experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Remove more parts of DCASGD missed in the first pass. (47949b)

Browse Source 
PiperOrigin-RevId: 164914552
This commit is contained in:
A. Unique TensorFlower 2017-08-10 15:48:42 -07:00 committed by TensorFlower Gardener
parent 73b3d52c7b
commit b8d13d218f
4 changed files with 0 additions and 145 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
tensorflow/core/kernels/training_ops.h
View File

@ -34,15 +34,6 @@ struct ApplyGradientDescent {
typename TTypes<T>::ConstFlat delta);
};
template <typename Device, typename T>
struct ApplyDelayCompensatedGradientDescent {
void operator()(const Device& d, typename TTypes<T>::Flat var,
typename TTypes<T>::ConstScalar alpha,
typename TTypes<T>::ConstFlat delta,
typename TTypes<T>::ConstScalar lambda,
typename TTypes<T>::Flat shadow);
};
template <typename Device, typename T>
struct ApplyAdadelta {
void operator()(const Device& d, typename TTypes<T>::Flat var,

53
tensorflow/core/ops/compat/ops_history.v1.pbtxt
View File

@ -1035,59 +1035,6 @@ op {
}
}
}
op {
name: "ApplyDelayCompensatedGradientDescent"
input_arg {
name: "var"
type: DT_RESOURCE
}
input_arg {
name: "alpha"
type_attr: "T"
}
input_arg {
name: "delta"
type_attr: "T"
}
input_arg {
name: "lambda"
type_attr: "T"
}
input_arg {
name: "shadow"
type: DT_RESOURCE
}
attr {
name: "T"
type: "type"
allowed_values {
list {
type: DT_FLOAT
type: DT_DOUBLE
type: DT_INT64
type: DT_INT32
type: DT_UINT8
type: DT_UINT16
type: DT_INT16
type: DT_INT8
type: DT_COMPLEX64
type: DT_COMPLEX128
type: DT_QINT8
type: DT_QUINT8
type: DT_QINT32
type: DT_HALF
}
}
}
attr {
name: "use_locking"
type: "bool"
default_value {
b: false
}
}
is_stateful: true
}
op {
name: "ApplyFtrl"
input_arg {

61
tensorflow/core/ops/ops.pbtxt
View File

@ -1049,67 +1049,6 @@ op {
summary: "Update \'*var\' according to the centered RMSProp algorithm."
description: "The centered RMSProp algorithm uses an estimate of the centered second moment\n(i.e., the variance) for normalization, as opposed to regular RMSProp, which\nuses the (uncentered) second moment. This often helps with training, but is\nslightly more expensive in terms of computation and memory.\n\nNote that in dense implementation of this algorithm, mg, ms, and mom will\nupdate even if the grad is zero, but in this sparse implementation, mg, ms,\nand mom will not update in iterations during which the grad is zero.\n\nmean_square = decay * mean_square + (1-decay) * gradient ** 2\nmean_grad = decay * mean_grad + (1-decay) * gradient\n\nDelta = learning_rate * gradient / sqrt(mean_square + epsilon - mean_grad ** 2)\n\nmg <- rho * mg_{t-1} + (1-rho) * grad\nms <- rho * ms_{t-1} + (1-rho) * grad * grad\nmom <- momentum * mom_{t-1} + lr * grad / sqrt(ms - mg * mg + epsilon)\nvar <- var - mom"
}
op {
name: "ApplyDelayCompensatedGradientDescent"
input_arg {
name: "var"
description: "Should be from a Variable()."
type: DT_RESOURCE
}
input_arg {
name: "alpha"
description: "Scaling factor. Must be a scalar."
type_attr: "T"
}
input_arg {
name: "delta"
description: "The change."
type_attr: "T"
}
input_arg {
name: "lambda"
description: "The variance parameter."
type_attr: "T"
}
input_arg {
name: "shadow"
description: "Same as \"var\"."
type: DT_RESOURCE
}
attr {
name: "T"
type: "type"
allowed_values {
list {
type: DT_FLOAT
type: DT_DOUBLE
type: DT_INT64
type: DT_INT32
type: DT_UINT8
type: DT_UINT16
type: DT_INT16
type: DT_INT8
type: DT_COMPLEX64
type: DT_COMPLEX128
type: DT_QINT8
type: DT_QUINT8
type: DT_QINT32
type: DT_HALF
}
}
}
attr {
name: "use_locking"
type: "bool"
default_value {
b: false
}
description: "If `True`, the subtraction will be protected by a lock;\notherwise the behavior is undefined, but may exhibit less contention."
}
summary: "var -= alpha * (delta + lambda * delta * (var - shadow))"
description: "Update \'*shadow\' by changing it to the new value of \'var\'"
is_stateful: true
}
op {
name: "ApplyFtrl"
input_arg {

22
tensorflow/core/ops/training_ops.cc
View File

@ -103,28 +103,6 @@ use_locking: If `True`, the subtraction will be protected by a lock;
otherwise the behavior is undefined, but may exhibit less contention.
)doc");
REGISTER_OP("ApplyDelayCompensatedGradientDescent")
.Input("var: resource")
.Input("alpha: T")
.Input("delta: T")
.Input("lambda: T")
.Input("shadow: resource")
.Attr("T: numbertype")
.Attr("use_locking: bool = false")
.SetShapeFn(ApplyGradientDescentShapeFn)
.Doc(R"doc(
var -= alpha * (delta + lambda * delta * (var - shadow))
Update '*shadow' by changing it to the new value of 'var'
var: Should be from a Variable().
alpha: Scaling factor. Must be a scalar.
delta: The change.
lambda: The variance parameter.
shadow: Same as "var".
use_locking: If `True`, the subtraction will be protected by a lock;
otherwise the behavior is undefined, but may exhibit less contention.
)doc");
static Status ApplyProximalGradientDescentShapeFn(InferenceContext* c,
bool sparse) {
ShapeHandle unused;