From 61d4654f60dbc70ca9bcfb45826f3b68f2de8f75 Mon Sep 17 00:00:00 2001 From: Travis CI Date: Mon, 18 Dec 2017 06:33:17 +0000 Subject: [PATCH] Deploy to GitHub Pages: 7456d737b295e687579a6d6f42a645d5dceee24d --- develop/doc/operators.json | 997 +++++++++++++++++++------------------ 1 file changed, 513 insertions(+), 484 deletions(-) diff --git a/develop/doc/operators.json b/develop/doc/operators.json index 1e8c456f81d..f41769d142c 100644 --- a/develop/doc/operators.json +++ b/develop/doc/operators.json @@ -387,35 +387,6 @@ "comment" : "(int, default:1) the contextStride of SequenceConvOp represents the stride length of convolution kernel. Currently, SequenceConvOp only supportscontextStride=1.", "generated" : 0 } ] -},{ - "type" : "sequence_pool", - "comment" : "\nSequence Pool Operator.\n\nThe SequencePoolOp pools features of all time-steps of each instance.\nIt supports six pooling types:\n1. AVERAGE: $$Out[i] = \\frac{\\sum_i X_i}{N}$$\n2. SUM: $$Out[i] = \\sum_jX_{ij}$$\n3. SQRT: $$Out[i] = \\frac{\\sum_jX_{ij}}{\\sqrt{len(X_i)}}$$\n4. LAST: Out[i] = last instance in i-th sequence X[i]\n5. FIRST: Out[i] = first instance in i-th sequence X[i]\n6. MAX: $$Out[i] = max(X_i)$$\n\nThe following example explains how this works:\nFor a mini-batch of 3 variable-length sentences,\ncontaining 2, 3, and 2 time-steps:\n\nAssume X is a [7,M,N] LoDTensor, and X->lod()[0] = [0, 2, 5, 7], 7=2+3+2.\nBesides, for the sake of simplicity, we assume M=1 and N=1,\nand the value of X = [[1, 3], [2, 4, 6], [5, 1]].\n\nThus, Out is a [3,1,1] Tensor without LoD infomation.\nAnd for different pooltype, the value of Out is as follows:\n\n- AVERAGE: [2, 4, 3], where 2=(1+3)/2, 4=(2+4+6)/3, 3=(5+1)/2\n- SUM: [4, 12, 6], where 4=1+3, 12=2+4+6, 6=5+1\n- SQRT: [2.82, 6.93, 4.24], where 2.82=(1+3)/sqrt(2),\n 6.93=(2+4+6)/sqrt(3), 4.24=(5+1)/sqrt(2)\n- MAX: [3, 6, 5], where 3=max(1,3), 6=max(2,4,6), 5=max(5,1)\n- LAST: [3, 6, 1], where 3=last(1,3), 6=last(2,4,6), 1=last(5,1)\n- FIRST: [1, 2, 5], where 1=first(1,3), 2=first(2,4,6), 5=first(5,1)\n\n ", - "inputs" : [ - { - "name" : "X", - "comment" : "(LoDTensor) The variable-length input of SequencePoolOp", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Out", - "comment" : "(Tensor) The output of SequencePoolOp does not contain LoD infomation.", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "MaxIndex", - "comment" : "(Tensor) This tensor is used for the sequence max-pooling to record the max indexes.", - "duplicable" : 0, - "intermediate" : 1 - } ], - "attrs" : [ - { - "name" : "pooltype", - "type" : "string", - "comment" : "(int, default AVERAGE) the pooling pooltype of SequencePoolOp.", - "generated" : 0 - } ] },{ "type" : "lstm", "comment" : "\nLong-Short Term Memory (LSTM) Operator.\n\nThe defalut implementation is diagonal/peephole connection\n(https://arxiv.org/pdf/1402.1128.pdf), the formula is as follows:\n\n$$\ni_t = \\sigma(W_{ix}x_{t} + W_{ih}h_{t-1} + W_{ic}c_{t-1} + b_i) \\\\\n\nf_t = \\sigma(W_{fx}x_{t} + W_{fh}h_{t-1} + W_{fc}c_{t-1} + b_f) \\\\\n\n\\tilde{c_t} = act_g(W_{cx}x_t + W_{ch}h_{t-1} + b_c) \\\\\n\no_t = \\sigma(W_{ox}x_{t} + W_{oh}h_{t-1} + W_{oc}c_t + b_o) \\\\\n\nc_t = f_t \\odot c_{t-1} + i_t \\odot \\tilde{c_t} \\\\\n\nh_t = o_t \\odot act_h(c_t)\n$$\n\nwhere the W terms denote weight matrices (e.g. $W_{xi}$ is the matrix\nof weights from the input gate to the input), $W_{ic}, W_{fc}, W_{oc}$\nare diagonal weight matrices for peephole connections. In our implementation,\nwe use vectors to reprenset these diagonal weight matrices. The b terms\ndenote bias vectors ($b_i$ is the input gate bias vector), $\\sigma$\nis the non-line activations, such as logistic sigmoid function, and\n$i, f, o$ and $c$ are the input gate, forget gate, output gate,\nand cell activation vectors, respectively, all of which have the same size as\nthe cell output activation vector $h$.\n\nThe $\\odot$ is the element-wise product of the vectors. $act_g$ and $act_h$\nare the cell input and cell output activation functions and `tanh` is usually\nused for them. $\\tilde{c_t}$ is also called candidate hidden state,\nwhich is computed based on the current input and the previous hidden state.\n\nSet `use_peepholes` False to disable peephole connection. The formula\nis omitted here, please refer to the paper\nhttp://www.bioinf.jku.at/publications/older/2604.pdf for details.\n\nNote that these $W_{xi}x_{t}, W_{xf}x_{t}, W_{xc}x_{t}, W_{xo}x_{t}$\noperations on the input $x_{t}$ are NOT included in this operator.\nUsers can choose to use fully-connect operator before LSTM operator.\n\n", @@ -1009,6 +980,47 @@ "intermediate" : 0 } ], "attrs" : [ ] +},{ + "type" : "read_from_array", + "comment" : "\nReadFromArray Operator.\n\nRead a LoDTensor from a LoDTensor Array.\n\nAssume $T$ is LoDTensor, $i$ is the subscript of the array, and $A$ is the array. The\nequation is\n\n$$T = A[i]$$\n\n", + "inputs" : [ + { + "name" : "X", + "comment" : "(TensorArray) the array will be read from.", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "I", + "comment" : "(Tensor) the subscript index in tensor array. The number of element should be 1", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "Out", + "comment" : "(LoDTensor) the tensor will be read from.", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ ] +},{ + "type" : "softplus", + "comment" : "\nSoftplus Activation Operator.\n\n$y = \\ln(1 + e^{x})$\n\n", + "inputs" : [ + { + "name" : "X", + "comment" : "Input of Softplus operator", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "Y", + "comment" : "Output of Softplus operator", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ ] },{ "type" : "logical_xor", "comment" : "logical_xor Operator\n\nIt operates element-wise on X and Y, and returns the Out. X, Y and Out are N-dim boolean tensors.\nEach element of Out is calculated by $$Out = (X || Y) \\, \\&\\& \\, !(X \\&\\& Y)$$\n", @@ -1331,24 +1343,6 @@ "comment" : "The exponential factor of Pow", "generated" : 0 } ] -},{ - "type" : "l1_norm", - "comment" : "\nL1 Norm Operator.\n\nComputes the L1 norm of a tensor.\n\n$$Out = \\sum{|X|}$$\n\n", - "inputs" : [ - { - "name" : "X", - "comment" : "(Tensor) The input of l1_norm op.", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Out", - "comment" : "(Scalar) The output of l1_norm op.", - "duplicable" : 0, - "intermediate" : 0 - } ], - "attrs" : [ ] },{ "type" : "sqrt", "comment" : "\nSqrt Activation Operator.\n\n$y = \\sqrt{x}$\n\n", @@ -1590,42 +1584,37 @@ } ], "attrs" : [ ] },{ - "type" : "merge_lod_tensor", - "comment" : "\n Merge True and False branches of LoDTensor into a single Output,\n with a mask at certain lod level. X is used to obtain complete\n lod information. Please refer to SplitLoDTensorOp.", + "type" : "reduce_min", + "comment" : "\n{ReduceOp} Operator.\n\nThis operator computes the min of input tensor along the given dimension. \nThe result tensor has 1 fewer dimension than the input unless keep_dim is true.\nIf reduce_all is true, just reduce along all dimensions and output a scalar.\n\n", "inputs" : [ { "name" : "X", - "comment" : "The input LoDTensor, contains complete lod information to construct the output", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Mask", - "comment" : "A bool column vector which mask the input", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "InTrue", - "comment" : "The True branch to be merged", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "InFalse", - "comment" : "The False branch to be merged", + "comment" : "(Tensor) The input tensor. Tensors with rank at most 6 are supported.", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "The merged output LoDTensor", + "comment" : "(Tensor) The result tensor.", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ { - "name" : "level", + "name" : "dim", "type" : "int", - "comment" : "(int) the specific lod level to rank.", + "comment" : "(int, default 0) The dimension to reduce. Must be in the range [-rank(input), rank(input)). If `dim < 0`, the dim to reduce is `rank + dim`. Note that reducing on the first dim will make the LoD info lost.", + "generated" : 0 + }, { + "name" : "keep_dim", + "type" : "bool", + "comment" : "(bool, default false) If true, retain the reduced dimension with length 1.", + "generated" : 0 + }, { + "name" : "reduce_all", + "type" : "bool", + "comment" : "(bool, default false) If true, output a scalar reduced along all dimensions.", "generated" : 0 } ] },{ @@ -1811,6 +1800,50 @@ "comment" : "(int, default 5 (FP32)) Output data type", "generated" : 0 } ] +},{ + "type" : "smooth_l1_loss", + "comment" : "\nSmooth L1 Loss Operator.