CompositeComponent is a component representing a sequence of [simple] components. More...
#include <nnet-simple-component.h>
Public Member Functions | |
virtual int32 | InputDim () const |
Returns input-dimension of this component. More... | |
virtual int32 | OutputDim () const |
Returns output-dimension of this component. More... | |
virtual std::string | Info () const |
Returns some text-form information about this component, for diagnostics. More... | |
virtual void | InitFromConfig (ConfigLine *cfl) |
Initialize, from a ConfigLine object. More... | |
virtual Component * | Copy () const |
Copies component (deep copy). More... | |
CompositeComponent () | |
void | Init (const std::vector< Component *> &components, int32 max_rows_process) |
virtual std::string | Type () const |
Returns a string such as "SigmoidComponent", describing the type of the object. More... | |
virtual int32 | Properties () const |
Return bitmask of the component's properties. More... | |
virtual void * | Propagate (const ComponentPrecomputedIndexes *indexes, const CuMatrixBase< BaseFloat > &in, CuMatrixBase< BaseFloat > *out) const |
Propagate function. More... | |
virtual void | Backprop (const std::string &debug_info, const ComponentPrecomputedIndexes *indexes, const CuMatrixBase< BaseFloat > &in_value, const CuMatrixBase< BaseFloat > &, const CuMatrixBase< BaseFloat > &out_deriv, void *memo, Component *to_update, CuMatrixBase< BaseFloat > *in_deriv) const |
Backprop function; depending on which of the arguments 'to_update' and 'in_deriv' are non-NULL, this can compute input-data derivatives and/or perform model update. More... | |
virtual void | ZeroStats () |
Components that provide an implementation of StoreStats should also provide an implementation of ZeroStats(), to set those stats to zero. More... | |
virtual void | Read (std::istream &is, bool binary) |
Read function (used after we know the type of the Component); accepts input that is missing the token that describes the component type, in case it has already been consumed. More... | |
virtual void | Write (std::ostream &os, bool binary) const |
Write component to stream. More... | |
virtual void | SetUnderlyingLearningRate (BaseFloat lrate) |
Sets the learning rate of gradient descent- gets multiplied by learning_rate_factor_. More... | |
virtual void | SetActualLearningRate (BaseFloat lrate) |
Sets the learning rate directly, bypassing learning_rate_factor_. More... | |
virtual void | SetAsGradient () |
Sets is_gradient_ to true and sets learning_rate_ to 1, ignoring learning_rate_factor_. More... | |
virtual void | Scale (BaseFloat scale) |
This virtual function when called on – an UpdatableComponent scales the parameters by "scale" when called by an UpdatableComponent. More... | |
virtual void | Add (BaseFloat alpha, const Component &other) |
This virtual function when called by – an UpdatableComponent adds the parameters of another updatable component, times some constant, to the current parameters. More... | |
virtual void | PerturbParams (BaseFloat stddev) |
This function is to be used in testing. More... | |
virtual BaseFloat | DotProduct (const UpdatableComponent &other) const |
Computes dot-product between parameters of two instances of a Component. More... | |
virtual int32 | NumParameters () const |
The following new virtual function returns the total dimension of the parameters in this class. More... | |
virtual void | Vectorize (VectorBase< BaseFloat > *params) const |
Turns the parameters into vector form. More... | |
virtual void | UnVectorize (const VectorBase< BaseFloat > ¶ms) |
Converts the parameters from vector form. More... | |
virtual void | FreezeNaturalGradient (bool freeze) |
virtual More... | |
int32 | NumComponents () const |
const Component * | GetComponent (int32 i) const |
Gets the ith component in this component. More... | |
void | SetComponent (int32 i, Component *component) |
Sets the ith component. More... | |
virtual | ~CompositeComponent () |
Public Member Functions inherited from UpdatableComponent | |
UpdatableComponent (const UpdatableComponent &other) | |
UpdatableComponent () | |
virtual | ~UpdatableComponent () |
virtual BaseFloat | LearningRateFactor () |
virtual void | SetLearningRateFactor (BaseFloat lrate_factor) |
void | SetUpdatableConfigs (const UpdatableComponent &other) |
BaseFloat | LearningRate () const |
Gets the learning rate to be used in gradient descent. More... | |
BaseFloat | MaxChange () const |
Returns the per-component max-change value, which is interpreted as the maximum change (in l2 norm) in parameters that is allowed per minibatch for this component. More... | |
void | SetMaxChange (BaseFloat max_change) |
BaseFloat | L2Regularization () const |
Returns the l2 regularization constant, which may be set in any updatable component (usually from the config file). More... | |
