SpliceMaxComponent Class Reference

This is as SpliceComponent but outputs the max of any of the inputs (taking the max across time). More...

#include <nnet-component.h>

Inheritance diagram for SpliceMaxComponent:

Collaboration diagram for SpliceMaxComponent:

Public Member Functions
	SpliceMaxComponent ()
void	Init (int32 dim, std::vector< int32 > context)
virtual std::string	Type () const
virtual std::string	Info () const
virtual void	InitFromString (std::string args)
	Initialize, typically from a line of a config file. More...
virtual int32	InputDim () const
	Get size of input vectors. More...
virtual int32	OutputDim () const
	Get size of output vectors. More...
virtual std::vector< int32 >	Context () const
	Return a vector describing the temporal context this component requires for each frame of output, as a sorted list. More...
virtual void	Propagate (const ChunkInfo &in_info, const ChunkInfo &out_info, const CuMatrixBase< BaseFloat > &in, CuMatrixBase< BaseFloat > *out) const
	Perform forward pass propagation Input->Output. More...
virtual void	Backprop (const ChunkInfo &in_info, const ChunkInfo &out_info, const CuMatrixBase< BaseFloat > &in_value, const CuMatrixBase< BaseFloat > &out_value, const CuMatrixBase< BaseFloat > &out_deriv, Component *to_update, CuMatrix< BaseFloat > *in_deriv) const
	Perform backward pass propagation of the derivative, and also either update the model (if to_update == this) or update another model or compute the model derivative (otherwise). More...
virtual bool	BackpropNeedsInput () const
virtual bool	BackpropNeedsOutput () const
virtual Component *	Copy () const
	Copy component (deep copy). More...
virtual void	Read (std::istream &is, bool binary)
virtual void	Write (std::ostream &os, bool binary) const
	Write component to stream. More...
Public Member Functions inherited from Component
	Component ()
virtual int32	Index () const
	Returns the index in the sequence of layers in the neural net; intended only to be used in debugging information. More...
virtual void	SetIndex (int32 index)
void	Propagate (const ChunkInfo &in_info, const ChunkInfo &out_info, const CuMatrixBase< BaseFloat > &in, CuMatrix< BaseFloat > *out) const
	A non-virtual propagate function that first resizes output if necessary. More...
virtual	~Component ()

Private Member Functions
	KALDI_DISALLOW_COPY_AND_ASSIGN (SpliceMaxComponent)

Private Attributes
int32	dim_
std::vector< int32 >	context_

Additional Inherited Members
Static Public Member Functions inherited from Component
static Component *	ReadNew (std::istream &is, bool binary)
	Read component from stream. More...
static Component *	NewFromString (const std::string &initializer_line)
	Initialize the Component from one line that will contain first the type, e.g. More...
static Component *	NewComponentOfType (const std::string &type)
	Return a new Component of the given type e.g. More...

Detailed Description

This is as SpliceComponent but outputs the max of any of the inputs (taking the max across time).

Definition at line 1133 of file nnet-component.h.

Constructor & Destructor Documentation

SpliceMaxComponent()

SpliceMaxComponent ( )

inline

Definition at line 1135 of file nnet-component.h.

{ }  // called only prior to Read() or Init().

Member Function Documentation

Backprop()

void Backprop ( const ChunkInfo &  in_info, const ChunkInfo &  out_info, const CuMatrixBase< BaseFloat > &  in_value, const CuMatrixBase< BaseFloat > &  out_value, const CuMatrixBase< BaseFloat > &  out_deriv, Component *  to_update, CuMatrix< BaseFloat > *  in_deriv  ) const

virtual

Perform backward pass propagation of the derivative, and also either update the model (if to_update == this) or update another model or compute the model derivative (otherwise).

Note: in_value and out_value are the values of the input and output of the component, and these may be dummy variables if respectively BackpropNeedsInput() or BackpropNeedsOutput() return false for that component (not all components need these).

num_chunks lets us treat the input matrix as contiguous-in-time chunks of equal size; it only matters if splicing is involved.