\n\nThis operator computes the smooth l1 loss for X and Y.\nThe operator takes the first dimension of X and Y as batch size.\nFor each instance, it computes the smooth l1 loss element by element first\nand then sums all the losses. So the shape of Out is [batch_size, 1].\n\nThe equation is:\n$$\nOut_{\\sigma}(X, Y)_i = \\begin{cases}\n0.5 * (\\sigma * (X_i - Y_i)) ^ 2\n\\quad |X_i - Y_i| \\lt \\frac{1} {{\\sigma} ^ 2} \\\\\n\\frac{|X_i - Y_i| - 0.5}{{\\sigma}^2},\n\\quad otherwise\n\\end{cases}\n$$\n\nIn the above equation, $Out_{\\sigma}(X, Y)_i$, $X_i$ and $Y_i$ represent the ith\nelement of Out, X and Y.\n\n", + "inputs" : [ + { + "name" : "X", + "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. The input value of smooth l1 loss op with shape [batch_size, dim1, ..., dimN].", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Y", + "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. The target value of smooth l1 loss op with same shape as X.", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "InsideWeight", + "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. This input is optional and should have same shape with X. If provided, the result of (X - Y) will be multiplied by this tensor element by element.", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "OutsideWeight", + "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. This input is optional and should have same shape with X. If provided, the out smooth l1 loss will be multiplied by this tensor element by element.", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "Diff", + "comment" : "Intermediate variable to cache InsideWeight * (X - Y).", + "duplicable" : 0, + "intermediate" : 1 + }, { + "name" : "Out", + "comment" : "(Tensor, default Tensor) A tensor with rank be 2. The output smooth l1 loss with shape [batch_size, 1].", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ + { + "name" : "sigma", + "type" : "float", + "comment" : "Hyper parameter of smooth l1 loss op.A float scalar with default value 3.0.", + "generated" : 0 + } ] },{ "type" : "pad", "comment" : "\nPad Operator.\n\nPad input into output, as specified by paddings and pad_value. \nThe input should be a k-D tensor(k > 0 and k < 7). As an example:\n\nGiven:\n\nX = [[1, 2],\n [3, 4]],\n\npaddings = [0, 1, 1, 2],\n\nand\n\npad_value = 0,\n\nwe have:\n\nOut = [[0, 1, 2, 0, 0]\n [0, 3, 4, 0, 0]\n [0, 0, 0, 0, 0]]\n\n", @@ -1993,168 +2026,70 @@ "generated" : 0 } ] },{ - "type" : "conv2d_transpose_cudnn", - "comment" : "\nConvolution2D Transpose Operator.\n\nThe convolution transpose operation calculates the output based on the input, filter\nand strides, paddings, groups parameters. The size of each dimension of the\nparameters is checked in the infer-shape.\nInput(Input) and output(Output) are in NCHW format. Where N is batchsize, C is the\nnumber of channels, H is the height of the feature, and W is the width of the feature.\nFilter(Input) is in MCHW format. Where M is the number of input feature channels,\nC is the number of output feature channels, H is the height of the filter,\nand W is the width of the filter.\nParameters(strides, paddings) are two elements. These two elements represent height\nand width, respectively.\nThe input(X) size and output(Out) size may be different.\n\nExample:\n Input:\n Input shape: $(N, C_{in}, H_{in}, W_{in})$\n Filter shape: $(C_{in}, C_{out}, H_f, W_f)$\n Output:\n Output shape: $(N, C_{out}, H_{out}, W_{out})$\n Where\n $$\n H_{out} = (H_{in} - 1) * strides[0] - 2 * paddings[0] + H_f \\\\\n W_{out} = (W_{in} - 1) * strides[1] - 2 * paddings[1] + W_f\n $$\n", + "type" : "increment", + "comment" : "\nIncrement Operator.\n\nThe equation is: \n$$Out = X + step$$\n\n", "inputs" : [ { - "name" : "Input", - "comment" : "(Tensor) The input tensor of convolution transpose operator. The format of input tensor is NCHW. Where N is batch size, C is the number of input channels, H is the height of the feature, and W is the width of the feature.", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Filter", - "comment" : "(Tensor) The filter tensor of convolution transpose operator. The format of the filter tensor is MCHW, where M is the number of input feature channels, C is the number of output feature channels,H is the height of the filter, and W is the width of the filter. We enforce groups number == 1 in the convolution transpose scenario.", + "name" : "X", + "comment" : "(Tensor) The input tensor of increment operator", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { - "name" : "Output", - "comment" : "(Tensor) The output tensor of convolution transpose operator. The format of output tensor is also NCHW.", + "name" : "Out", + "comment" : "(Tensor) The output tensor of increment operator.", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ { - "name" : "strides", - "type" : "int array", - "comment" : "(vector default:{1, 1}), the strides(h_stride, w_stride) of convolution transpose operator.", - "generated" : 0 - }, { - "name" : "paddings", - "type" : "int array", - "comment" : "(vector default:{0, 0}), the paddings(h_pad, w_pad) of convolution transpose operator.", - "generated" : 0 - }, { - "name" : "dilations", - "type" : "int array", - "comment" : "dilations of convolution operator.", - "generated" : 0 - }, { - "name" : "workspace_size_MB", - "type" : "int", - "comment" : "workspace size for cudnn, in MB, workspace is a section of GPU memory which will be allocated/freed each time the operator runs, larger workspace size can increase performance but also requires better hardward. This size should be carefully setted.", + "name" : "step", + "type" : "float", + "comment" : "(float, default 1.0) The step size by which the input tensor will be incremented.", "generated" : 0 } ] },{ - "type" : "reduce_min", - "comment" : "\n{ReduceOp} Operator.\n\nThis operator computes the min of input tensor along the given dimension. \nThe result tensor has 1 fewer dimension than the input unless keep_dim is true.\nIf reduce_all is true, just reduce along all dimensions and output a scalar.\n\n", + "type" : "log_loss", + "comment" : "\nLogLoss Operator.\n\nLog loss is a loss function used for binary classification. Log Loss quantifies\nthe accuracy of a classifier by penalising false classifications. Minimising the\nLog Loss is equivalent to maximising the accuracy of the classifier. We define\nPredicted as the values predicted by our model and Labels as the target ground\ntruth value. Log loss can evaluate how close the predicted values are to the\ntarget. The shapes of Predicted and Labels are both [batch_size, 1].