape(* _output_shapes : �d* use_locking(* T0* _class loc:@h2/kernel m h2/kernel/readIdentity h2/kernel* T0* _class loc:@h2/kernel* _output_shapes : �d � h2/bias/Initializer/zerosConst* dtype0* _output_shapes :d* _class loc:@h2/bias* valueBd* � h2/bias VariableV2* _output_shapes :d* shared_name* _class loc:@h2/bias* container* shape:d* dtype0 � h2/bias/AssignAssignh2/biash2/bias/Initializer/zeros* _class loc:@h2/bias* validate_shape(* _output_shapes :d* use_locking(* T0 b h2/bias/readIdentityh2/bias* _class loc:@h2/bias* _output_shapes :d* T0 � network/h2/MatMulMatMulnetwork/h1/Reluh2/kernel/read* T0* transpose_a(*' _output_shapes :��d* transpose_b( � network/h2/BiasAddBiasAddnetwork/h2/MatMulh2/bias/read* T0* data_formatNHWC*' _output_shapes :��d ] network/h2/ReluRelunetwork/h2/BiasAdd*' _output_shapes :��d* T0 � .output/kernel/Initializer/random_uniform/shapeConst* _class loc:@output/kernel* valueB"d * dtype0* _output_shapes : � ,output/kernel/Initializer/random_uniform/minConst* _class loc:@output/kernel* valueB *�'o�* dtype0* _output_shapes : � ,output/kernel/Initializer/random_uniform/maxConst* _output_shapes :* _class loc:@output/kernel* valueB *�'o>* dtype0 � 6output/kernel/Initializer/random_uniform/RandomUniform RandomUniform.output/kernel/Initializer/random_uniform/shape* dtype0* _output_shapes :d * seed* T0* _class loc:@output/kernel* seed2 � ,output/kernel/Initializer/random_uniform/subSub,output/kernel/Initializer/random_uniform/max,output/kernel/Initializer/random_uniform/min* T0* _class loc:@output/kernel* _output_shapes : � ,output/kernel/Initializer/random_uniform/mulMul6output/kernel/Initializer/random_uniform/RandomUniform,output/kernel/Initializer/random_uniform/sub* _output_shapes :d * T0* _class loc:@output/kernel � (output/kernel/Initializer/random_uniformAdd,output/kernel/Initializer/random_uniform/mul,output/kernel/Initializer/random_uniform/min* T0* _class loc:@output/kernel* _output_shapes :d � output/kernel VariableV2* shared_name* _class loc:@output/kernel* container* shape :d * dtype0* _output_shapes :d � output/kernel/AssignAssign output/kernel(output/kernel/Initializer/random_uniform* T0* _class loc:@output/kernel* validate_shape(* _output_shapes :d * use_locking( x output/kernel/readIdentity output/kernel* T0* _class loc:@output/kernel* _output_shapes :d � output/bias/Initializer/zerosConst* _class loc:@output/bias* valueB ** dtype0* _output_shapes : � output/bias VariableV2* shared_name* _class loc:@output/bias* container* shape: * dtype0* _output_shapes : � output/bias/AssignAssignoutput/biasoutput/bias/Initializer/zeros* validate_shape(* _output_shapes : * use_locking(* T0* _class loc:@output/bias n output/bias/readIdentityoutput/bias* T0* _class loc:@output/bias* _output_shapes : � network/output/MatMulMatMulnetwork/h2/Reluoutput/kernel/read* transpose_a(*' _output_shapes :�� * transpose_b(* T0 � network/output/BiasAddBiasAddnetwork/output/MatMuloutput/bias/read* data_formatNHWC*' _output_shapes :�� * T0 � /train/SparseSoftmaxCrossEntropyWithLogits/ShapeShape network/y*# _output_shapes : ��* T0 * out_type0 � Mtrain/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits#SparseSoftmaxCrossEntropyWithLogitsnetwork/output/BiasAdd network/y* T0* Tlabels0 *6 _output_shapes$ ": ��:�� U train/ConstConst* _output_shapes :* value B:* dtype0 � train/lossMeanMtrain/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogitstrain/Const* Tidx0* keep_dims(* T0* _output_shapes : X train/gradients/ShapeConst* valueB* dtype0* _output_shapes : ^ train/gradients/grad_ys_0Const* valueB *�?