Remove empty description fields in tf_generated_ops.td and tf_ops.td. (NFC)
PiperOrigin-RevId: 317466704 Change-Id: Ic0bbec0c7013c2f2238a4f4e5763632c846f1337
This commit is contained in:
Andy Ly
TensorFlower Gardener
parent
7a0531c5aa
commit
6d2ce43b03
@ -52,9 +52,6 @@ an output element, this operation computes \\(y = |x|\\).
|
||||
def TF_AcosOp : TF_Op<"Acos", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes acos of x element-wise.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32, F64, I32, I64, TF_Complex128, TF_Complex64]>:$x
|
||||
);
|
||||
@ -371,9 +368,6 @@ retained with length 1.
|
||||
def TF_ApproximateEqualOp : TF_Op<"ApproximateEqual", [Commutative, NoSideEffect]> {
|
||||
let summary = "Returns the truth value of abs(x-y) < tolerance element-wise.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32, F64, I16, I32, I64, I8, TF_Complex128, TF_Complex64, TF_Qint32, TF_Qint8, TF_Quint8, TF_Uint16, TF_Uint32, TF_Uint64, TF_Uint8]>:$x,
|
||||
TensorOf<[BF16, F16, F32, F64, I16, I32, I64, I8, TF_Complex128, TF_Complex64, TF_Qint32, TF_Qint8, TF_Quint8, TF_Uint16, TF_Uint32, TF_Uint64, TF_Uint8]>:$y,
|
||||
@ -734,9 +728,6 @@ window in `value`.
|
||||
def TF_AvgPoolGradOp : TF_Op<"AvgPoolGrad", [NoSideEffect]> {
|
||||
let summary = "Computes gradients of the average pooling function.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I32Tensor:$orig_input_shape,
|
||||
TF_FpTensor:$grad,
|
||||
@ -1402,9 +1393,6 @@ An n-way switch statement, implementing the following:
|
||||
def TF_CastOp : TF_Op<"Cast", [NoSideEffect, SameOperandsAndResultShape]> {
|
||||
let summary = "Cast x of type SrcT to y of DstT.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_Tensor:$x,
|
||||
|
||||
@ -1424,9 +1412,6 @@ def TF_CastOp : TF_Op<"Cast", [NoSideEffect, SameOperandsAndResultShape]> {
|
||||
def TF_CeilOp : TF_Op<"Ceil", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Returns element-wise smallest integer not less than x.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$x
|
||||
);
|
||||
@ -1485,9 +1470,6 @@ greater than `clip_value_max` are set to `clip_value_max`.
|
||||
def TF_CollectiveBcastRecvOp : TF_Op<"CollectiveBcastRecv", []> {
|
||||
let summary = "Receives a tensor value broadcast from another device.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I64Attr:$group_size,
|
||||
I64Attr:$group_key,
|
||||
@ -1507,9 +1489,6 @@ def TF_CollectiveBcastRecvOp : TF_Op<"CollectiveBcastRecv", []> {
|
||||
def TF_CollectiveBcastSendOp : TF_Op<"CollectiveBcastSend", []> {
|
||||
let summary = "Broadcasts a tensor value to one or more other devices.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[F16, F32, F64, I1, I32, I64]>:$input,
|
||||
|
||||
@ -1533,9 +1512,6 @@ def TF_CollectiveGatherOp : TF_Op<"CollectiveGather", []> {
|
||||
Mutually accumulates multiple tensors of identical type and shape.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[F16, F32, F64, I32, I64]>:$input,
|
||||
|
||||
@ -1559,9 +1535,6 @@ def TF_CollectiveReduceOp : TF_Op<"CollectiveReduce", [SameOperandsAndResultType
|
||||
Mutually reduces multiple tensors of identical type and shape.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[F16, F32, F64, I32, I64]>:$input,
|
||||
|
||||
@ -1641,9 +1614,6 @@ value is computed as \\( \sqrt{a^2 + b^2}\\).
|
||||
def TF_ConcatOp : TF_Op<"Concat", [NoSideEffect]> {
|
||||
let summary = "Concatenates tensors along one dimension.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I32Tensor:$concat_dim,
|
||||
Variadic<TF_Tensor>:$values
|
||||
@ -1700,9 +1670,6 @@ This is typically used by gradient computations for a concat operation.
