Module autograd::ops

A collection of functions to manipulate ag::Tensor objects

Modules

gradient_descent_ops

Provides gradient descent optimizers.

Functions

_hessian_vector_product	(Experimental) Computes hessian vector product
_range
abs	Returns the largest integer less than or equal to a number, element-wise.
acos	Elementwise arccos
acosh	Elementwise hyperbolic arccos
add	Addition.
add_n	Adds all input tensors, element-wise.
argmax	Takes argmax along specified axis.
asin	Elementwise arcsin
asinh	Elementwise hyperbolic arcsin
atan	Elementwise arctan
atanh	Elementwise hyperbolic arctan
batch_matmul	Batched matrix multiplication.
batch_matmul_t	Batched matrix multiplication with inputs's transposition.
batch_norm	Applies batch normalization.
bernoulli	Outputs values sampled from the bernoulli distribution.
bernoulli_rng	Outputs values sampled from the bernoulli distribution.
ceil	Returns the smallest integer greater than or equal to a number, element-wise.
clip	Limits all elements of x so as to be within [min, max]
concat	Concatenates input tensors along specified axis.
constant	Creates a constant tensor.
conv2d	2D convolution.
conv2d_transpose	2D transposed convolution.
convert_to_tensor	Converts an ndarray::Array to a ag::Tensor.
cos	Elementwise cosine
cosh	Elementwise hyperbolic cosine
dilated_conv2d	2D convolution with dilation.
dilated_conv2d_transpose	2D transposed convolution with dilation.
div	Division.
elu	Elementwise exponential linear unit.
equal	Compares two tensors and returns a binary tensor.
exp	Elementwise exponential
expand_dims	Expands specified dims.
flatten	Flattens input tensor into 1-ranked (vector).
floor	Returns the largest integer less than or equal to a number, element-wise.
gather	Gathers subviews from the input tensor.
gather_common	Gathers subviews from the input tensor.
grad	Returns gradient tensors wrt input tensors.
grad_with_default⚠	Computes gradients with ys's already known gradients.
greater	Returns a binary tensor.
greater_equal	Returns a binary tensor.
identity	Identity function without copy.
jacobians	Computes jacobians for variables.
leaky_relu	Elementwise leaky relu.
lesser	Returns a binary tensor.
lesser_equal	Returns a binary tensor.
log	Elementwise log
log_normal	Outputs values sampled from the log-normal distribution.
log_normal_rng	Outputs values sampled from the log-normal distribution.
log_softmax	Log softmax function.
matmul	Matrix multiplication.
matmul_t	Matrix multiplication with inputs's transposition.
max_pool2d	2D max pooling.
maximum	Returns the max of x and y (i.e. x > y ? x : y) element-wise.
minimum	Returns the min of x and y (i.e. x > y ? y : x) element-wise.
mul	Multiplication.
neg	Performs the - operation.
normalize	Normalizes the input tensor with its mean and variance along specified axis.
not_equal	Compares two tensors and returns a binary tensor.
ones	Returns ones with given shape.
placeholder	Creates a placeholder tensor.
pow	Elementwise pow
random_exp	Outputs values sampled from the exponential distribution.
random_exp_rng	Outputs values sampled from the exponential distribution.
random_gamma	Outputs values sampled from the gamma distribution.
random_gamma_rng	Outputs values sampled from the gamma distribution.
random_normal	Outputs values sampled from the normal distribution.
random_normal_rng	Outputs values sampled from the normal distribution.
random_uniform	Outputs values sampled from the uniform distribution.
random_uniform_rng	Outputs values sampled from the uniform distribution.
range	Returns a range.
rank	Returns the (symbolic) rank of input tensor
reciprocal	Returns the 1/x, element-wise.
reduce_logsumexp	Computes log(sum(exp(x))) along specified axis.
reduce_max	Takes max along specified axes.
reduce_mean	Takes mean along specified axes.
reduce_min	Takes min along specified axes.
reduce_prod	Takes product along specified axes.
reduce_sum	Takes sum along specified axes.
reduce_sum_to_scalar	Sum up all the elements to a scalar value (0-D Tensor).
relu	Elementwise rectified linear unit.
reshape	Reshapes input tensor.
scalar	Generates a zero-ranked tensor from a scalar value.
setdiff1d	Takes diff between two tensors.
shape	Returns the (symbolic) shape of input tensor
sigmoid	Elementwise logistic sigmoid function.
sigmoid_cross_entropy	Computes binary_cross_entropy(sigmoid(y), t).
sign	Returns -1 if x < 0, 0 if x==0, 1 if x > 0, element-wise.
sin	Elementwise sine
sinh	Elementwise hyperbolic sine
size	Returns the (symbolic) size of input tensor
slice	Slices the input tensor.
softmax	Computes softmax along specified axis
softmax_cross_entropy	Computes categorical_cross_entropy(softmax(y), t).
softplus	Elementwise softplus.
sparse_softmax_cross_entropy	A variant of softmax_cross_entropy.
split	Splits input tensors into parts.
sqrt	Elementwise sqrt
square	Takes square of the input.
squeeze	Squeezes specified dims.
standard_normal	Outputs values sampled from the standard normal distribution.
standard_normal_rng	Outputs values sampled from the standard normal distribution.
standard_uniform	Outputs values sampled from the standard uniform distribution.
standard_uniform_rng	Outputs values sampled from the standard uniform distribution.
stop_gradient	Stops gradient propagation.
sub	Subtraction.
tan	Elementwise tangent
tanh	Elementwise hyperbolic tangent
tensordot	Computes tensor-dot-product (tensor contraction) along specified axes.
tile	Tiles input tensor along specified axis.
transpose	Permutes dimensions.
variable	Creates a shared variable tensor from an ndarray.
zeros	Returns zeros with given shape.