* dtype0* _output_shapes : � train/gradients/FillFilltrain/gradients/Shapetrain/gradients/grad_ys_0* T0* index_type0* _output_shapes : w -train/gradients/train/loss_grad/Reshape/shapeConst* value B:* dtype0* _output_shapes : � 'train/gradients/train/loss_grad/ReshapeReshapetrain/gradients/Fill-train/gradients/train/loss_grad/Reshape/shape* T0* Tshape0* _output_shapes : � %train/gradients/train/loss_grad/ShapeShapeMtrain/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits* _output_shapes :* T0* out_type0 � $train/gradients/train/loss_grad/TileTile'train/gradients/train/loss_grad/Reshape%train/gradients/train/loss_grad/Shape* Tmultiples0* T0*# _output_shapes : �� 'train/gradients/train/loss_grad/Shape_1ShapeMtrain/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits* T0* out_type0* _output_shapes : j 'train/gradients/train/loss_grad/Shape_2Const* valueB* dtype0* _output_shapes : o %train/gradients/train/loss_grad/ConstConst* value B:* dtype0* _output_shapes : � $train/gradients/train/loss_grad/ProdProd'train/gradients/train/loss_grad/Shape_1%train/gradients/train/loss_grad/Const* T0* _output_shapes :* Tidx0* keep_dims( q 'train/gradients/train/loss_grad/Const_1Const* value B:* dtype0* _output_shapes : � &train/gradients/train/loss_grad/Prod_1Prod'train/gradients/train/loss_grad/Shape_2'train/gradients/train/loss_grad/Const_1* _output_shapes :* Tidx0* keep_dims(* T0 k )train/gradients/train/loss_grad/Maximum/yConst* dtype0* _output_shapes :* value B: � 'train/gradients/train/loss_grad/MaximumMaximum&train/gradients/train/loss_grad/Prod_1)train/gradients/train/loss_grad/Maximum/y* _output_shapes :* T0 � (train/gradients/train/loss_grad/floordivFloorDiv$train/gradients/train/loss_grad/Prod'train/gradients/train/loss_grad/Maximum* _output_shapes :* T0 � $train/gradients/train/loss_grad/CastCast(train/gradients/train/loss_grad/floordiv* SrcT0* Truncate(* DstT0* _output_shapes : � 'train/gradients/train/loss_grad/truedivRealDiv$train/gradients/train/loss_grad/Tile$train/gradients/train/loss_grad/Cast*# _output_shapes : ��* T0 � train/gradients/zeros_like ZerosLikeOtrain/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits:1* T0*' _output_shapes :�� rtrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/PreventGradientPreventGradientOtrain/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits:1* T0*' _output_shapes :�� *� message��Currently there is no way to take the second derivative of sparse_softmax_cross_entropy_with_logits due to the fused implementation's interaction with tf.gradients() � qtrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/ExpandDims/dimConst* valueB: ��* dtype0* _output_shapes : � mtrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/ExpandDims ExpandDims'train/gradients/train/loss_grad/truedivqtrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/ExpandDims/dim*' _output_shapes :��* Tdim0* T0 � ftrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/mulMulmtrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/ExpandDimsrtrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/PreventGradient* T0*' _output_shapes :�� 7train/gradients/network/output/BiasAdd_grad/BiasAddGradBiasAddGradftrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/mul* T0* data_formatNHWC* _output_shapes : � <train/gradients/network/output/BiasAdd_grad/tuple/group_depsNoOp8^train/gradients/network/output/BiasAdd_grad/BiasAddGradg^train/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/mul � Dtrain/gradients/network/output/BiasAdd_grad/tuple/control_dependencyIdentityftrain/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/mul=^train/gradients/network/output/BiasAdd_grad/tuple/group_deps* T0*y _classo mkloc:@train/gradients/train/SparseSoftmaxCrossEntropyWithLogits/SparseSoftmaxCrossEntropyWithLogits_grad/mul*' _output_shapes :�� Ftrain/gradients/network/output/BiasAdd_grad/tuple/control_dependency_1Identity7train/gradients/network/output/BiasAdd_grad/BiasAddGrad=^train/gradients/network/output/BiasAdd_grad/tuple/group_deps* T0*J _class@ ><loc:@train/gradients/network/output/BiasAdd_grad/BiasAddGrad* _output_shapes : � 1train/gradients/network/output/MatMul_grad/MatMulMatMulDtrain/gradients/network/output/BiasAdd_grad/tuple/control_dependencyoutput/kernel/read* transpose_a(*' _output_shapes :��d* transpose_b(* T0 � 3train/gradients/network/output/MatMul_grad/MatMul_1MatMulnetwork/h2/ReluDtrain/gradients/network/output/BiasAdd_grad/tuple/control_dependency* T0* transpose_a(* _output_shapes :d * transpose_b( � ;train/gradients/network/output/MatMul_grad/tuple/group_depsNoOp2^train/gradients/network/output/MatMul_grad/MatMul4^train/gradients/network/output/MatMul_grad/MatMul_1 � Ctrain/gradients/network/output/MatMul_grad/tuple/control_dependencyIdentity1train/gradients/network/output/MatMul_grad/MatMul<^train/gradients/network/output/MatMul_grad/tuple/group_deps* T0*D _class: 86loc:@train/gradients/network/output/MatMul_grad/MatMul*' _output_shapes :��d � Etrain/gradients/network/output/MatMul_grad/tuple/control_dependency_1Identity3train/gradients/network/output/MatMul_grad/MatMul_1<^train/gradients/network/output/MatMul_grad/tuple/group_deps* T0*F _class< :8loc:@train/gradients/network/output/MatMul_grad/MatMul_1* _output_shapes :d � -train/gradients/network/h2/Relu_grad/ReluGradReluGradCtrain/gradients/network/output/MatMul_grad/tuple/control_dependencynetwork/h2/Relu* T0*' _output_shapes :��d � 3train/gradients/network/h2/BiasAdd_grad/BiasAddGradBiasAddGrad-train/gradients/network/h2/Relu_grad/ReluGrad* T0* data_formatNHWC* _output_shapes :d � 8train/gradients/network/h2/BiasAdd_grad/tuple/group_depsNoOp4^train/gradients/network/h2/BiasAdd_grad/BiasAddGrad.^train/gradients/network/h2/Relu_grad/ReluGrad � @train/gradients/network/h2/BiasAdd_grad/tuple/control_dependencyIdentity-train/gradients/network/h2/Relu_grad/ReluGrad9^train/gradients/network/h2/BiasAdd_grad/tuple/group_deps* T0*@ _class6 42loc:@train/gradients/network/h2/Relu_grad/ReluGrad*' _output_shapes :��d � Btrain/gradients/network/h2/BiasAdd_grad/tuple/control_dependency_1Identity3train/gradients/network/h2/BiasAdd_grad/BiasAddGrad9^train/gradients/network/h2/BiasAdd_grad/tuple/group_deps* T0*F _class< :8loc:@train/gradients/network/h2/BiasAdd_grad/BiasAddGrad* _output_shapes :d � -train/gradients/network/h2/MatMul_grad/MatMulMatMul@train/gradients/network/h2/BiasAdd_grad/tuple/control_dependencyh2/kernel/read* T0* transpose_a(*( _output_shapes :��* transpose_b( � /train/gradients/network/h2/MatMul_grad/MatMul_1MatMulnetwork/h1/Relu@train/gradients/network/h2/BiasAdd_grad/tuple/control_dependency* transpose_a(* _output_shapes : �d* transpose_b(* T0 � 7train/gradients/network/h2/MatMul_grad/tuple/group_depsNoOp.^train/gradients/network/h2/MatMul_grad/MatMul0^train/gradients/network/h2/MatMul_grad/MatMul_1 � ?train/gradients/network/h2/MatMul_grad/tuple/control_dependencyIdentity-train/gradients/network/h2/MatMul_grad/MatMul8^train/gradients/network/h2/MatMul_grad/tuple/group_deps* T0*@ _class6 42loc:@train/gradients/network/h2/MatMul_grad/MatMul*( _output_shapes :�� Atrain/gradients/network/h2/MatMul_grad/tuple/control_dependency_1Identity/train/gradients/network/h2/MatMul_grad/MatMul_18^train/gradients/network/h2/MatMul_grad/tuple/group_deps* T0*B _class8 64loc:@train/gradients/network/h2/MatMul_grad/MatMul_1* _output_shapes : �d � -train/gradients/network/h1/Relu_grad/ReluGradReluGrad?train/gradients/network/h2/MatMul_grad/tuple/control_dependencynetwork/h1/Relu* T0*( _output_shapes :�� 3train/gradients/network/h1/BiasAdd_grad/BiasAddGradBiasAddGrad-train/gradients/network/h1/Relu_grad/ReluGrad* T0* data_formatNHWC* _output_shapes :� � 8train/gradients/network/h1/BiasAdd_grad/tuple/group_depsNoOp4^train/gradients/network/h1/BiasAdd_grad/BiasAddGrad.