|
||||
def TF_ConcatV2Op : TF_Op<"ConcatV2", [NoSideEffect]> {
|
||||
let summary = "Concatenates tensors along one dimension.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
Variadic<TF_Tensor>:$values,
|
||||
TF_I32OrI64Tensor:$axis
|
||||
@ -1842,9 +1809,6 @@ def TF_Conv2DBackpropFilterOp : TF_Op<"Conv2DBackpropFilter", [NoSideEffect, TF_
|
||||
Computes the gradients of convolution with respect to the filter.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$input,
|
||||
I32Tensor:$filter_sizes,
|
||||
@ -1878,9 +1842,6 @@ def TF_Conv2DBackpropInputOp : TF_Op<"Conv2DBackpropInput", [NoSideEffect, TF_La
|
||||
Computes the gradients of convolution with respect to the input.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I32Tensor:$input_sizes,
|
||||
TensorOf<[BF16, F16, F32, F64, I32]>:$filter,
|
||||
@ -1952,9 +1913,6 @@ def TF_Conv3DBackpropFilterV2Op : TF_Op<"Conv3DBackpropFilterV2", [NoSideEffect]
|
||||
Computes the gradients of 3-D convolution with respect to the filter.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$input,
|
||||
I32Tensor:$filter_sizes,
|
||||
@ -1978,9 +1936,6 @@ def TF_Conv3DBackpropInputV2Op : TF_Op<"Conv3DBackpropInputV2", [NoSideEffect]>
|
||||
Computes the gradients of 3-D convolution with respect to the input.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_I32OrI64Tensor:$input_sizes,
|
||||
TF_FpTensor:$filter,
|
||||
@ -2465,9 +2420,6 @@ horizontal and vertices strides, `strides = [1, stride, stride, 1]`.
|
||||
def TF_DeviceIndexOp : TF_Op<"DeviceIndex", [NoSideEffect]> {
|
||||
let summary = "Return the index of device the op runs.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
StrArrayAttr:$device_names
|
||||
);
|
||||
@ -2792,9 +2744,6 @@ def TF_EluGradOp : TF_Op<"EluGrad", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
Computes gradients for the exponential linear (Elu) operation.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$gradients,
|
||||
TF_FpTensor:$outputs
|
||||
@ -2814,9 +2763,6 @@ Creates a tensor with the given shape.
|
||||
This operation creates a tensor of `shape` and `dtype`.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I32Tensor:$shape,
|
||||
|
||||
@ -2946,9 +2892,6 @@ tf.math.equal(x, y) ==> array([True, True])
|
||||
def TF_ErfOp : TF_Op<"Erf", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes the Gauss error function of `x` element-wise.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$x
|
||||
);
|
||||
@ -2965,9 +2908,6 @@ def TF_ErfcOp : TF_Op<"Erfc", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
Computes the complementary error function of `x` element-wise.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$x
|
||||
);
|
||||
@ -2982,9 +2922,6 @@ Computes the complementary error function of `x` element-wise.
|
||||
def TF_ErfinvOp : TF_Op<"Erfinv", [NoSideEffect]> {
|
||||
let summary = "";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$x
|
||||
);
|
||||
@ -3190,9 +3127,6 @@ def TF_FakeParamOp : TF_Op<"FakeParam", [NoSideEffect]> {
|
||||
intermediate output needed for the gradient computation of the other branch).