void | SetL2Regularization (BaseFloat a) |
Public Member Functions inherited from Component | |
virtual void | StoreStats (const CuMatrixBase< BaseFloat > &in_value, const CuMatrixBase< BaseFloat > &out_value, void *memo) |
This function may store stats on average activation values, and for some component types, the average value of the derivative of the nonlinearity. More... | |
virtual void | GetInputIndexes (const MiscComputationInfo &misc_info, const Index &output_index, std::vector< Index > *desired_indexes) const |
This function only does something interesting for non-simple Components. More... | |
virtual bool | IsComputable (const MiscComputationInfo &misc_info, const Index &output_index, const IndexSet &input_index_set, std::vector< Index > *used_inputs) const |
This function only does something interesting for non-simple Components, and it exists to make it possible to manage optionally-required inputs. More... | |
virtual void | ReorderIndexes (std::vector< Index > *input_indexes, std::vector< Index > *output_indexes) const |
This function only does something interesting for non-simple Components. More... | |
virtual ComponentPrecomputedIndexes * | PrecomputeIndexes (const MiscComputationInfo &misc_info, const std::vector< Index > &input_indexes, const std::vector< Index > &output_indexes, bool need_backprop) const |
This function must return NULL for simple Components. More... | |
virtual void | DeleteMemo (void *memo) const |
This virtual function only needs to be overwritten by Components that return a non-NULL memo from their Propagate() function. More... | |
virtual void | ConsolidateMemory () |
This virtual function relates to memory management, and avoiding fragmentation. More... | |
Component () | |
virtual | ~Component () |
Private Member Functions | |
MatrixStrideType | GetStrideType (int32 i) const |
bool | IsUpdatable () const |
Private Attributes | |
int32 | max_rows_process_ |
std::vector< Component * > | components_ |
Additional Inherited Members | |
Static Public Member Functions inherited from Component | |
static Component * | ReadNew (std::istream &is, bool binary) |
Read component from stream (works out its type). Dies on error. More... | |
static Component * | NewComponentOfType (const std::string &type) |
Returns a new Component of the given type e.g. More... | |
Protected Member Functions inherited from UpdatableComponent | |
void | InitLearningRatesFromConfig (ConfigLine *cfl) |
std::string | ReadUpdatableCommon (std::istream &is, bool binary) |
void | WriteUpdatableCommon (std::ostream &is, bool binary) const |
Protected Attributes inherited from UpdatableComponent | |
BaseFloat | learning_rate_ |
learning rate (typically 0.0..0.01) More... | |
BaseFloat | learning_rate_factor_ |
learning rate factor (normally 1.0, but can be set to another < value so that when < you call SetLearningRate(), that value will be scaled by this factor. More... | |
BaseFloat | l2_regularize_ |
L2 regularization constant. More... | |
bool | is_gradient_ |
True if this component is to be treated as a gradient rather than as parameters. More... | |
BaseFloat | max_change_ |
configuration value for imposing max-change More... | |
CompositeComponent is a component representing a sequence of [simple] components.
The config line would be something like the following (imagine this is all on one line):
component name=composite1 type=CompositeComponent max-rows-process=2048 num-components=3 \ component1='type=BlockAffineComponent input-dim=1000 output-dim=10000 num-blocks=100' \ component2='type=RectifiedLinearComponent dim=10000' \ component3='type=BlockAffineComponent input-dim=10000 output-dim=1000 num-blocks=100'
The reason you might want to use this component, instead of directly using the same sequence of components in the config file, is to save GPU memory (at the expense of more compute)– because doing it like this means we have to re-do parts of the forward pass in the backprop phase, but we avoid using much memory for very long (and you can make the memory usage very small by making max-rows-process small). We inherit from UpdatableComponent just in case one or more of the components in the sequence are updatable.
It is an error to nest a CompositeComponent inside a CompositeComponent. The same effect can be accomplished by specifying a smaller max-rows-process in a single CompositeComponent.
Definition at line 1971 of file nnet-simple-component.h.
|
inline |
Definition at line 1982 of file nnet-simple-component.h.
References PnormComponent::Init().
|
inlinevirtual |
Definition at line 2050 of file nnet-simple-component.h.
References kaldi::DeletePointers().
This virtual function when called by – an UpdatableComponent adds the parameters of another updatable component, times some constant, to the current parameters.