\nThe equation is:\n\n$$\nLoss = - Labels * log(Predicted + \\epsilon) -\n (1 - Labels) * log(1 - Predicted + \\epsilon)\n$$\n\n", "inputs" : [ { - "name" : "X", - "comment" : "(Tensor) The input tensor. Tensors with rank at most 6 are supported.", + "name" : "Predicted", + "comment" : "The input value (Predicted) of Log loss op.Predicted is a 2-D tensor with shape [batch_size, 1].", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Labels", + "comment" : "The target value (Labels) of Log loss op.Labels is a 2-D tensor with shape [batch_size, 1].", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { - "name" : "Out", - "comment" : "(Tensor) The result tensor.", + "name" : "Loss", + "comment" : "The output tensor with shape [batch_size, 1] which represents the log loss.", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ { - "name" : "dim", - "type" : "int", - "comment" : "(int, default 0) The dimension to reduce. Must be in the range [-rank(input), rank(input)). If `dim < 0`, the dim to reduce is `rank + dim`. Note that reducing on the first dim will make the LoD info lost.", - "generated" : 0 - }, { - "name" : "keep_dim", - "type" : "bool", - "comment" : "(bool, default false) If true, retain the reduced dimension with length 1.", - "generated" : 0 - }, { - "name" : "reduce_all", - "type" : "bool", - "comment" : "(bool, default false) If true, output a scalar reduced along all dimensions.", + "name" : "epsilon", + "type" : "float", + "comment" : "Epsilon in log loss.", "generated" : 0 } ] },{ - "type" : "smooth_l1_loss", - "comment" : "\nSmooth L1 Loss Operator.\n\nThis operator computes the smooth l1 loss for X and Y.\nThe operator takes the first dimension of X and Y as batch size.\nFor each instance, it computes the smooth l1 loss element by element first\nand then sums all the losses. So the shape of Out is [batch_size, 1].\n\nThe equation is:\n$$\nOut_{\\sigma}(X, Y)_i = \\begin{cases}\n0.5 * (\\sigma * (X_i - Y_i)) ^ 2\n\\quad |X_i - Y_i| \\lt \\frac{1} {{\\sigma} ^ 2} \\\\\n\\frac{|X_i - Y_i| - 0.5}{{\\sigma}^2},\n\\quad otherwise\n\\end{cases}\n$$\n\nIn the above equation, $Out_{\\sigma}(X, Y)_i$, $X_i$ and $Y_i$ represent the ith\nelement of Out, X and Y.\n\n", + "type" : "unpool", + "comment" : "\n \"Input shape: $(N, C_{in}, H_{in}, W_{in})$\n Output shape: $(N, C_{out}, H_{out}, W_{out})$\n Where\n $$\n H_{out} = (H_{in}−1) * strides[0] − 2 * paddings[0] + ksize[0] \\\\\n W_{out} = (W_{in}−1) * strides[1] − 2 * paddings[1] + ksize[1]\n $$\n Paper: http://www.matthewzeiler.com/wp-content/uploads/2017\n /07/iccv2011.pdf\n ", "inputs" : [ { "name" : "X", - "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. The input value of smooth l1 loss op with shape [batch_size, dim1, ..., dimN].", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Y", - "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. The target value of smooth l1 loss op with same shape as X.", + "comment" : "(Tensor) The input tensor of unpool operator. The format of input tensor is NCHW. Where N is batch size, C is the number of channels, H and W is the height and width of feature.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "InsideWeight", - "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. This input is optional and should have same shape with X. If provided, the result of (X - Y) will be multiplied by this tensor element by element.", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "OutsideWeight", - "comment" : "(Tensor, default Tensor) A tensor with rank at least 2. This input is optional and should have same shape with X. If provided, the out smooth l1 loss will be multiplied by this tensor element by element.", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Diff", - "comment" : "Intermediate variable to cache InsideWeight * (X - Y).", - "duplicable" : 0, - "intermediate" : 1 - }, { - "name" : "Out", - "comment" : "(Tensor, default Tensor) A tensor with rank be 2. The output smooth l1 loss with shape [batch_size, 1].", - "duplicable" : 0, - "intermediate" : 0 - } ], - "attrs" : [ - { - "name" : "sigma", - "type" : "float", - "comment" : "Hyper parameter of smooth l1 loss op.A float scalar with default value 3.0.", - "generated" : 0 - } ] -},{ - "type" : "log_loss", - "comment" : "\nLogLoss Operator.\n\nLog loss is a loss function used for binary classification. Log Loss quantifies\nthe accuracy of a classifier by penalising false classifications. Minimising the\nLog Loss is equivalent to maximising the accuracy of the classifier. We define\nPredicted as the values predicted by our model and Labels as the target ground\ntruth value. Log loss can evaluate how close the predicted values are to the\ntarget. The shapes of Predicted and Labels are both [batch_size, 1].\nThe equation is:\n\n$$\nLoss = - Labels * log(Predicted + \\epsilon) -\n (1 - Labels) * log(1 - Predicted + \\epsilon)\n$$\n\n", - "inputs" : [ - { - "name" : "Predicted", - "comment" : "The input value (Predicted) of Log loss op.Predicted is a 2-D tensor with shape [batch_size, 1].", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Labels", - "comment" : "The target value (Labels) of Log loss op.Labels is a 2-D tensor with shape [batch_size, 1].", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Loss", - "comment" : "The output tensor with shape [batch_size, 1] which represents the log loss.", - "duplicable" : 0, - "intermediate" : 0 - } ], - "attrs" : [ - { - "name" : "epsilon", - "type" : "float", - "comment" : "Epsilon in log loss.", - "generated" : 0 - } ] -},{ - "type" : "unpool", - "comment" : "\n \"Input shape: $(N, C_{in}, H_{in}, W_{in})$\n Output shape: $(N, C_{out}, H_{out}, W_{out})$\n Where\n $$\n H_{out} = (H_{in}−1) * strides[0] − 2 * paddings[0] + ksize[0] \\\\\n W_{out} = (W_{in}−1) * strides[1] − 2 * paddings[1] + ksize[1]\n $$\n Paper: http://www.matthewzeiler.com/wp-content/uploads/2017\n /07/iccv2011.pdf\n ", - "inputs" : [ - { - "name" : "X", - "comment" : "(Tensor) The input tensor of unpool operator. The format of input tensor is NCHW. Where N is batch size, C is the number of channels, H and W is the height and width of feature.", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Indices", - "comment" : "(Tensor) The input tensor of the indices given out by MaxPool2d. The format of input tensor is NCHW. Where N is batch size, C is the number of channels, H and W is the height and width of feature.", + "name" : "Indices", + "comment" : "(Tensor) The input tensor of the indices given out by MaxPool2d. The format of input tensor is NCHW. Where N is batch size, C is the number of channels, H and W is the height and width of feature.", "duplicable" : 0, "intermediate" : 0 } ], @@ -2245,48 +2180,6 @@ "comment" : "(int, default 5 (FP32)) Output data type", "generated" : 0 } ] -},{ - "type" : "swish", - "comment" : "\nSwish Activation Operator.\n\n$$y = \\frac{x}{1 + e^{- \\beta x}}$$\n\n", - "inputs" : [ - { - "name" : "X", - "comment" : "Input of Swish operator", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Y", - "comment" : "Output of Swish operator", - "duplicable" : 0, - "intermediate" : 0 - } ], - "attrs" : [ - { - "name" : "beta", - "type" : "float", - "comment" : "Constant beta of swish operator", - "generated" : 0 - } ] -},{ - "type" : "is_empty", - "comment" : "\nIsEmpty Operator which checks whether a tensor is empty.\n\nIt will just return product(tensor.ddims()) > 0;\n ", - "inputs" : [ - { - "name" : "X", - "comment" : "(Tensor) Tensor which is to be checked.", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Out", - "comment" : "(Tensor) a boolean Tensor that indicate empty or not.", - "duplicable" : 0, - "intermediate" : 0 - } ], - "attrs" : [ ] },{ "type" : "reduce_max", "comment" : "\n{ReduceOp} Operator.\n\nThis operator computes the max of input tensor along the given dimension. \nThe result tensor has 1 fewer dimension than the input unless keep_dim is true.\nIf reduce_all is true, just reduce along all dimensions and output a scalar.\n\n", @@ -2322,431 +2215,474 @@ "generated" : 0 } ] },{ - "type" : "rank_loss", - "comment" : "\nRankLoss Operator.\n\nRankLoss operator for RankNet\n(http://icml.cc/2015/wp-content/uploads/2015/06/icml_ranking.pdf). \nRankNet is a pairwise ranking model with\none training sample consisting of a pair of doc A and B, and the label P\nindicating that A is ranked higher than B or not:\n\nP = {0, 1} or {0, 0.5, 1}, where 0.5 means no information about the rank of\nthe input pair.