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ShapeAttr:$shape
|
||||
);
|
||||
@ -3402,9 +3336,6 @@ fill([2, 3], 9) ==> [[9, 9, 9]
|
||||
def TF_FloorOp : TF_Op<"Floor", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Returns element-wise largest integer not greater than x.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$x
|
||||
);
|
||||
@ -4212,9 +4143,6 @@ def TF_IgammaGradAOp : TF_Op<"IgammaGradA", [NoSideEffect, ResultsBroadcastableS
|
||||
WithBroadcastableBinOpBuilder {
|
||||
let summary = "Computes the gradient of `igamma(a, x)` wrt `a`.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_F32OrF64Tensor:$a,
|
||||
TF_F32OrF64Tensor:$x
|
||||
@ -4487,9 +4415,6 @@ tf.math.is_nan(x) ==> [False, True, False, True, False]
|
||||
def TF_IteratorGetNextOp : TF_Op<"IteratorGetNext", []> {
|
||||
let summary = "Gets the next output from the given iterator .";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$iterator
|
||||
);
|
||||
@ -4558,9 +4483,6 @@ convolutional neural networks (NIPS 2012)](http://papers.nips.cc/paper/4824-imag
|
||||
def TF_LRNGradOp : TF_Op<"LRNGrad", [NoSideEffect]> {
|
||||
let summary = "Gradients for Local Response Normalization.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32]>:$input_grads,
|
||||
TensorOf<[BF16, F16, F32]>:$input_image,
|
||||
@ -4582,9 +4504,6 @@ def TF_LRNGradOp : TF_Op<"LRNGrad", [NoSideEffect]> {
|
||||
def TF_LeakyReluOp : TF_Op<"LeakyRelu", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes rectified linear: `max(features, features * alpha)`.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$features,
|
||||
|
||||
@ -4605,9 +4524,6 @@ def TF_LeakyReluGradOp : TF_Op<"LeakyReluGrad", [NoSideEffect, SameOperandsAndRe
|
||||
Computes rectified linear gradients for a LeakyRelu operation.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$gradients,
|
||||
TF_FpTensor:$features,
|
||||
@ -4888,9 +4804,6 @@ def TF_LogicalAndOp : TF_Op<"LogicalAnd", [Commutative, NoSideEffect, ResultsBro
|
||||
def TF_LogicalNotOp : TF_Op<"LogicalNot", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Returns the truth value of `NOT x` element-wise.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I1Tensor:$x
|
||||
);
|
||||
@ -4971,9 +4884,6 @@ The tensor `values` must be of the type of the table values.
|
||||
def TF_LookupTableSizeV2Op : TF_Op<"LookupTableSizeV2", []> {
|
||||
let summary = "Computes the number of elements in the given table.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$table_handle
|
||||
);
|
||||
@ -5658,9 +5568,6 @@ retained with length 1.
|
||||
def TF_MaxPoolOp : TF_Op<"MaxPool", [NoSideEffect, TF_FoldOperandsTransposeInterface]> {
|
||||
let summary = "Performs max pooling on the input.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32, F64, I16, I32, I64, I8, TF_Qint8, TF_Uint16, TF_Uint8]>:$input,
|
||||
|
||||
@ -5687,9 +5594,6 @@ def TF_MaxPoolOp : TF_Op<"MaxPool", [NoSideEffect, TF_FoldOperandsTransposeInter
|
||||
def TF_MaxPool3DOp : TF_Op<"MaxPool3D", [NoSideEffect]> {
|
||||
let summary = "Performs 3D max pooling on the input.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32]>:$input,
|
||||
|
||||
@ -5709,9 +5613,6 @@ def TF_MaxPool3DOp : TF_Op<"MaxPool3D", [NoSideEffect]> {
|
||||
def TF_MaxPool3DGradOp : TF_Op<"MaxPool3DGrad", [NoSideEffect]> {
|
||||
let summary = "Computes gradients of 3D max pooling function.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32]>:$orig_input,
|
||||
TensorOf<[BF16, F16, F32]>:$orig_output,
|
||||
@ -5734,9 +5635,6 @@ def TF_MaxPool3DGradOp : TF_Op<"MaxPool3DGrad", [NoSideEffect]> {
|
||||
def TF_MaxPoolGradOp : TF_Op<"MaxPoolGrad", [NoSideEffect]> {
|
||||
let summary = "Computes gradients of the maxpooling function.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_IntOrFpTensor:$orig_input,
|
||||
TF_IntOrFpTensor:$orig_output,
|
||||
@ -6015,9 +5913,6 @@ Returns x * y element-wise. Returns zero if y is zero, even if x if infinite or
|
||||
def TF_MultinomialOp : TF_Op<"Multinomial", []> {
|
||||
let summary = "Draws samples from a multinomial distribution.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_IntOrFpTensor:$logits,
|
||||
I32Tensor:$num_samples,
|
||||
@ -6037,9 +5932,6 @@ def TF_MultinomialOp : TF_Op<"Multinomial", []> {
|
||||
def TF_NdtriOp : TF_Op<"Ndtri", [NoSideEffect]> {
|
||||
let summary = "";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$x
|
||||
);
|
||||
@ -6074,9 +5966,6 @@ I.e., \\(y = -x\\).