– a NonlinearComponent (or another component that stores stats, like BatchNormComponent)– it relates to adding stats. Otherwise it will normally do nothing.
Reimplemented from Component.
Definition at line 4423 of file nnet-simple-component.cc.
References PerElementScaleComponent::Add(), CompositeComponent::components_, rnnlm::i, and KALDI_ASSERT.
|
virtual |
Backprop function; depending on which of the arguments 'to_update' and 'in_deriv' are non-NULL, this can compute input-data derivatives and/or perform model update.
[in] | debug_info | The component name, to be printed out in any warning messages. |
[in] | indexes | A pointer to some information output by this class's PrecomputeIndexes function (will be NULL for simple components, i.e. those that don't do things like splicing). |
[in] | in_value | The matrix that was given as input to the Propagate function. Will be ignored (and may be empty) if Properties()&kBackpropNeedsInput == 0. |
[in] | out_value | The matrix that was output from the Propagate function. Will be ignored (and may be empty) if Properties()&kBackpropNeedsOutput == 0 |
[in] | out_deriv | The derivative at the output of this component. |
[in] | memo | This will normally be NULL, but for component types that set the flag kUsesMemo, this will be the return value of the Propagate() function that corresponds to this Backprop() function. Ownership of any pointers is not transferred to the Backprop function; DeleteMemo() will be called to delete it. |
[out] | to_update | If model update is desired, the Component to be updated, else NULL. Does not have to be identical to this. If supplied, you can assume that to_update->Properties() & kUpdatableComponent is nonzero. |
[out] | in_deriv | The derivative at the input of this component, if needed (else NULL). If Properties()&kBackpropInPlace, may be the same matrix as out_deriv. If Properties()&kBackpropAdds, this is added to by the Backprop routine, else it is set. The component code chooses which mode to work in, based on convenience. |
Implements Component.
Definition at line 4277 of file nnet-simple-component.cc.
References PerElementScaleComponent::Backprop(), rnnlm::i, PerElementScaleComponent::InputDim(), KALDI_ASSERT, kaldi::nnet3::kBackpropAdds, kaldi::nnet3::kBackpropNeedsInput, kaldi::nnet3::kBackpropNeedsOutput, kaldi::nnet3::kPropagateAdds, kaldi::kSetZero, kaldi::nnet3::kStoresStats, kaldi::kUndefined, kaldi::nnet3::kUpdatableComponent, kaldi::nnet3::kUsesMemo, CuMatrixBase< Real >::NumCols(), CuMatrixBase< Real >::NumRows(), NVTX_RANGE, PerElementScaleComponent::OutputDim(), PerElementScaleComponent::Properties(), and Component::StoreStats().
|
virtual |
Copies component (deep copy).
Implements Component.
Definition at line 4567 of file nnet-simple-component.cc.
References NaturalGradientPerElementScaleComponent::Copy(), rnnlm::i, and CompositeComponent::Init().
|
virtual |
Computes dot-product between parameters of two instances of a Component.
Can be used for computing parameter-norm of an UpdatableComponent.
Implements UpdatableComponent.
Definition at line 4534 of file nnet-simple-component.cc.
References CompositeComponent::components_, UpdatableComponent::DotProduct(), rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, and PerElementScaleComponent::Properties().
|
virtual |
virtual
Reimplemented from UpdatableComponent.
Definition at line 4555 of file nnet-simple-component.cc.
References UpdatableComponent::FreezeNaturalGradient(), rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, and PerElementScaleComponent::Properties().
Gets the ith component in this component.
The ordering is the same as in the config line. The caller does not own the received component.
Definition at line 4633 of file nnet-simple-component.cc.
References rnnlm::i, and KALDI_ASSERT.
Referenced by kaldi::nnet3::ConvertRepeatedToBlockAffine(), and kaldi::nnet3::UnitTestConvertRepeatedToBlockAffineComposite().
|
inlineprivate |
Definition at line 4125 of file nnet-simple-component.cc.
References kaldi::kDefaultStride, kaldi::nnet3::kInputContiguous, kaldi::nnet3::kOutputContiguous, kaldi::kStrideEqualNumCols, and PerElementScaleComponent::Properties().
|
virtual |
Returns some text-form information about this component, for diagnostics.
Starts with the type of the component. E.g. "SigmoidComponent dim=900", although most components will have much more info.
Reimplemented from UpdatableComponent.
Definition at line 4405 of file nnet-simple-component.cc.