\n\nThe RankLoss operator takes three inputs: Left (o_i), Right (o_j) and Label\n(P_{i,j}), which represent the output score of RankNet for the two docs and \nthe label respectively, and yields the rank loss C_{i,j} using the following \nequation:\n\n$$\n C_{i,j} = -\\tilde{P_{ij}} * o_{i,j} + \\log(1 + e^{o_{i,j}}) \\\\\n o_{i,j} = o_i - o_j \\\\\n \\tilde{P_{i,j}} = \\left \\{0, 0.5, 1 \\right \\} \\ or \\ \\left \\{0, 1 \\right \\}\n$$\n\nThe operator can take batch inputs with size batch_size (batch_size >= 1).\n\n", + "type" : "shrink_rnn_memory", + "comment" : "\n In dynamic RNN, we are able to handle sequences of different lengths. \n Because of the multiple lengths, the size of each step input can be \n different, which may lead to a mismatching between the input of\n the current step and the memory generated by the previous one. This \n operator shrinks memory according to the size of the next step input, \n to make sure that they can match each other.\n ", "inputs" : [ { - "name" : "Label", - "comment" : "(2-D Tensor with shape [batch_size x 1]) The label indicating A ranked higher than B or not.", + "name" : "X", + "comment" : "(LoDTensor) The RNN step memory to be shrinked.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "Left", - "comment" : "(2-D Tensor with shape [batch_size x 1]) The output of RankNet for doc A.", + "name" : "RankTable", + "comment" : "(LoDRankTable) The lod_rank_table of dynamic RNN.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "Right", - "comment" : "(2-D Tensor with shape [batch_size x 1]) The output of RankNet for doc B.", + "name" : "I", + "comment" : "(LoDTensor) The step index. The RNN step memory 'X' will be shrinked to match the size of the input of the index'th step.", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(2-D Tensor with shape [batch_size x 1]) The output loss of RankLoss operator.", + "comment" : "(LoDTensor) The shrinked RNN step memory.", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ ] },{ - "type" : "greater_than", - "comment" : "greater_than Operator\n\nIt operates element-wise on X and Y, and returns the Out. Each of them is a\nN-dim tensor. X and Y could be any type. The each element of the Out tensor is\ncalculated by Out = X > Y\n", + "type" : "lod_reset", + "comment" : "LoDReset operator\n\nReset LoD of Input(X) into a new one specified by Input(TargetLoD) or\nAttr(target_lod), or set LoD for Input(X) if it doesn't have one.\nCurrently the lod_reset operator only supports the reset of level 0 LoD.\nAt least one of Input(TargetLoD) and Attr(target_lod) must be set,\nand if both of them are set, Input(TargetLoD) will be chosen as the\ntarget LoD.\n\nAn example:\nGiven a float LoDTensor X with shape (6, 1), its transpose form represents\n\n [1.0, 2.0, 3.0, 4.0, 5.0, 6.0],\n\nwith LoD = [[0, 2, 5, 6]] and the three (transposed) sequences look like\n\n [1.0, 2.0], [3.0, 4.0, 5.0], [6.0].\n\nIf target LoD = [0, 4, 6], the lod_reset operator will reset the LoD and\nthe sequences that the LoDTensor Output(Out) contains becomes:\n\n [1.0, 2.0, 3.0, 4.0], [5.0, 6.0].\n\n", "inputs" : [ { "name" : "X", - "comment" : "(LoDTensor) the left hand operand of greater_than operator", + "comment" : "(LoDTensor) The input tensor of lod_reset operator.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "Y", - "comment" : "(LoDTensor) the right hand operand of greater_than operator", + "name" : "TargetLoD", + "comment" : "(Tensor, optional) The target level 0 LoD from Input().", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(LoDTensor) n-dim bool tensor. Each element is Out = X > Y", + "comment" : "(LoDTensor) The output tensor of lod_reset operator.", "duplicable" : 0, "intermediate" : 0 } ], - "attrs" : [ ] + "attrs" : [ + { + "name" : "target_lod", + "type" : "int array", + "comment" : "The target level 0 LoD from Attr().", + "generated" : 0 + } ] },{ - "type" : "sequence_softmax", - "comment" : "\nSequence Softmax Operator.\n\nSequenceSoftmaxOp computes the softmax activation among all time-steps for each\nsequence. The dimension of each time-step should be 1. Thus, the shape of\ninput Tensor can be either [N, 1] or [N], where N is the sum of the length\nof all sequences.\n\nThe algorithm works as follows:\n for i-th sequence in a mini-batch:\n $$Out(X[lod[i]:lod[i+1]], :) =\n \\frac{\\exp(X[lod[i]:lod[i+1], :])}\n {\\sum(\\exp(X[lod[i]:lod[i+1], :]))}$$\n\nFor example, for a mini-batch of 3 sequences with variable-length,\neach containing 2, 3, 2 time-steps, the lod of which is [0, 2, 5, 7],\nthen softmax will be computed among X[0:2, :], X[2:5, :], X[5:7, :]\nand N turns out to be 7.\n\n", + "type" : "write_to_array", + "comment" : "\nWriteToArray Operator.\n\nThis operator writes a LoDTensor to a LoDTensor array.\n\nAssume $T$ is LoDTensor, $i$ is the subscript of the array, and $A$ is the array. The\nequation is\n\n$$A[i] = T$$\n\n", "inputs" : [ { "name" : "X", - "comment" : "(LoDTensor) 1-D or 2-D input LoDTensor with the 2-nd dimension of length 1.", + "comment" : "(LoDTensor) the tensor will be written to tensor array", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "I", + "comment" : "(Tensor) the subscript index in tensor array. The number of element should be 1", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(LoDTensor) 1-D or 2-D output LoDTensor with the 2-nd dimension of length 1.", + "comment" : "(TensorArray) the tensor array will be written", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ ] },{ - "type" : "momentum", - "comment" : "\nMomentum Optimizer.\n\nThis optimizer has a flag for Nestrov Momentum.\nThe update equations are as follows:\n\n$$\nvelocity = mu * velocity + gradient \\\\\nif (use\\_nesterov): \\\\\n param = param - gradient * learning\\_rate + mu * velocity * learning\\_rate \\\\\nelse: \\\\\n param = param - learning\\_rate * velocity. \\\\\n$$\n\n", + "type" : "sequence_pool", + "comment" : "\nSequence Pool Operator.\n\nThe SequencePoolOp pools features of all time-steps of each instance.\nIt supports six pooling types:\n1. AVERAGE: $$Out[i] = \\frac{\\sum_i X_i}{N}$$\n2. SUM: $$Out[i] = \\sum_jX_{ij}$$\n3. SQRT: $$Out[i] = \\frac{\\sum_jX_{ij}}{\\sqrt{len(X_i)}}$$\n4. LAST: Out[i] = last instance in i-th sequence X[i]\n5. FIRST: Out[i] = first instance in i-th sequence X[i]\n6. MAX: $$Out[i] = max(X_i)$$\n\nThe following example explains how this works:\nFor a mini-batch of 3 variable-length sentences,\ncontaining 2, 3, and 2 time-steps:\n\nAssume X is a [7,M,N] LoDTensor, and X->lod()[0] = [0, 2, 5, 7], 7=2+3+2.\nBesides, for the sake of simplicity, we assume M=1 and N=1,\nand the value of X = [[1, 3], [2, 4, 6], [5, 1]].\n\nThus, Out is a [3,1,1] Tensor without LoD infomation.\nAnd for different pooltype, the value of Out is as follows:\n\n- AVERAGE: [2, 4, 3], where 2=(1+3)/2, 4=(2+4+6)/3, 3=(5+1)/2\n- SUM: [4, 12, 6], where 4=1+3, 12=2+4+6, 6=5+1\n- SQRT: [2.82, 6.93, 4.24], where 2.82=(1+3)/sqrt(2),\n 6.93=(2+4+6)/sqrt(3), 4.24=(5+1)/sqrt(2)\n- MAX: [3, 6, 5], where 3=max(1,3), 6=max(2,4,6), 5=max(5,1)\n- LAST: [3, 6, 1], where 3=last(1,3), 6=last(2,4,6), 1=last(5,1)\n- FIRST: [1, 2, 5], where 1=first(1,3), 2=first(2,4,6), 5=first(5,1)\n\n ", "inputs" : [ { - "name" : "Param", - "comment" : "(Tensor, default Tensor) Input parameter that has to be updated", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Grad", - "comment" : "(Tensor, default Tensor) Input gradient of the parameter", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Velocity", - "comment" : "(Tensor, default Tensor) Input velocity (corresponding to the parameter) that has to be updated", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "LearningRate", - "comment" : "(Tensor, default Tensor) Input learning rate", + "name" : "X", + "comment" : "(LoDTensor) The variable-length input of SequencePoolOp", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { - "name" : "ParamOut", - "comment" : "(Tensor) This output is updated parameter. It shared memory with Input(Param).", + "name" : "Out", + "comment" : "(Tensor) The output of SequencePoolOp does not contain LoD infomation.