|
||||
def TF_NoOp : TF_Op<"NoOp", [NoSideEffect]> {
|
||||
let summary = "Does nothing. Only useful as a placeholder for control edges.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins);
|
||||
|
||||
let results = (outs);
|
||||
@ -6330,9 +6219,6 @@ output =
|
||||
def TF_OutfeedEnqueueTupleOp : TF_Op<"OutfeedEnqueueTuple", []> {
|
||||
let summary = "Enqueue multiple Tensor values on the computation outfeed.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
Variadic<TF_Tensor>:$inputs
|
||||
);
|
||||
@ -6617,9 +6503,6 @@ q_full, r_full = qr(a, full_matrices=True)
|
||||
def TF_QuantizeAndDequantizeOp : TF_Op<"QuantizeAndDequantize", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Use QuantizeAndDequantizeV2 instead.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$input,
|
||||
|
||||
@ -6871,9 +6754,6 @@ def TF_RandomGammaGradOp : TF_Op<"RandomGammaGrad", [NoSideEffect, ResultsBroadc
|
||||
Computes the derivative of a Gamma random sample w.r.t. `alpha`.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_F32OrF64Tensor:$alpha,
|
||||
TF_F32OrF64Tensor:$sample
|
||||
@ -7203,9 +7083,6 @@ array([ 0., 0., -0., 3.], dtype=float32)
|
||||
def TF_Relu6Op : TF_Op<"Relu6", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes rectified linear 6: `min(max(features, 0), 6)`.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_IntOrFpTensor:$features
|
||||
);
|
||||
@ -7220,9 +7097,6 @@ def TF_Relu6Op : TF_Op<"Relu6", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
def TF_Relu6GradOp : TF_Op<"Relu6Grad", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes rectified linear 6 gradients for a Relu6 operation.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_IntOrFpTensor:$gradients,
|
||||
TF_IntOrFpTensor:$features
|
||||
@ -7238,9 +7112,6 @@ def TF_Relu6GradOp : TF_Op<"Relu6Grad", [NoSideEffect, SameOperandsAndResultType
|
||||
def TF_ReluGradOp : TF_Op<"ReluGrad", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes rectified linear gradients for a Relu operation.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_IntOrFpTensor:$gradients,
|
||||
TF_IntOrFpTensor:$features
|
||||
@ -7365,9 +7236,6 @@ Input images can be of different types but output images are always float.
|
||||
def TF_ResizeBilinearGradOp : TF_Op<"ResizeBilinearGrad", [NoSideEffect]> {
|
||||
let summary = "Computes the gradient of bilinear interpolation.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
F32Tensor:$grads,
|
||||
TF_FpTensor:$original_image,
|
||||
@ -7388,9 +7256,6 @@ def TF_ResizeNearestNeighborOp : TF_Op<"ResizeNearestNeighbor", [NoSideEffect]>
|
||||
Resize `images` to `size` using nearest neighbor interpolation.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32, F64, I16, I32, I64, I8, TF_Uint16, TF_Uint8]>:$images,
|
||||
I32Tensor:$size,
|
||||
@ -7507,9 +7372,6 @@ var <- var - mom
|
||||
def TF_ResourceApplyGradientDescentOp : TF_Op<"ResourceApplyGradientDescent", []> {
|
||||
let summary = "Update '*var' by subtracting 'alpha' * 'delta' from it.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$var,
|
||||
TensorOf<[BF16, F16, F32, F64, I16, I32, I64, I8, TF_Complex128, TF_Complex64, TF_Qint32, TF_Qint8, TF_Quint8, TF_Uint16, TF_Uint32, TF_Uint64, TF_Uint8]>:$alpha,
|
||||
@ -8292,9 +8154,6 @@ select(condition, t, e) ==> [[1, 2],
|
||||
def TF_SelectV2Op : TF_Op<"SelectV2", [NoSideEffect, ResultsBroadcastableShape]> {
|
||||
let summary = "";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I1Tensor:$condition,
|
||||
TF_Tensor:$t,
|
||||
@ -8343,9 +8202,6 @@ def TF_SeluGradOp : TF_Op<"SeluGrad", [NoSideEffect, SameOperandsAndResultType]>
|
||||
Computes gradients for the scaled exponential linear (Selu) operation.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$gradients,
|
||||
TF_FpTensor:$outputs
|
||||
@ -8596,9 +8452,6 @@ whose values are extracted from 'input' starting at the offsets in
|
||||
def TF_SnapshotOp : TF_Op<"Snapshot", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Returns a copy of the input tensor.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_Tensor:$input
|
||||
);
|
||||
@ -8663,9 +8516,6 @@ Inputs are the logits, not probabilities.
