diff options
author | 2018-04-18 19:49:12 -0700 | |
---|---|---|
committer | 2018-04-18 19:52:22 -0700 | |
commit | 542edb6dd64bd18d63ef1fd64c55a645c406f170 (patch) | |
tree | 61abe6f27c36f347a995193c7ef773292fefbf03 /tensorflow/go | |
parent | 6c85471ee06bf10f5034e2a8fb1fd6ab84dd7fbc (diff) |
Go: Update generated wrapper functions for TensorFlow ops.
PiperOrigin-RevId: 193456151
Diffstat (limited to 'tensorflow/go')
-rw-r--r-- | tensorflow/go/op/wrappers.go | 212 |
1 files changed, 106 insertions, 106 deletions
diff --git a/tensorflow/go/op/wrappers.go b/tensorflow/go/op/wrappers.go index a5b293ce75..f270eadc32 100644 --- a/tensorflow/go/op/wrappers.go +++ b/tensorflow/go/op/wrappers.go @@ -13107,6 +13107,112 @@ func FusedResizeAndPadConv2D(scope *Scope, input tf.Output, size tf.Output, padd return op.Output(0) } +// Returns a list of tensors with the same shapes and contents as the input +// +// tensors. +// +// This op can be used to override the gradient for complicated functions. For +// example, suppose y = f(x) and we wish to apply a custom function g for backprop +// such that dx = g(dy). In Python, +// +// ```python +// with tf.get_default_graph().gradient_override_map( +// {'IdentityN': 'OverrideGradientWithG'}): +// y, _ = identity_n([f(x), x]) +// +// @tf.RegisterGradient('OverrideGradientWithG') +// def ApplyG(op, dy, _): +// return [None, g(dy)] # Do not backprop to f(x). +// ``` +func IdentityN(scope *Scope, input []tf.Output) (output []tf.Output) { + if scope.Err() != nil { + return + } + opspec := tf.OpSpec{ + Type: "IdentityN", + Input: []tf.Input{ + tf.OutputList(input), + }, + } + op := scope.AddOperation(opspec) + if scope.Err() != nil { + return + } + var idx int + var err error + if output, idx, err = makeOutputList(op, idx, "output"); err != nil { + scope.UpdateErr("IdentityN", err) + return + } + return output +} + +// ResourceApplyCenteredRMSPropAttr is an optional argument to ResourceApplyCenteredRMSProp. +type ResourceApplyCenteredRMSPropAttr func(optionalAttr) + +// ResourceApplyCenteredRMSPropUseLocking sets the optional use_locking attribute to value. +// +// value: If `True`, updating of the var, mg, ms, and mom tensors is +// protected by a lock; otherwise the behavior is undefined, but may exhibit less +// contention. +// If not specified, defaults to false +func ResourceApplyCenteredRMSPropUseLocking(value bool) ResourceApplyCenteredRMSPropAttr { + return func(m optionalAttr) { + m["use_locking"] = value + } +} + +// Update '*var' according to the centered RMSProp algorithm. +// +// The centered RMSProp algorithm uses an estimate of the centered second moment +// (i.e., the variance) for normalization, as opposed to regular RMSProp, which +// uses the (uncentered) second moment. This often helps with training, but is +// slightly more expensive in terms of computation and memory. +// +// Note that in dense implementation of this algorithm, mg, ms, and mom will +// update even if the grad is zero, but in this sparse implementation, mg, ms, +// and mom will not update in iterations during which the grad is zero. +// +// mean_square = decay * mean_square + (1-decay) * gradient ** 2 +// mean_grad = decay * mean_grad + (1-decay) * gradient +// +// Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2) +// +// mg <- rho * mg_{t-1} + (1-rho) * grad +// ms <- rho * ms_{t-1} + (1-rho) * grad * grad +// mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon) +// var <- var - mom +// +// Arguments: +// var_: Should be from a Variable(). +// mg: Should be from a Variable(). +// ms: Should be from a Variable(). +// mom: Should be from a Variable(). +// lr: Scaling factor. Must be a scalar. +// rho: Decay rate. Must be a scalar. +// +// epsilon: Ridge term. Must be a scalar. +// grad: The gradient. +// +// Returns the created operation. +func ResourceApplyCenteredRMSProp(scope *Scope, var_ tf.Output, mg tf.Output, ms tf.Output, mom tf.Output, lr tf.Output, rho tf.Output, momentum tf.Output, epsilon tf.Output, grad tf.Output, optional ...ResourceApplyCenteredRMSPropAttr) (o *tf.Operation) { + if scope.Err() != nil { + return + } + attrs := map[string]interface{}{} + for _, a := range