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "VelocityOut", - "comment" : "(Tensor) This output is updated velocity. It shared memory with Input(Velocity).", + "name" : "MaxIndex", + "comment" : "(Tensor) This tensor is used for the sequence max-pooling to record the max indexes.", "duplicable" : 0, - "intermediate" : 0 + "intermediate" : 1 } ], "attrs" : [ { - "name" : "mu", - "type" : "float", - "comment" : "(float) Momentum coefficient", - "generated" : 0 - }, { - "name" : "use_nesterov", - "type" : "bool", - "comment" : "(bool, default false) Use Nesterov Momentum", + "name" : "pooltype", + "type" : "string", + "comment" : "(int, default AVERAGE) the pooling pooltype of SequencePoolOp.", "generated" : 0 } ] },{ - "type" : "scatter", - "comment" : "\nScatter Operator.\n\nThis operator obtains output by updating the input on selected indices on the first axis:\n\n$$\nOut = Ref \\\\\nOut[Index] = Ref[Index] + Updates\n$$\n\n", + "type" : "spp", + "comment" : "\n \"With spatial pyramid pooling, the input image can\n be of any sizes. This not only allows arbitrary aspect\n ratios, but also allows arbitrary scales. We can resize\n the input image to any scale (e.g., min(w, h)=180, 224,\n ...) and apply the same deep network. When the\n input image is at different scales, the network (with\n the same filter sizes) will extract features at different\n scales. The scales play important roles in traditional\n methods.\n Input shape: $(N, C_{in}, H_{in}, W_{in})$\n Output shape: $(H_{out}, W_{out})$\n Where\n $$\n H_{out} = N \\\\\n W_{out} = (((4^pyramid_height) - 1) / (4 - 1))$ * C_{in}\n $$\n paper https://arxiv.org/pdf/1406.4729v4.pdf\n ", "inputs" : [ { - "name" : "Ref", - "comment" : "The source input of scatter op", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Index", - "comment" : "The index input of scatter op where Ref will be updated", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Updates", - "comment" : "The updated value of updates op", + "name" : "X", + "comment" : "(Tensor) The input tensor of spp operator. The format of input tensor is NCHW. Where N is batch size, C is the number of channels, H and W is the height and width of feature.", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "The output of add op", + "comment" : "(Tensor) The output tensor of spp operator.N * M.M = C * H * W", "duplicable" : 0, "intermediate" : 0 } ], - "attrs" : [ ] + "attrs" : [ + { + "name" : "pyramid_height", + "type" : "int", + "comment" : "(int), multi level pooling", + "generated" : 0 + }, { + "name" : "pooling_type", + "type" : "string", + "comment" : "(string), pooling type, can be \"max\" for max-pooling and \"avg\" for average-pooling.", + "generated" : 0 + } ] },{ - "type" : "shrink_rnn_memory", - "comment" : "\n In dynamic RNN, we are able to handle sequences of different lengths. \n Because of the multiple lengths, the size of each step input can be \n different, which may lead to a mismatching between the input of\n the current step and the memory generated by the previous one. This \n operator shrinks memory according to the size of the next step input, \n to make sure that they can match each other.\n ", + "type" : "conv2d_transpose_cudnn", + "comment" : "\nConvolution2D Transpose Operator.\n\nThe convolution transpose operation calculates the output based on the input, filter\nand strides, paddings, groups parameters. The size of each dimension of the\nparameters is checked in the infer-shape.\nInput(Input) and output(Output) are in NCHW format. Where N is batchsize, C is the\nnumber of channels, H is the height of the feature, and W is the width of the feature.\nFilter(Input) is in MCHW format. Where M is the number of input feature channels,\nC is the number of output feature channels, H is the height of the filter,\nand W is the width of the filter.\nParameters(strides, paddings) are two elements. These two elements represent height\nand width, respectively.\nThe input(X) size and output(Out) size may be different.\n\nExample:\n Input:\n Input shape: $(N, C_{in}, H_{in}, W_{in})$\n Filter shape: $(C_{in}, C_{out}, H_f, W_f)$\n Output:\n Output shape: $(N, C_{out}, H_{out}, W_{out})$\n Where\n $$\n H_{out} = (H_{in} - 1) * strides[0] - 2 * paddings[0] + H_f \\\\\n W_{out} = (W_{in} - 1) * strides[1] - 2 * paddings[1] + W_f\n $$\n", "inputs" : [ { - "name" : "X", - "comment" : "(LoDTensor) The RNN step memory to be shrinked.", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "RankTable", - "comment" : "(LoDRankTable) The lod_rank_table of dynamic RNN.", + "name" : "Input", + "comment" : "(Tensor) The input tensor of convolution transpose operator. The format of input tensor is NCHW. Where N is batch size, C is the number of input channels, H is the height of the feature, and W is the width of the feature.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "I", - "comment" : "(LoDTensor) The step index. The RNN step memory 'X' will be shrinked to match the size of the input of the index'th step.", + "name" : "Filter", + "comment" : "(Tensor) The filter tensor of convolution transpose operator. The format of the filter tensor is MCHW, where M is the number of input feature channels, C is the number of output feature channels,H is the height of the filter, and W is the width of the filter. We enforce groups number == 1 in the convolution transpose scenario.", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { - "name" : "Out", - "comment" : "(LoDTensor) The shrinked RNN step memory.", + "name" : "Output", + "comment" : "(Tensor) The output tensor of convolution transpose operator. The format of output tensor is also NCHW.", "duplicable" : 0, "intermediate" : 0 } ], - "attrs" : [ ] + "attrs" : [ + { + "name" : "strides", + "type" : "int array", + "comment" : "(vector default:{1, 1}), the strides(h_stride, w_stride) of convolution transpose operator.", + "generated" : 0 + }, { + "name" : "paddings", + "type" : "int array", + "comment" : "(vector default:{0, 0}), the paddings(h_pad, w_pad) of convolution transpose operator.", + "generated" : 0 + }, { + "name" : "dilations", + "type" : "int array", + "comment" : "dilations of convolution operator.", + "generated" : 0 + }, { + "name" : "workspace_size_MB", + "type" : "int", + "comment" : "workspace size for cudnn, in MB, workspace is a section of GPU memory which will be allocated/freed each time the operator runs, larger workspace size can increase performance but also requires better hardward. This size should be carefully setted.", + "generated" : 0 + } ] },{ - "type" : "lod_reset", - "comment" : "LoDReset operator\n\nReset LoD of Input(X) into a new one specified by Input(TargetLoD) or\nAttr(target_lod), or set LoD for Input(X) if it doesn't have one.\nCurrently the lod_reset operator only supports the reset of level 0 LoD.\nAt least one of Input(TargetLoD) and Attr(target_lod) must be set,\nand if both of them are set, Input(TargetLoD) will be chosen as the\ntarget LoD.\n\nAn example:\nGiven a float LoDTensor X with shape (6, 1), its transpose form represents\n\n [1.0, 2.0, 3.0, 4.0, 5.0, 6.0],\n\nwith LoD = [[0, 2, 5, 6]] and the three (transposed) sequences look like\n\n [1.0, 2.0], [3.0, 4.0, 5.0], [6.0].\n\nIf target LoD = [0, 4, 6], the lod_reset operator will reset the LoD and\nthe sequences that the LoDTensor Output(Out) contains becomes:\n\n [1.0, 2.0, 3.0, 4.0], [5.0, 6.0].\n\n", + "type" : "merge_lod_tensor", + "comment" : "\n Merge True and False branches of LoDTensor into a single Output,\n with a mask at certain lod level. X is used to obtain complete\n lod information. Please refer to SplitLoDTensorOp.", "inputs" : [ { "name" : "X", - "comment" : "(LoDTensor) The input tensor of lod_reset operator.", + "comment" : "The input LoDTensor, contains complete lod information to construct the output", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "TargetLoD", - "comment" : "(Tensor, optional) The target level 0 LoD from Input().", + "name" : "Mask", + "comment" : "A bool column vector which mask the input", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "InTrue", + "comment" : "The True branch to be merged", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "InFalse", + "comment" : "The False branch to be merged", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(LoDTensor) The output tensor of lod_reset operator.", + "comment" : "The merged output LoDTensor", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ { - "name" : "target_lod", - "type" : "int array", - "comment" : "The target level 0 LoD from Attr().", + "name" : "level", + "type" : "int", + "comment" : "(int) the specific lod level to rank.", "generated" : 0 } ] },{ - "type" : "write_to_array", - "comment" : "\nWriteToArray Operator.