|
||||
def TF_SoftplusOp : TF_Op<"Softplus", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes softplus: `log(exp(features) + 1)`.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$features
|
||||
);
|
||||
@ -8680,9 +8530,6 @@ def TF_SoftplusOp : TF_Op<"Softplus", [NoSideEffect, SameOperandsAndResultType]>
|
||||
def TF_SoftplusGradOp : TF_Op<"SoftplusGrad", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes softplus gradients for a softplus operation.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$gradients,
|
||||
TF_FpTensor:$features
|
||||
@ -8698,9 +8545,6 @@ def TF_SoftplusGradOp : TF_Op<"SoftplusGrad", [NoSideEffect, SameOperandsAndResu
|
||||
def TF_SoftsignOp : TF_Op<"Softsign", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes softsign: `features / (abs(features) + 1)`.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$features
|
||||
);
|
||||
@ -8715,9 +8559,6 @@ def TF_SoftsignOp : TF_Op<"Softsign", [NoSideEffect, SameOperandsAndResultType]>
|
||||
def TF_SoftsignGradOp : TF_Op<"SoftsignGrad", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Computes softsign gradients for a softsign operation.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_FpTensor:$gradients,
|
||||
TF_FpTensor:$features
|
||||
@ -8965,9 +8806,6 @@ are checked during execution.
|
||||
def TF_SplitOp : TF_Op<"Split", [NoSideEffect]> {
|
||||
let summary = "Splits a tensor into `num_split` tensors along one dimension.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I32Tensor:$split_dim,
|
||||
TF_Tensor:$value
|
||||
@ -8986,9 +8824,6 @@ def TF_SplitOp : TF_Op<"Split", [NoSideEffect]> {
|
||||
def TF_SplitVOp : TF_Op<"SplitV", [NoSideEffect]> {
|
||||
let summary = "Splits a tensor into `num_split` tensors along one dimension.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_Tensor:$value,
|
||||
TF_I32OrI64Tensor:$size_splits,
|
||||
@ -9052,11 +8887,11 @@ I.e., \\(y = x * x = x^2\\).
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[BF16, F16, F32, F64, I32, I64, TF_Complex128, TF_Complex64]>:$x
|
||||
TensorOf<[BF16, F16, F32, F64, I16, I32, I64, I8, TF_Complex128, TF_Complex64]>:$x
|
||||
);
|
||||
|
||||
let results = (outs
|
||||
TensorOf<[BF16, F16, F32, F64, I32, I64, TF_Complex128, TF_Complex64]>:$y
|
||||
TensorOf<[BF16, F16, F32, F64, I16, I32, I64, I8, TF_Complex128, TF_Complex64]>:$y
|
||||
);
|
||||
|
||||
TF_DerivedOperandTypeAttr T = TF_DerivedOperandTypeAttr<0>;
|
||||
@ -9125,9 +8960,6 @@ shape(squeeze(t, [2, 4])) ==> [1, 2, 3, 1]
|
||||
def TF_StackCloseV2Op : TF_Op<"StackCloseV2", []> {
|
||||
let summary = "Delete the stack from its resource container.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$handle
|
||||
);
|
||||
@ -9138,9 +8970,6 @@ def TF_StackCloseV2Op : TF_Op<"StackCloseV2", []> {
|
||||
def TF_StackPopV2Op : TF_Op<"StackPopV2", []> {
|
||||
let summary = "Pop the element at the top of the stack.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$handle
|
||||
);
|
||||
@ -9155,9 +8984,6 @@ def TF_StackPopV2Op : TF_Op<"StackPopV2", []> {
|
||||
def TF_StackPushV2Op : TF_Op<"StackPushV2", []> {
|
||||
let summary = "Push an element onto the stack.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$handle,
|
||||
TF_Tensor:$elem,
|
||||
@ -9175,9 +9001,6 @@ def TF_StackPushV2Op : TF_Op<"StackPushV2", []> {
|
||||
def TF_StackV2Op : TF_Op<"StackV2", []> {
|
||||
let summary = "A stack that produces elements in first-in last-out order.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
I32Tensor:$max_size,
|
||||
|
||||
@ -9895,9 +9718,6 @@ calculation gets its own TensorArray accumulator.
