提交 004df46f 编写于 作者: X xuwei06

Make print_op able to show the value of bool tensor

And some minor fixes on comments.
上级 432d2b5d
......@@ -314,7 +314,6 @@ EIGEN_FUNCTOR(Div, EIGEN_DIV);
template <typename DeviceContext, typename T, typename functor,
typename broadcastfunctor, typename broadcast2functor>
void ElementwiseGradCompute(const framework::ExecutionContext& ctx,
const framework::Tensor* x,
const framework::Tensor* y,
const framework::Tensor* out,
......
......@@ -46,7 +46,7 @@ struct Formater {
}
private:
void PrintMessage() { CLOG << std::time(nullptr) << "\t" << message; }
void PrintMessage() { CLOG << std::time(nullptr) << "\t" << message << "\t"; }
void PrintName() {
if (!name.empty()) {
CLOG << "Tensor[" << name << "]" << std::endl;
......@@ -85,15 +85,16 @@ struct Formater {
// print float
if (dtype.hash_code() == typeid(float).hash_code()) {
Display<float>(size);
}
if (dtype.hash_code() == typeid(double).hash_code()) {
} else if (dtype.hash_code() == typeid(double).hash_code()) {
Display<double>(size);
}
if (dtype.hash_code() == typeid(int).hash_code()) {
} else if (dtype.hash_code() == typeid(int).hash_code()) {
Display<int>(size);
}
if (dtype.hash_code() == typeid(int64_t).hash_code()) {
} else if (dtype.hash_code() == typeid(int64_t).hash_code()) {
Display<int64_t>(size);
} else if (dtype.hash_code() == typeid(bool).hash_code()) {
Display<bool>(size);
} else {
CLOG << "\tdata: unprintable type: " << dtype.name() << std::endl;
}
}
......@@ -182,6 +183,7 @@ class TensorPrintOp : public framework::OperatorBase {
}
Formater formater;
formater.message = Attr<std::string>("message");
if (Attr<bool>("print_tensor_name")) {
formater.name = printed_var_name;
}
......
......@@ -174,7 +174,7 @@ def Print(input,
print_tensor_type (bool): Print the tensor type.
print_tensor_shape (bool): Print the tensor shape.
print_tensor_lod (bool): Print the tensor lod.
print_phase (bool): Which phase to displace, including 'forward',
print_phase (str): Which phase to displace, including 'forward',
'backward' and 'both'. If set to 'backward' or 'both', will
print the gradients of input tensor.
......
......@@ -1579,7 +1579,7 @@ def layer_norm(input,
"""
**Layer Normalization**
Assume feature vectors exist on dimensions
Assume feature vectors exist on dimensions
:attr:`begin_norm_axis ... rank(input)` and calculate the moment statistics
along these dimensions for each feature vector :math:`a` with size
:math:`H`, then normalize each feature vector using the corresponding
......@@ -1600,13 +1600,13 @@ def layer_norm(input,
Args:
input(Variable): The input tensor variable.
scale(bool): Whether to learn the adaptive gain :math:`g` after
scale(bool): Whether to learn the adaptive gain :math:`g` after
normalization.
shift(bool): Whether to learn the adaptive bias :math:`b` after
shift(bool): Whether to learn the adaptive bias :math:`b` after
normalization.
begin_norm_axis(bool): The normalization will be performed along
begin_norm_axis(bool): The normalization will be performed along
dimensions from :attr:`begin_norm_axis` to :attr:`rank(input)`.
epsilon(float): The small value added to the variance to prevent
epsilon(float): The small value added to the variance to prevent
division by zero.
param_attr(ParamAttr|None): The parameter attribute for the learnable
gain :math:`g`.
......@@ -2070,7 +2070,7 @@ def reduce_sum(input, dim=None, keep_dim=False, name=None):
Tensor variable with a single element, otherwise must be in the
range :math:`[-rank(input), rank(input))`. If :math:`dim < 0`,
the dimension to reduce is :math:`rank + dim`.
keep_dim (bool): Whether to reserve the reduced dimension in the
keep_dim (bool|False): Whether to reserve the reduced dimension in the
output Tensor. The result tensor will have one fewer dimension
than the :attr:`input` unless :attr:`keep_dim` is true.
name(str|None): A name for this layer(optional). If set None, the layer
......@@ -3098,33 +3098,33 @@ def multiplex(inputs, index):
def softmax_with_cross_entropy(logits, label, soft_label=False):
"""
**Softmax With Cross Entropy Operator.**
Cross entropy loss with softmax is used as the output layer extensively. This
operator computes the softmax normalized values for each row of the input
tensor, after which cross-entropy loss is computed. This provides a more
numerically stable gradient.
Because this operator performs a softmax on logits internally, it expects
unscaled logits. This operator should not be used with the output of
softmax operator since that would produce incorrect results.
When the attribute soft_label is set false, this operators expects mutually
exclusive hard labels, each sample in a batch is in exactly one class with a
probability of 1.0. Each sample in the batch will have a single label.
The equation is as follows:
1) Hard label (one-hot label, so every sample has exactly one class)
.. math::
loss_j = -\\text{logit}_{label_j} +
\\log\\left(\\sum_{i=0}^{K}\\exp(\\text{logit}_i)\\right), j = 1,..., K
2) Soft label (each sample can have a distribution over all classes)
.. math::
loss_j = -\\sum_{i=0}^{K}\\text{label}_i
\\left(\\text{logit}_i - \\log\\left(\\sum_{i=0}^{K}
\\exp(\\text{logit}_i)\\right)\\right), j = 1,...,K
......@@ -3169,7 +3169,7 @@ def smooth_l1(x, y, inside_weight=None, outside_weight=None, sigma=None):
The operator takes the first dimension of X and Y as batch size.
For each instance, it computes the smooth l1 loss element by element first
and then sums all the losses. So the shape of Out is [batch_size, 1].
Args:
x (Variable): A tensor with rank at least 2. The input value of smooth
l1 loss op with shape [batch_size, dim1, ..., dimN].
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册