\n\nThis operator writes a LoDTensor to a LoDTensor array.\n\nAssume $T$ is LoDTensor, $i$ is the subscript of the array, and $A$ is the array. The\nequation is\n\n$$A[i] = T$$\n\n", + "type" : "rank_loss", + "comment" : "\nRankLoss Operator.\n\nRankLoss operator for RankNet\n(http://icml.cc/2015/wp-content/uploads/2015/06/icml_ranking.pdf). \nRankNet is a pairwise ranking model with\none training sample consisting of a pair of doc A and B, and the label P\nindicating that A is ranked higher than B or not:\n\nP = {0, 1} or {0, 0.5, 1}, where 0.5 means no information about the rank of\nthe input pair.\n\nThe RankLoss operator takes three inputs: Left (o_i), Right (o_j) and Label\n(P_{i,j}), which represent the output score of RankNet for the two docs and \nthe label respectively, and yields the rank loss C_{i,j} using the following \nequation:\n\n$$\n C_{i,j} = -\\tilde{P_{ij}} * o_{i,j} + \\log(1 + e^{o_{i,j}}) \\\\\n o_{i,j} = o_i - o_j \\\\\n \\tilde{P_{i,j}} = \\left \\{0, 0.5, 1 \\right \\} \\ or \\ \\left \\{0, 1 \\right \\}\n$$\n\nThe operator can take batch inputs with size batch_size (batch_size >= 1).\n\n", "inputs" : [ { - "name" : "X", - "comment" : "(LoDTensor) the tensor will be written to tensor array", + "name" : "Label", + "comment" : "(2-D Tensor with shape [batch_size x 1]) The label indicating A ranked higher than B or not.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "I", - "comment" : "(Tensor) the subscript index in tensor array. The number of element should be 1", + "name" : "Left", + "comment" : "(2-D Tensor with shape [batch_size x 1]) The output of RankNet for doc A.", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Right", + "comment" : "(2-D Tensor with shape [batch_size x 1]) The output of RankNet for doc B.", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(TensorArray) the tensor array will be written", + "comment" : "(2-D Tensor with shape [batch_size x 1]) The output loss of RankLoss operator.", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ ] },{ - "type" : "logical_or", - "comment" : "logical_or Operator\n\nIt operates element-wise on X and Y, and returns the Out. X, Y and Out are N-dim boolean tensors.\nEach element of Out is calculated by $$Out = X || Y$$\n", + "type" : "greater_than", + "comment" : "greater_than Operator\n\nIt operates element-wise on X and Y, and returns the Out. Each of them is a\nN-dim tensor. X and Y could be any type. The each element of the Out tensor is\ncalculated by Out = X > Y\n", "inputs" : [ { "name" : "X", - "comment" : "(LoDTensor) Left hand operand of logical_or operator", + "comment" : "(LoDTensor) the left hand operand of greater_than operator", "duplicable" : 0, "intermediate" : 0 }, { "name" : "Y", - "comment" : "(LoDTensor) Right hand operand of logical_or operator", + "comment" : "(LoDTensor) the right hand operand of greater_than operator", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(LoDTensor) n-dim bool tensor. Each element is $$Out = X || Y$$", + "comment" : "(LoDTensor) n-dim bool tensor. Each element is Out = X > Y", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ ] },{ - "type" : "seq_expand", - "comment" : "\nSeq Expand Operator.\n\nThis operator expands input(X) according to LOD of input(Y).\nFollowing are cases to better explain how this works:\nCase 1:\n\nGiven 2-level a LoDTensor input(X)\n X.lod = [[0, 2, 3],\n [0, 1, 3, 4]]\n X.data = [a, b, c, d]\n X.dims = [4, 1]\nand input(Y)\n Y.lod = [[0, 2, 4],\n [0, 3, 6, 7, 8]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 2-level LoDTensor\n Out.lod = [[0, 2, 4],\n [0, 3, 6, 7, 8]]\n Out.data = [a, a, a, b, b, b, c, d]\n Out.dims = [8, 1]\n\nCase 2:\n\nGiven a 0-level LoDTensor input(X)\n X.data = [a, b, c]\n X.lod = NULL\n X.dims = [3, 1]\nand input(Y)\n Y.lod = [[0, 2, 3, 6]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 1-level LoDTensor\n Out.lod = [[0, 2, 3, 6]]\n Out.data = [a, a, b, c, c, c]\n Out.dims = [6, 1]\n\nCase 3:\n\nGiven a 0-level LoDTensor input(X)\n X.data = [[a, b], [c, d], [e, f]]\n X.lod = NULL\n X.dims = [3, 2]\nand input(Y)\n Y.lod = [[0, 2, 3, 6]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 1-level LoDTensor\n Out.lod = [[0, 2, 3, 6]]\n Out.data = [[a,b], [a,b] [c,d], [e, f], [e, f], [e, f]]\n Out.dims = [6, 2]\n\nCase 4:\n\nGiven 2-level a LoDTensor input(X)\n X.lod = [[0, 2, 3],\n [0, 1, 3, 4]]\n X.data = [a, b, c, d]\n X.dims = [4, 1]\nand input(Y)\n Y.lod = [[0, 2, 4],\n [0, 3, 6, 6, 8]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 2-level LoDTensor\n Out.lod = [[0, 2, 4],\n [0, 3, 6, 6, 8]]\n Out.data = [a, a, a, b, b, b, d, d]\n Out.dims = [8, 1]\n\n\n", + "type" : "sequence_softmax", + "comment" : "\nSequence Softmax Operator.\n\nSequenceSoftmaxOp computes the softmax activation among all time-steps for each\nsequence. The dimension of each time-step should be 1. Thus, the shape of\ninput Tensor can be either [N, 1] or [N], where N is the sum of the length\nof all sequences.\n\nThe algorithm works as follows:\n for i-th sequence in a mini-batch:\n $$Out(X[lod[i]:lod[i+1]], :) =\n \\frac{\\exp(X[lod[i]:lod[i+1], :])}\n {\\sum(\\exp(X[lod[i]:lod[i+1], :]))}$$\n\nFor example, for a mini-batch of 3 sequences with variable-length,\neach containing 2, 3, 2 time-steps, the lod of which is [0, 2, 5, 7],\nthen softmax will be computed among X[0:2, :], X[2:5, :], X[5:7, :]\nand N turns out to be 7.\n\n", "inputs" : [ { "name" : "X", - "comment" : "(Tensor or LoDTensor) The input(X) of this operator can be a LoDTensor or a base Tensor.", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Y", - "comment" : "(LoDTensor)The reference input(Y) of seq_expand op.It must be a LoDTensor with k-level(k>0).The input(X) will be expanded according to LOD of input(Y).The element numbers of last level in input(Y) must be equal to dims[0] of input(X).", + "comment" : "(LoDTensor) 1-D or 2-D input LoDTensor with the 2-nd dimension of length 1.", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(LodTensor)The output of seq_expand op.The lod of output will be as same as input(Y)'s lod.", + "comment" : "(LoDTensor) 1-D or 2-D output LoDTensor with the 2-nd dimension of length 1.", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ ] },{ - "type" : "scale", - "comment" : "\nScale operator\n\n$$Out = scale*X$$\n", + "type" : "momentum", + "comment" : "\nMomentum Optimizer.\n\nThis optimizer has a flag for Nestrov Momentum.\nThe update equations are as follows:\n\n$$\nvelocity = mu * velocity + gradient \\\\\nif (use\\_nesterov): \\\\\n param = param - gradient * learning\\_rate + mu * velocity * learning\\_rate \\\\\nelse: \\\\\n param = param - learning\\_rate * velocity. \\\\\n$$\n\n", "inputs" : [ { - "name" : "X", - "comment" : "(Tensor) Input tensor of scale operator.", + "name" : "Param", + "comment" : "(Tensor, default Tensor) Input parameter that has to be updated", "duplicable" : 0, "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Out", - "comment" : "(Tensor) Output tensor of scale operator.", + }, { + "name" : "Grad", + "comment" : "(Tensor, default Tensor) Input gradient of the parameter", "duplicable" : 0, "intermediate" : 0 - } ], - "attrs" : [ - { - "name" : "scale", - "type" : "float", - "comment" : "(float, default 0)The scaling factor of the scale operator.", - "generated" : 0 - } ] -},{ - "type" : "reduce_sum", - "comment" : "\n{ReduceOp} Operator.