|
||||
def TF_TensorArrayReadV3Op : TF_Op<"TensorArrayReadV3", []> {
|
||||
let summary = "Read an element from the TensorArray into output `value`.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$handle,
|
||||
I32Tensor:$index,
|
||||
@ -9937,9 +9757,6 @@ Scatter the data from the input value into specific TensorArray elements.
|
||||
def TF_TensorArraySizeV3Op : TF_Op<"TensorArraySizeV3", []> {
|
||||
let summary = "Get the current size of the TensorArray.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$handle,
|
||||
F32Tensor:$flow_in
|
||||
@ -10016,9 +9833,6 @@ Write data via Write and read via Read or Pack.
|
||||
def TF_TensorArrayWriteV3Op : TF_Op<"TensorArrayWriteV3", []> {
|
||||
let summary = "Push an element onto the tensor_array.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ResourceTensor:$handle,
|
||||
I32Tensor:$index,
|
||||
@ -10139,9 +9953,6 @@ values: The tensor.
|
||||
def TF_TensorListGetItemOp : TF_Op<"TensorListGetItem", [NoSideEffect]> {
|
||||
let summary = "";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_VariantTensor:$input_handle,
|
||||
I32Tensor:$index,
|
||||
@ -10271,9 +10082,6 @@ output_handle: The TensorList.
|
||||
def TF_TensorListSetItemOp : TF_Op<"TensorListSetItem", [NoSideEffect]> {
|
||||
let summary = "";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_VariantTensor:$input_handle,
|
||||
I32Tensor:$index,
|
||||
@ -11063,9 +10871,6 @@ def TF_XdivyOp : TF_Op<"Xdivy", [NoSideEffect, ResultsBroadcastableShape]>,
|
||||
WithBroadcastableBinOpBuilder {
|
||||
let summary = "Returns 0 if x == 0, and x / y otherwise, elementwise.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[F16, F32, F64, TF_Complex128, TF_Complex64]>:$x,
|
||||
TensorOf<[F16, F32, F64, TF_Complex128, TF_Complex64]>:$y
|
||||
@ -11242,9 +11047,6 @@ def TF_XlaHostComputeOp : TF_Op<"XlaHostCompute", []> {
|
||||
A pseudo-op to represent host-side computation in an XLA program.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
Variadic<TF_Tensor>:$inputs,
|
||||
|
||||
@ -11315,9 +11117,6 @@ https://www.tensorflow.org/performance/xla/operation_semantics#pad
|
||||
def TF_XlaRecvFromHostOp : TF_Op<"XlaRecvFromHost", []> {
|
||||
let summary = "An op to receive a tensor from the host.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_ShapeAttr:$shape,
|
||||
StrAttr:$key
|
||||
@ -11355,9 +11154,6 @@ https://www.tensorflow.org/performance/xla/operation_semantics#reduce .
|
||||
def TF_XlaReplicaIdOp : TF_Op<"XlaReplicaId", [NoSideEffect]> {
|
||||
let summary = "Replica ID.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins);
|
||||
|
||||
let results = (outs
|
||||
@ -11397,9 +11193,6 @@ i=0...N-1.