^train/gradients/network/h1/Relu_grad/ReluGrad � @train/gradients/network/h1/BiasAdd_grad/tuple/control_dependencyIdentity-train/gradients/network/h1/Relu_grad/ReluGrad9^train/gradients/network/h1/BiasAdd_grad/tuple/group_deps* T0*@ _class6 42loc:@train/gradients/network/h1/Relu_grad/ReluGrad*( _output_shapes :�� Btrain/gradients/network/h1/BiasAdd_grad/tuple/control_dependency_1Identity3train/gradients/network/h1/BiasAdd_grad/BiasAddGrad9^train/gradients/network/h1/BiasAdd_grad/tuple/group_deps* _output_shapes :�* T0*F _class< :8loc:@train/gradients/network/h1/BiasAdd_grad/BiasAddGrad � -train/gradients/network/h1/MatMul_grad/MatMulMatMul@train/gradients/network/h1/BiasAdd_grad/tuple/control_dependencyh1/kernel/read* T0* transpose_a(*( _output_shapes :��* transpose_b( � /train/gradients/network/h1/MatMul_grad/MatMul_1MatMul network/X@train/gradients/network/h1/BiasAdd_grad/tuple/control_dependency* T0* transpose_a(* _output_shapes : ��* transpose_b( � 7train/gradients/network/h1/MatMul_grad/tuple/group_depsNoOp.^train/gradients/network/h1/MatMul_grad/MatMul0^train/gradients/network/h1/MatMul_grad/MatMul_1 � ?train/gradients/network/h1/MatMul_grad/tuple/control_dependencyIdentity-train/gradients/network/h1/MatMul_grad/MatMul8^train/gradients/network/h1/MatMul_grad/tuple/group_deps*( _output_shapes :��* T0*@ _class6 42loc:@train/gradients/network/h1/MatMul_grad/MatMul � Atrain/gradients/network/h1/MatMul_grad/tuple/control_dependency_1Identity/train/gradients/network/h1/MatMul_grad/MatMul_18^train/gradients/network/h1/MatMul_grad/tuple/group_deps* T0*B _class8 64loc:@train/gradients/network/h1/MatMul_grad/MatMul_1* _output_shapes : �� h #train/GradientDescent/learning_rateConst* valueB * �#<* dtype0* _output_shapes : � ;train/GradientDescent/update_h1/kernel/ApplyGradientDescentApplyGradientDescent h1/kernel#train/GradientDescent/learning_rateAtrain/gradients/network/h1/MatMul_grad/tuple/control_dependency_1* _output_shapes : ��* use_locking(* T0* _class loc:@h1/kernel � 9train/GradientDescent/update_h1/bias/ApplyGradientDescentApplyGradientDescenth1/bias#train/GradientDescent/learning_rateBtrain/gradients/network/h1/BiasAdd_grad/tuple/control_dependency_1* T0* _class loc:@h1/bias* _output_shapes :�* use_locking( � ;train/GradientDescent/update_h2/kernel/ApplyGradientDescentApplyGradientDescent h2/kernel#train/GradientDescent/learning_rateAtrain/gradients/network/h2/MatMul_grad/tuple/control_dependency_1* _output_shapes : �d* use_locking(* T0* _class loc:@h2/kernel � 9train/GradientDescent/update_h2/bias/ApplyGradientDescentApplyGradientDescenth2/bias#train/GradientDescent/learning_rateBtrain/gradients/network/h2/BiasAdd_grad/tuple/control_dependency_1* use_locking(* T0* _class loc:@h2/bias* _output_shapes :d � ?train/GradientDescent/update_output/kernel/ApplyGradientDescentApplyGradientDescent output/kernel#train/GradientDescent/learning_rateEtrain/gradients/network/output/MatMul_grad/tuple/control_dependency_1* _output_shapes :d * use_locking(* T0* _class loc:@output/kernel � =train/GradientDescent/update_output/bias/ApplyGradientDescentApplyGradientDescentoutput/bias#train/GradientDescent/learning_rateFtrain/gradients/network/output/BiasAdd_grad/tuple/control_dependency_1* T0* _class loc:@output/bias* _output_shapes : * use_locking( � train/GradientDescentNoOp:^train/GradientDescent/update_h1/bias/ApplyGradientDescent<^train/GradientDescent/update_h1/kernel/ApplyGradientDescent:^train/GradientDescent/update_h2/bias/ApplyGradientDescent<^train/GradientDescent/update_h2/kernel/ApplyGradientDescent>^train/GradientDescent/update_output/bias/ApplyGradientDescent@^train/GradientDescent/update_output/kernel/ApplyGradientDescent