References rnnlm::i, NaturalGradientPerElementScaleComponent::Info(), and NaturalGradientPerElementScaleComponent::Type().
Definition at line 4184 of file nnet-simple-component.cc.
References kaldi::DeletePointers(), rnnlm::i, PerElementScaleComponent::InputDim(), KALDI_ASSERT, kaldi::nnet3::kSimpleComponent, PerElementScaleComponent::OutputDim(), and PerElementScaleComponent::Properties().
Referenced by CompositeComponent::Copy().
|
virtual |
Initialize, from a ConfigLine object.
[in] | cfl | A ConfigLine containing any parameters that are needed for initialization. For example: "dim=100 param-stddev=0.1" |
Implements Component.
Definition at line 4578 of file nnet-simple-component.cc.
References kaldi::DeletePointers(), ConfigLine::FirstToken(), ConfigLine::GetValue(), ConfigLine::HasUnusedValues(), rnnlm::i, NaturalGradientPerElementScaleComponent::Init(), Component::InitFromConfig(), KALDI_ERR, kaldi::nnet3::kRandomComponent, kaldi::nnet3::kSimpleComponent, Component::NewComponentOfType(), ConfigLine::ParseLine(), Component::Properties(), Component::Type(), ConfigLine::UnusedValues(), and ConfigLine::WholeLine().
|
virtual |
Returns input-dimension of this component.
Implements Component.
Definition at line 4091 of file nnet-simple-component.cc.
References KALDI_ASSERT.
|
private |
Definition at line 4082 of file nnet-simple-component.cc.
References kaldi::nnet3::kUpdatableComponent.
|
inline |
Definition at line 2040 of file nnet-simple-component.h.
References rnnlm::i.
Referenced by kaldi::nnet3::ConvertRepeatedToBlockAffine(), and kaldi::nnet3::UnitTestConvertRepeatedToBlockAffineComposite().
|
virtual |
The following new virtual function returns the total dimension of the parameters in this class.
Reimplemented from UpdatableComponent.
Definition at line 4486 of file nnet-simple-component.cc.
References rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, UpdatableComponent::NumParameters(), and PerElementScaleComponent::Properties().
|
virtual |
Returns output-dimension of this component.
Implements Component.
Definition at line 4097 of file nnet-simple-component.cc.
References KALDI_ASSERT.
|
virtual |
This function is to be used in testing.
It adds unit noise times "stddev" to the parameters of the component.
Implements UpdatableComponent.
Definition at line 4433 of file nnet-simple-component.cc.
References rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, UpdatableComponent::PerturbParams(), and PerElementScaleComponent::Properties().
|
virtual |
Propagate function.
[in] | indexes | A pointer to some information output by this class's PrecomputeIndexes function (will be NULL for simple components, i.e. those that don't do things like splicing). |
[in] | in | The input to this component. Num-columns == InputDim(). |
[out] | out | The output of this component. Num-columns == OutputDim(). Note: output of this component will be added to the initial value of "out" if Properties()&kPropagateAdds != 0; otherwise the output will be set and the initial value ignored. Each Component chooses whether it is more convenient implementation-wise to add or set, and the calling code has to deal with it. |
Implements Component.
Definition at line 4137 of file nnet-simple-component.cc.
References rnnlm::i, PerElementScaleComponent::InputDim(), KALDI_ASSERT, kaldi::nnet3::kPropagateAdds, kaldi::kSetZero, kaldi::kUndefined, CuMatrixBase< Real >::NumCols(), CuMatrixBase< Real >::NumRows(), PerElementScaleComponent::OutputDim(), and PerElementScaleComponent::Propagate().
|
virtual |
Return bitmask of the component's properties.
These properties depend only on the component's type. See enum ComponentProperties.
Implements Component.
Definition at line 4103 of file nnet-simple-component.cc.
References KALDI_ASSERT, kaldi::nnet3::kBackpropAdds, kaldi::nnet3::kBackpropNeedsInput, kaldi::nnet3::kBackpropNeedsOutput, kaldi::nnet3::kInputContiguous, kaldi::nnet3::kOutputContiguous, kaldi::nnet3::kPropagateAdds, kaldi::nnet3::kSimpleComponent, kaldi::nnet3::kStoresStats, and kaldi::nnet3::kUpdatableComponent.
|
virtual |
Read function (used after we know the type of the Component); accepts input that is missing the token that describes the component type, in case it has already been consumed.
Implements Component.
Definition at line 4203 of file nnet-simple-component.cc.