|
||||
def TF_XlaSendToHostOp : TF_Op<"XlaSendToHost", []> {
|
||||
let summary = "An op to send a tensor to the host.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_Tensor:$input,
|
||||
|
||||
@ -11443,9 +11236,6 @@ tensor such that tensor[...,:,:] = u[..., :, :] * Diag(s[..., :]) * Transpose(v[
|
||||
def TF_Xlog1pyOp : TF_Op<"Xlog1py", [NoSideEffect]> {
|
||||
let summary = "Returns 0 if x == 0, and x * log1p(y) otherwise, elementwise.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[F16, F32, F64, TF_Complex128, TF_Complex64]>:$x,
|
||||
TensorOf<[F16, F32, F64, TF_Complex128, TF_Complex64]>:$y
|
||||
@ -11462,9 +11252,6 @@ def TF_XlogyOp : TF_Op<"Xlogy", [NoSideEffect, ResultsBroadcastableShape]>,
|
||||
WithBroadcastableBinOpBuilder {
|
||||
let summary = "Returns 0 if x == 0, and x * log(y) otherwise, elementwise.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TensorOf<[F16, F32, F64, TF_Complex128, TF_Complex64]>:$x,
|
||||
TensorOf<[F16, F32, F64, TF_Complex128, TF_Complex64]>:$y
|
||||
@ -11480,9 +11267,6 @@ def TF_XlogyOp : TF_Op<"Xlogy", [NoSideEffect, ResultsBroadcastableShape]>,
|
||||
def TF_ZerosLikeOp : TF_Op<"ZerosLike", [NoSideEffect, SameOperandsAndResultType]> {
|
||||
let summary = "Returns a tensor of zeros with the same shape and type as x.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_Tensor:$x
|
||||
);
|
||||
@ -11589,9 +11373,6 @@ expected to create these operators.
|
||||
def TF__HostComputeMlirOp : TF_Op<"_HostComputeMlir", []> {
|
||||
let summary = "A host-side computation called from a TPU device.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
Variadic<TF_Tensor>:$inputs,
|
||||
|
||||
@ -11671,9 +11452,6 @@ def TF__XlaRecvAtHostOp : TF_Op<"_XlaRecvAtHost", []> {
|
||||
A placeholder op to receive values from a running XLA computation.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_StrTensor:$dynamic_key,
|
||||
|
||||
@ -11691,9 +11469,6 @@ A placeholder op to receive values from a running XLA computation.
|
||||
def TF__XlaSendFromHostOp : TF_Op<"_XlaSendFromHost", []> {
|
||||
let summary = "A placeholder op to send values to a running XLA computation.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
Variadic<TF_Tensor>:$inputs,
|
||||
TF_StrTensor:$dynamic_key,
|
||||
|
||||
@ -232,6 +232,7 @@ else_branch: A function that takes 'inputs' and returns a list of
|
||||
|
||||
def TF_YieldOp : TF_Op<"Yield", [Terminator]> {
|
||||
let summary = "Yield operation";
|
||||
|
||||
let description = [{
|
||||
The "yield" operation represents a return operation within the conditional
|
||||
and body of structured control flow (e.g., if and while). The operation
|
||||
@ -497,6 +498,7 @@ Inserts a placeholder for a tensor that will be always fed.
|
||||
|
||||
def TF_PlaceholderWithDefaultOp : TF_Op<"PlaceholderWithDefault", [NoSideEffect]> {
|
||||
let summary = "Placeholder op";
|
||||
|
||||
let description = [{
|
||||
A placeholder op that passes through input when its output is not fed.
|
||||
}];
|
||||
@ -839,9 +841,6 @@ def TF_XlaShardingOp : TF_Op<"XlaSharding", [NoSideEffect]> {
|
||||
An op which shards the input based on the given sharding attribute.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_Tensor:$input,
|
||||
|
||||
@ -858,9 +857,6 @@ An op which shards the input based on the given sharding attribute.
|
||||
def TF_InfeedDequeueTupleOp : TF_Op<"InfeedDequeueTuple", []> {
|
||||
let summary = "Fetches multiple values from infeed as an XLA tuple.";
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
OptionalAttr<StrAttr>:$_XlaSharding
|
||||
);
|
||||
@ -904,9 +900,6 @@ def TF_BatchDatasetV2Op : TF_Op<"BatchDatasetV2", [NoSideEffect]> {
|
||||
Creates a dataset that batches `batch_size` elements from `input_dataset`.
|
||||
}];
|
||||
|
||||
let description = [{
|
||||
}];
|
||||
|
||||
let arguments = (ins
|
||||
TF_VariantTensor:$input_dataset,
|
||||
I64Tensor:$batch_size,
|
||||
|
||||
Loading…
Reference in New Issue
Block a user