Z eval/in_top_k/InTopKV2/kConst* value B R* dtype0 * _output_shapes : � eval/in_top_k/InTopKV2InTopKV2network/output/BiasAdd network/yeval/in_top_k/InTopKV2/k* T0 *# _output_shapes : �� v eval/CastCasteval/in_top_k/InTopKV2* SrcT0 * Truncate(* DstT0*# _output_shapes : �� T eval/ConstConst* value B:* dtype0* _output_shapes : f eval/MeanMean eval/Cast eval/Const* _output_shapes :* Tidx0* keep_dims(* T0 � initNoOp^h1/bias/Assign^h1/kernel/Assign^h2/bias/Assign^h2/kernel/Assign^output/bias/Assign^output/kernel/Assign Y save/filename/inputConst* value BBmodel* dtype0* _output_shapes : n save/filenamePlaceholderWithDefaultsave/filename/input* shape:* dtype0* _output_shapes : e save/ConstPlaceholderWithDefault save/filename* shape:* dtype0* _output_shapes : � save/SaveV2/tensor_namesConst*W valueNBLBh1/biasB h1/kernelBh2/biasB h2/kernelBoutput/biasB output/kernel* dtype0* _output_shapes : o save/SaveV2/shape_and_slicesConst* dtype0* _output_shapes :* valueBBBBBBB � save/SaveV2SaveV2 save/Constsave/SaveV2/tensor_namessave/SaveV2/shape_and_slicesh1/bias h1/kernelh2/bias h2/kerneloutput/bias output/kernel* dtypes 2 } save/control_dependencyIdentity save/Const^save/SaveV2* T0* _class loc:@save/Const* _output_shapes : � save/RestoreV2/tensor_namesConst" /device:CPU:0* _output_shapes :*W valueNBLBh1/biasB h1/kernelBh2/biasB h2/kernelBoutput/biasB output/kernel* dtype0 � save/RestoreV2/shape_and_slicesConst" /device:CPU:0* valueBBBBBBB* dtype0* _output_shapes : � save/RestoreV2 RestoreV2 save/Constsave/RestoreV2/tensor_namessave/RestoreV2/shape_and_slices" /device:CPU:0*, _output_shapes ::::::* dtypes 2 � save/AssignAssignh1/biassave/RestoreV2* _class loc:@h1/bias* validate_shape(* _output_shapes :�* use_locking(* T0 � save/Assign_1Assign h1/kernelsave/RestoreV2:1* _class loc:@h1/kernel* validate_shape(* _output_shapes : ��* use_locking(* T0 � save/Assign_2Assignh2/biassave/RestoreV2:2* use_locking(* T0* _class loc:@h2/bias* validate_shape(* _output_shapes :d � save/Assign_3Assign h2/kernelsave/RestoreV2:3* validate_shape(* _output_shapes : �d* use_locking(* T0* _class loc:@h2/kernel � save/Assign_4Assignoutput/biassave/RestoreV2:4* use_locking(* T0* _class loc:@output/bias* validate_shape(* _output_shapes : � save/Assign_5Assign output/kernelsave/RestoreV2:5* use_locking(* T0* _class loc:@output/kernel* validate_shape(* _output_shapes :d v save/restore_allNoOp^save/Assign^save/Assign_1^save/Assign_2^save/Assign_3^save/Assign_4^save/Assign_5"D save/Const:0save/control_dependency:0save/restore_all 5@F8"% train_op train/GradientDescent"� variables�� [ h1/kernel:0h1/kernel/Assignh1/kernel/read:02&h1/kernel/Initializer/random_uniform:08 J h1/bias:0h1/bias/Assignh1/bias/read:02h1/bias/Initializer/zeros:08 [ h2/kernel:0h2/kernel/Assignh2/kernel/read:02&h2/kernel/Initializer/random_uniform:08 J h2/bias:0h2/bias/Assignh2/bias/read:02h2/bias/Initializer/zeros:08 k output/kernel:0output/kernel/Assignoutput/kernel/read:02*output/kernel/Initializer/random_uniform:08 Z output/bias:0output/bias/Assignoutput/bias/read:02output/bias/Initializer/zeros:08"� trainable_variables�� [ h1/kernel:0h1/kernel/Assignh1/kernel/read:02&h1/kernel/Initializer/random_uniform:08 J h1/bias:0h1/bias/Assignh1/bias/read:02h1/bias/Initializer/zeros:08 [ h2/kernel:0h2/kernel/Assignh2/kernel/read:02&h2/kernel/Initializer/random_uniform:08 J h2/bias:0h2/bias/Assignh2/bias/read:02h2/bias/Initializer/zeros:08 k output/kernel:0output/kernel/Assignoutput/kernel/read:02*output/kernel/Initializer/random_uniform:08 Z output/bias:0output/bias/Assignoutput/bias/read:02output/bias/Initializer/zeros:08

cli/jobs/parallel/3a_mnist_batch_identification/mnist_models/mnist-tf.model.meta (1,297 lines of code) (raw):