optional { + a(attrs) + } + opspec := tf.OpSpec{ + Type: "ResourceApplyCenteredRMSProp", + Input: []tf.Input{ + var_, mg, ms, mom, lr, rho, momentum, epsilon, grad, + }, + Attrs: attrs, + } + return scope.AddOperation(opspec) +} + // Adds `bias` to `value`. // // This is a deprecated version of BiasAdd and will be soon removed. @@ -23167,112 +23273,6 @@ func DecodeGif(scope *Scope, contents tf.Output) (image tf.Output) { return op.Output(0) } -// ResourceApplyCenteredRMSPropAttr is an optional argument to ResourceApplyCenteredRMSProp. -type ResourceApplyCenteredRMSPropAttr func(optionalAttr) - -// ResourceApplyCenteredRMSPropUseLocking sets the optional use_locking attribute to value. -// -// value: If `True`, updating of the var, mg, ms, and mom tensors is -// protected by a lock; otherwise the behavior is undefined, but may exhibit less -// contention. -// If not specified, defaults to false -func ResourceApplyCenteredRMSPropUseLocking(value bool) ResourceApplyCenteredRMSPropAttr { - return func(m optionalAttr) { - m["use_locking"] = value - } -} - -// Update '*var' according to the centered RMSProp algorithm. -// -// The centered RMSProp algorithm uses an estimate of the centered second moment -// (i.e., the variance) for normalization, as opposed to regular RMSProp, which -// uses the (uncentered) second moment. This often helps with training, but is -// slightly more expensive in terms of computation and memory. -// -// Note that in dense implementation of this algorithm, mg, ms, and mom will -// update even if the grad is zero, but in this sparse implementation, mg, ms, -// and mom will not update in iterations during which the grad is zero. -// -// mean_square = decay * mean_square + (1-decay) * gradient ** 2 -// mean_grad = decay * mean_grad + (1-decay) * gradient -// -// Delta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2) -// -// mg <- rho * mg_{t-1} + (1-rho) * grad -// ms <- rho * ms_{t-1} + (1-rho) * grad * grad -// mom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon) -// var <- var - mom -// -// Arguments: -// var_: Should be from a Variable(). -// mg: Should be from a Variable(). -// ms: Should be from a Variable(). -// mom: Should be from a Variable(). -// lr: Scaling factor. Must be a scalar. -// rho: Decay rate. Must be a scalar. -// -// epsilon: Ridge term. Must be a scalar. -// grad: The gradient. -// -// Returns the created operation. -func ResourceApplyCenteredRMSProp(scope *Scope, var_ tf.Output, mg tf.Output, ms tf.Output, mom tf.Output, lr tf.Output, rho tf.Output, momentum tf.Output, epsilon tf.Output, grad tf.Output, optional ...ResourceApplyCenteredRMSPropAttr) (o *tf.Operation) { - if scope.Err() != nil { - return - } - attrs := map[string]interface{}{} - for _, a := range optional { - a(attrs) - } - opspec := tf.OpSpec{ - Type: "ResourceApplyCenteredRMSProp", - Input: []tf.Input{ - var_, mg, ms, mom, lr, rho, momentum, epsilon, grad, - }, - Attrs: attrs, - } - return scope.AddOperation(opspec) -} - -// Returns a list of tensors with the same shapes and contents as the input -// -// tensors. -// -// This op can be used to override the gradient for complicated functions. For -// example, suppose y = f(x) and we wish to apply a custom function g for backprop -// such that dx = g(dy). In Python, -// -// ```python -// with tf.get_default_graph().gradient_override_map( -// {'IdentityN': 'OverrideGradientWithG'}): -// y, _ = identity_n([f(x), x]) -// -// @tf.RegisterGradient('OverrideGradientWithG') -// def ApplyG(op, dy, _): -// return [None, g(dy)] # Do not backprop to f(x). -// ``` -func IdentityN(scope *Scope, input []tf.Output) (output []tf.Output) { - if scope.Err() != nil { - return - } - opspec := tf.OpSpec{ - Type: "IdentityN", - Input: []tf.Input{ - tf.OutputList(input), - }, - } - op := scope.AddOperation(opspec) - if scope.Err() != nil { - return - } - var idx int - var err error - if output, idx, err = makeOutputList(op, idx, "output"); err != nil { - scope.UpdateErr("IdentityN", err) - return - } - return output -} - // Computes the gradient of the sigmoid of `x` wrt its input. // // Specifically, `grad = dy * y * (1 - y)`, where `y = sigmoid(x)`, and |