References kaldi::nnet3::ExpectToken(), rnnlm::i, NaturalGradientPerElementScaleComponent::Init(), UpdatableComponent::is_gradient_, KALDI_ERR, UpdatableComponent::learning_rate_, UpdatableComponent::learning_rate_factor_, kaldi::ReadBasicType(), Component::ReadNew(), kaldi::ReadToken(), and UpdatableComponent::ReadUpdatableCommon().
|
virtual |
This virtual function when called on – an UpdatableComponent scales the parameters by "scale" when called by an UpdatableComponent.
– a Nonlinear component (or another component that stores stats, like BatchNormComponent)– it relates to scaling activation stats, not parameters. Otherwise it will normally do nothing.
Reimplemented from Component.
Definition at line 4417 of file nnet-simple-component.cc.
References rnnlm::i, and PerElementScaleComponent::Scale().
|
virtual |
Sets the learning rate directly, bypassing learning_rate_factor_.
Reimplemented from UpdatableComponent.
Definition at line 4460 of file nnet-simple-component.cc.
References rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, PerElementScaleComponent::Properties(), and UpdatableComponent::SetActualLearningRate().
|
virtual |
Sets is_gradient_ to true and sets learning_rate_ to 1, ignoring learning_rate_factor_.
Reimplemented from UpdatableComponent.
Definition at line 4473 of file nnet-simple-component.cc.
References rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, PerElementScaleComponent::Properties(), and UpdatableComponent::SetAsGradient().
Sets the ith component.
After this call, CompositeComponent owns the reference to the argument component. Frees the previous ith component.
Definition at line 4638 of file nnet-simple-component.cc.
References rnnlm::i, and KALDI_ASSERT.
Referenced by kaldi::nnet3::ConvertRepeatedToBlockAffine().
|
virtual |
Sets the learning rate of gradient descent- gets multiplied by learning_rate_factor_.
Reimplemented from UpdatableComponent.
Definition at line 4444 of file nnet-simple-component.cc.
References rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, UpdatableComponent::LearningRate(), PerElementScaleComponent::Properties(), and UpdatableComponent::SetUnderlyingLearningRate().
|
inlinevirtual |
Returns a string such as "SigmoidComponent", describing the type of the object.
Implements Component.
Definition at line 1988 of file nnet-simple-component.h.
References Component::Add(), PnormComponent::Backprop(), kaldi::nnet3::DotProduct(), kaldi::nnet3::FreezeNaturalGradient(), kaldi::nnet3::NumParameters(), kaldi::nnet3::PerturbParams(), PnormComponent::Propagate(), PnormComponent::Properties(), PnormComponent::Read(), Component::Scale(), PnormComponent::Write(), and Component::ZeroStats().
|
virtual |
Converts the parameters from vector form.
Reimplemented from UpdatableComponent.
Definition at line 4517 of file nnet-simple-component.cc.
References VectorBase< Real >::Dim(), rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, UpdatableComponent::NumParameters(), PerElementScaleComponent::Properties(), and UpdatableComponent::UnVectorize().
|
virtual |
Turns the parameters into vector form.
We put the vector form on the CPU, because in the kinds of situations where we do this, we'll tend to use too much memory for the GPU.
Reimplemented from UpdatableComponent.
Definition at line 4500 of file nnet-simple-component.cc.
References VectorBase< Real >::Dim(), rnnlm::i, KALDI_ASSERT, kaldi::nnet3::kUpdatableComponent, UpdatableComponent::NumParameters(), PerElementScaleComponent::Properties(), and UpdatableComponent::Vectorize().
|
virtual |
Write component to stream.
Implements Component.
Definition at line 4263 of file nnet-simple-component.cc.
References rnnlm::i, NaturalGradientPerElementScaleComponent::Write(), kaldi::WriteBasicType(), kaldi::WriteToken(), and UpdatableComponent::WriteUpdatableCommon().
|
virtual |
Components that provide an implementation of StoreStats should also provide an implementation of ZeroStats(), to set those stats to zero.
Other components that store other types of statistics (e.g. regarding gradient clipping) should implement ZeroStats() also.
Reimplemented from Component.
Definition at line 4254 of file nnet-simple-component.cc.
References rnnlm::i, and Component::ZeroStats().
|
private |
Definition at line 2062 of file nnet-simple-component.h.
Referenced by CompositeComponent::Add(), and CompositeComponent::DotProduct().
|
private |
Definition at line 2061 of file nnet-simple-component.h.