\n\nThis operator computes the sum of input tensor along the given dimension. \nThe result tensor has 1 fewer dimension than the input unless keep_dim is true.\nIf reduce_all is true, just reduce along all dimensions and output a scalar.\n\n", - "inputs" : [ - { - "name" : "X", - "comment" : "(Tensor) The input tensor. Tensors with rank at most 6 are supported.", + }, { + "name" : "Velocity", + "comment" : "(Tensor, default Tensor) Input velocity (corresponding to the parameter) that has to be updated", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "LearningRate", + "comment" : "(Tensor, default Tensor) Input learning rate", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { - "name" : "Out", - "comment" : "(Tensor) The result tensor.", + "name" : "ParamOut", + "comment" : "(Tensor) This output is updated parameter. It shared memory with Input(Param).", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "VelocityOut", + "comment" : "(Tensor) This output is updated velocity. It shared memory with Input(Velocity).", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ { - "name" : "dim", - "type" : "int", - "comment" : "(int, default 0) The dimension to reduce. Must be in the range [-rank(input), rank(input)). If `dim < 0`, the dim to reduce is `rank + dim`. Note that reducing on the first dim will make the LoD info lost.", - "generated" : 0 - }, { - "name" : "keep_dim", - "type" : "bool", - "comment" : "(bool, default false) If true, retain the reduced dimension with length 1.", + "name" : "mu", + "type" : "float", + "comment" : "(float) Momentum coefficient", "generated" : 0 }, { - "name" : "reduce_all", + "name" : "use_nesterov", "type" : "bool", - "comment" : "(bool, default false) If true, output a scalar reduced along all dimensions.", + "comment" : "(bool, default false) Use Nesterov Momentum", "generated" : 0 } ] },{ - "type" : "stanh", - "comment" : "\nSTanh Activation Operator.\n\n$$y = b * \\frac{e^{a * x} - e^{-a * x}}{e^{a * x} + e^{-a * x}}$$\n\n", + "type" : "scatter", + "comment" : "\nScatter Operator.\n\nThis operator obtains output by updating the input on selected indices on the first axis:\n\n$$\nOut = Ref \\\\\nOut[Index] = Ref[Index] + Updates\n$$\n\n", "inputs" : [ { - "name" : "X", - "comment" : "Input of STanh operator", + "name" : "Ref", + "comment" : "The source input of scatter op", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Index", + "comment" : "The index input of scatter op where Ref will be updated", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Updates", + "comment" : "The updated value of updates op", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { - "name" : "Y", - "comment" : "Output of STanh operator", + "name" : "Out", + "comment" : "The output of add op", "duplicable" : 0, "intermediate" : 0 } ], - "attrs" : [ - { - "name" : "scale_a", - "type" : "float", - "comment" : "The scale parameter of a for the input", - "generated" : 0 - }, { - "name" : "scale_b", - "type" : "float", - "comment" : "The scale parameter of b for the input", - "generated" : 0 - } ] + "attrs" : [ ] },{ - "type" : "adamax", - "comment" : "\nAdamax Optimizer.\n\nWe implement the Adamax optimizer from Section 7 of the Adam\npaper: https://arxiv.org/abs/1412.6980. Adamax is a variant of the\nAdam algorithm based on the infinity norm.\n\nAdamax updates:\n\n$$\nmoment\\_out = \\beta_1 * moment + (1 - \\beta_1) * grad \\\\\ninf\\_norm\\_out = max(\\beta_2 * inf\\_norm + \\epsilon, |grad|) \\\\\nlearning\\_rate = \\frac{learning\\_rate}{1 - \\beta_{1\\_pow}} \\\\\nparam\\_out = param - learning\\_rate * \\frac{moment\\_out}{inf\\_norm\\_out}\n$$\n\nThe original paper does not have an epsilon attribute.\nHowever, it is added here for numerical stability to prevent the\ndivision by 0 error.\n\n", + "type" : "logical_or", + "comment" : "logical_or Operator\n\nIt operates element-wise on X and Y, and returns the Out. X, Y and Out are N-dim boolean tensors.\nEach element of Out is calculated by $$Out = X || Y$$\n", "inputs" : [ { - "name" : "Param", - "comment" : "(Tensor) Input parameter", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "Grad", - "comment" : "(Tensor) Input gradient", + "name" : "X", + "comment" : "(LoDTensor) Left hand operand of logical_or operator", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "LearningRate", - "comment" : "(Tensor) Learning rate", + "name" : "Y", + "comment" : "(LoDTensor) Right hand operand of logical_or operator", "duplicable" : 0, "intermediate" : 0 - }, { - "name" : "Moment", - "comment" : "(Tensor) First moment", + } ], + "outputs" : [ + { + "name" : "Out", + "comment" : "(LoDTensor) n-dim bool tensor. Each element is $$Out = X || Y$$", "duplicable" : 0, "intermediate" : 0 - }, { - "name" : "InfNorm", - "comment" : "(Tensor) Input exponentially weighted infinity norm", + } ], + "attrs" : [ ] +},{ + "type" : "seq_expand", + "comment" : "\nSeq Expand Operator.\n\nThis operator expands input(X) according to LOD of input(Y).\nFollowing are cases to better explain how this works:\nCase 1:\n\nGiven 2-level a LoDTensor input(X)\n X.lod = [[0, 2, 3],\n [0, 1, 3, 4]]\n X.data = [a, b, c, d]\n X.dims = [4, 1]\nand input(Y)\n Y.lod = [[0, 2, 4],\n [0, 3, 6, 7, 8]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 2-level LoDTensor\n Out.lod = [[0, 2, 4],\n [0, 3, 6, 7, 8]]\n Out.data = [a, a, a, b, b, b, c, d]\n Out.dims = [8, 1]\n\nCase 2:\n\nGiven a 0-level LoDTensor input(X)\n X.data = [a, b, c]\n X.lod = NULL\n X.dims = [3, 1]\nand input(Y)\n Y.lod = [[0, 2, 3, 6]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 1-level LoDTensor\n Out.lod = [[0, 2, 3, 6]]\n Out.data = [a, a, b, c, c, c]\n Out.dims = [6, 1]\n\nCase 3:\n\nGiven a 0-level LoDTensor input(X)\n X.data = [[a, b], [c, d], [e, f]]\n X.lod = NULL\n X.dims = [3, 2]\nand input(Y)\n Y.lod = [[0, 2, 3, 6]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 1-level LoDTensor\n Out.lod = [[0, 2, 3, 6]]\n Out.data = [[a,b], [a,b] [c,d], [e, f], [e, f], [e, f]]\n Out.dims = [6, 2]\n\nCase 4:\n\nGiven 2-level a LoDTensor input(X)\n X.lod = [[0, 2, 3],\n [0, 1, 3, 4]]\n X.data = [a, b, c, d]\n X.dims = [4, 1]\nand input(Y)\n Y.lod = [[0, 2, 4],\n [0, 3, 6, 6, 8]]\nwith condition len(Y.lod[-1]) -1 == X.dims[0]\nthen we get 2-level LoDTensor\n Out.lod = [[0, 2, 4],\n [0, 3, 6, 6, 8]]\n Out.data = [a, a, a, b, b, b, d, d]\n Out.dims = [8, 1]\n\n\n", + "inputs" : [ + { + "name" : "X", + "comment" : "(Tensor or LoDTensor) The input(X) of this operator can be a LoDTensor or a base Tensor.", "duplicable" : 0, "intermediate" : 0 }, { - "name" : "Beta1Pow", - "comment" : "(Tensor) Input beta1 power accumulator", + "name" : "Y", + "comment" : "(LoDTensor)The reference input(Y) of seq_expand op.It must be a LoDTensor with k-level(k>0).The input(X) will be expanded according to LOD of input(Y).The element numbers of last level in input(Y) must be equal to dims[0] of input(X).", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { - "name" : "ParamOut", - "comment" : "(Tensor) Output parameter", + "name" : "Out", + "comment" : "(LodTensor)The output of seq_expand op.The lod of output will be as same as input(Y)'s lod.", "duplicable" : 0, "intermediate" : 0 - }, { - "name" : "MomentOut", - "comment" : "(Tensor) Output first moment", + } ], + "attrs" : [ ] +},{ + "type" : "scale", + "comment" : "\nScale operator\n\n$$Out = scale*X$$\n", + "inputs" : [ + { + "name" : "X", + "comment" : "(Tensor) Input tensor of scale operator.", "duplicable" : 0, "intermediate" : 0 - }, { - "name" : "InfNormOut", - "comment" : "(Tensor) Output exponentially weighted infinity norm", + } ], + "outputs" : [ + { + "name" : "Out", + "comment" : "(Tensor) Output tensor of scale operator.", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ { - "name" : "beta1", + "name" : "scale", "type" : "float", - "comment" : "(float, default 0.9) Exponential decay rate for the 1st moment estimates.", + "comment" : "(float, default 0)The scaling factor of the scale operator.", + "generated" : 0 + } ] +},{ + "type" : "reduce_sum", + "comment" : "\n{ReduceOp} Operator.\n\nThis operator computes the sum of input tensor along the given dimension. \nThe result tensor has 1 fewer dimension than the input unless keep_dim is true.\nIf reduce_all is true, just reduce along all dimensions and output a scalar.\n\n", + "inputs" : [ + { + "name" : "X", + "comment" : "(Tensor) The input tensor. Tensors with rank at most 6 are supported.", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "Out", + "comment" : "(Tensor) The result tensor.", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ + { + "name" : "dim", + "type" : "int", + "comment" : "(int, default 0) The dimension to reduce. Must be in the range [-rank(input), rank(input)). If `dim < 0`, the dim to reduce is `rank + dim`. Note that reducing on the first dim will make the LoD info lost.", "generated" : 0 }, { - "name" : "beta2", - "type" : "float", - "comment" : "(float, default 0.999) exponential decay rate for the weighted infinity norm estimates.", + "name" : "keep_dim", + "type" : "bool", + "comment" : "(bool, default false) If true, retain the reduced dimension with length 1.", "generated" : 0 }, { - "name" : "epsilon", - "type" : "float", - "comment" : "(float, default 1.0e-8) Constant for numerical stability", + "name" : "reduce_all", + "type" : "bool", + "comment" : "(bool, default false) If true, output a scalar reduced along all dimensions.", "generated" : 0 } ] },{ @@ -3045,29 +2981,163 @@ } ], "attrs" : [ ] },{ - "type" : "increment", - "comment" : "\nIncrement Operator.\n\nThe equation is: \n$$Out = X + step$$\n\n", + "type" : "l1_norm", + "comment" : "\nL1 Norm Operator.\n\nComputes the L1 norm of a tensor.\n\n$$Out = \\sum{|X|}$$\n\n", "inputs" : [ { "name" : "X", - "comment" : "(Tensor) The input tensor of increment operator", + "comment" : "(Tensor) The input of l1_norm op.", "duplicable" : 0, "intermediate" : 0 } ], "outputs" : [ { "name" : "Out", - "comment" : "(Tensor) The output tensor of increment operator.", + "comment" : "(Scalar) The output of l1_norm op.", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ ] +},{ + "type" : "stanh", + "comment" : "\nSTanh Activation Operator.\n\n$$y = b * \\frac{e^{a * x} - e^{-a * x}}{e^{a * x} + e^{-a * x}}$$\n\n", + "inputs" : [ + { + "name" : "X", + "comment" : "Input of STanh operator", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "Y", + "comment" : "Output of STanh operator", "duplicable" : 0, "intermediate" : 0 } ], "attrs" : [ { - "name" : "step", + "name" : "scale_a", "type" : "float", - "comment" : "(float, default 1.0) The step size by which the input tensor will be incremented.", + "comment" : "The scale parameter of a for the input", + "generated" : 0 + }, { + "name" : "scale_b", + "type" : "float", + "comment" : "The scale parameter of b for the input", + "generated" : 0 + } ] +},{ + "type" : "adamax", + "comment" : "\nAdamax Optimizer.\n\nWe implement the Adamax optimizer from Section 7 of the Adam\npaper: https://arxiv.org/abs/1412.6980. Adamax is a variant of the\nAdam algorithm based on the infinity norm.\n\nAdamax updates:\n\n$$\nmoment\\_out = \\beta_1 * moment + (1 - \\beta_1) * grad \\\\\ninf\\_norm\\_out = max(\\beta_2 * inf\\_norm + \\epsilon, |grad|) \\\\\nlearning\\_rate = \\frac{learning\\_rate}{1 - \\beta_{1\\_pow}} \\\\\nparam\\_out = param - learning\\_rate * \\frac{moment\\_out}{inf\\_norm\\_out}\n$$\n\nThe original paper does not have an epsilon attribute.\nHowever, it is added here for numerical stability to prevent the\ndivision by 0 error.\n\n", + "inputs" : [ + { + "name" : "Param", + "comment" : "(Tensor) Input parameter", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Grad", + "comment" : "(Tensor) Input gradient", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "LearningRate", + "comment" : "(Tensor) Learning rate", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Moment", + "comment" : "(Tensor) First moment", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "InfNorm", + "comment" : "(Tensor) Input exponentially weighted infinity norm", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "Beta1Pow", + "comment" : "(Tensor) Input beta1 power accumulator", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "ParamOut", + "comment" : "(Tensor) Output parameter", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "MomentOut", + "comment" : "(Tensor) Output first moment", + "duplicable" : 0, + "intermediate" : 0 + }, { + "name" : "InfNormOut", + "comment" : "(Tensor) Output exponentially weighted infinity norm", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ + { + "name" : "beta1", + "type" : "float", + "comment" : "(float, default 0.9) Exponential decay rate for the 1st moment estimates.", + "generated" : 0 + }, { + "name" : "beta2", + "type" : "float", + "comment" : "(float, default 0.999) exponential decay rate for the weighted infinity norm estimates.", + "generated" : 0 + }, { + "name" : "epsilon", + "type" : "float", + "comment" : "(float, default 1.0e-8) Constant for numerical stability", + "generated" : 0 + } ] +},{ + "type" : "swish", + "comment" : "\nSwish Activation Operator.\n\n$$y = \\frac{x}{1 + e^{- \\beta x}}$$\n\n", + "inputs" : [ + { + "name" : "X", + "comment" : "Input of Swish operator", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "Y", + "comment" : "Output of Swish operator", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ + { + "name" : "beta", + "type" : "float", + "comment" : "Constant beta of swish operator", "generated" : 0 } ] +},{ + "type" : "is_empty", + "comment" : "\nIsEmpty Operator which checks whether a tensor is empty.\n\nIt will just return product(tensor.ddims()) > 0;\n ", + "inputs" : [ + { + "name" : "X", + "comment" : "(Tensor) Tensor which is to be checked.", + "duplicable" : 0, + "intermediate" : 0 + } ], + "outputs" : [ + { + "name" : "Out", + "comment" : "(Tensor) a boolean Tensor that indicate empty or not.", + "duplicable" : 0, + "intermediate" : 0 + } ], + "attrs" : [ ] },{ "type" : "rmsprop", "comment" : "\nRmsprop Optimizer. \n\n$$\nMeanSquareOut = decay * MeanSquare + (1 - decay) * Grad * Grad \\\\\nMomentOut = momentum * Moment +\n \\frac{LearningRate * Grad}{\\sqrt{MeanSquareOut + epsilon}} \\\\\nParamOut = Param - MomentOut\n$$\n\nThe original slides that proposed Rmsprop: Slide 29 of\nhttp://www.cs.toronto.edu/~tijmen/csc321/slides/lecture_slides_lec6.pdf)\n\n", @@ -4755,47 +4825,6 @@ "comment" : "The number of thresholds to use when discretizing the roc curve.", "generated" : 0 } ] -},{ - "type" : "read_from_array", - "comment" : "\nReadFromArray Operator.\n\nRead a LoDTensor from a LoDTensor Array.\n\nAssume $T$ is LoDTensor, $i$ is the subscript of the array, and $A$ is the array. The\nequation is\n\n$$T = A[i]$$\n\n", - "inputs" : [ - { - "name" : "X", - "comment" : "(TensorArray) the array will be read from.", - "duplicable" : 0, - "intermediate" : 0 - }, { - "name" : "I", - "comment" : "(Tensor) the subscript index in tensor array. The number of element should be 1", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Out", - "comment" : "(LoDTensor) the tensor will be read from.", - "duplicable" : 0, - "intermediate" : 0 - } ], - "attrs" : [ ] -},{ - "type" : "softplus", - "comment" : "\nSoftplus Activation Operator.\n\n$y = \\ln(1 + e^{x})$\n\n", - "inputs" : [ - { - "name" : "X", - "comment" : "Input of Softplus operator", - "duplicable" : 0, - "intermediate" : 0 - } ], - "outputs" : [ - { - "name" : "Y", - "comment" : "Output of Softplus operator", - "duplicable" : 0, - "intermediate" : 0 - } ], - "attrs" : [ ] },{ "type" : "split", "comment" : "\nSplit operator\n\nThis operator splits the input tensor into multiple sub-tensors.\n\nExample:\n Input = [[1,2],\n [3,4],\n [5,6]]\n sections = [2,1]\n axis = 0\n Output[0] = [[1,2],\n [3,4]]\n Output[1] = [[5,6]]\n\n ", -- GitLab