Implements Component.

Definition at line 2891 of file nnet-component.cc.

References ChunkInfo::Check(), ChunkInfo::CheckSize(), ChunkInfo::ChunkSize(), ChunkInfo::GetIndex(), ChunkInfo::GetOffset(), AffineComponent::InputDim(), KALDI_ASSERT, ChunkInfo::NumChunks(), ChunkInfo::NumCols(), CuMatrixBase< Real >::NumCols(), ChunkInfo::NumRows(), CuMatrixBase< Real >::NumRows(), and CuMatrix< Real >::Resize().

Referenced by kaldi::nnet2::BasicDebugTestForSpliceMax().

                                                                        {
  in_info.Check();
  out_info.Check();
  in_info.CheckSize(in_value);
  out_info.CheckSize(out_deriv);
  in_deriv->Resize(in_info.NumRows(), in_info.NumCols());
  KALDI_ASSERT(in_info.NumChunks() == out_info.NumChunks());

  int32 out_chunk_size = out_info.ChunkSize(),
         in_chunk_size = in_info.ChunkSize(),
                      dim = out_deriv.NumCols();

  KALDI_ASSERT(dim == InputDim());

  for (int32 chunk = 0; chunk < in_info.NumChunks(); chunk++) {
    CuSubMatrix<BaseFloat> in_deriv_chunk(*in_deriv,
                                        chunk * in_chunk_size,
                                        in_chunk_size,
                                        0, dim),
                         in_value_chunk(in_value,
                                        chunk * in_chunk_size,
                                        in_chunk_size,
                                        0, dim),
                        out_deriv_chunk(out_deriv,
                                        chunk * out_chunk_size,
                                        out_chunk_size,
                                        0, dim);
    for (int32 r = 0; r < out_deriv_chunk.NumRows(); r++) {
      int32 out_chunk_ind = r;
      int32 out_chunk_offset =
          out_info.GetOffset(out_chunk_ind);

      for (int32 c = 0; c < dim; c++) {
        int32 in_r_max = -1;
        BaseFloat max_input = -std::numeric_limits<BaseFloat>::infinity();
        for (int32 context_ind = 0;
             context_ind < context_.size(); context_ind++) {
          int32 in_r =
              in_info.GetIndex(out_chunk_offset + context_[context_ind]);
          BaseFloat input = in_value_chunk(in_r, c);
          if (input > max_input) {
            max_input = input;
            in_r_max = in_r;
          }
        }
        KALDI_ASSERT(in_r_max != -1);
        (*in_deriv)(in_r_max, c) += out_deriv_chunk(r, c);
      }
    }
  }
}

BackpropNeedsInput()

virtual bool BackpropNeedsInput ( ) const

inlinevirtual

Reimplemented from Component.

Definition at line 1156 of file nnet-component.h.

{ return true; }

BackpropNeedsOutput()

virtual bool BackpropNeedsOutput ( ) const

inlinevirtual

Reimplemented from Component.

Definition at line 1157 of file nnet-component.h.

References kaldi::cu::Copy(), and KALDI_DISALLOW_COPY_AND_ASSIGN.

{ return false; }

Context()

virtual std::vector<int32> Context ( ) const

inlinevirtual

Return a vector describing the temporal context this component requires for each frame of output, as a sorted list.

The default implementation returns a vector ( 0 ), but a splicing layer might return e.g. (-2, -1, 0, 1, 2), but it doesn't have to be contiguous. Note : The context needed by the entire network is a function of the contexts needed by all the components. It is required that Context().front() <= 0 and Context().back() >= 0.

Reimplemented from Component.

Definition at line 1143 of file nnet-component.h.

References Component::Propagate().

{ return context_; }

Copy()

Component * Copy ( ) const

virtual

Copy component (deep copy).

Implements Component.

Definition at line 2949 of file nnet-component.cc.

References SpliceMaxComponent::Init().

                                          {
  SpliceMaxComponent *ans = new SpliceMaxComponent();
  ans->Init(dim_, context_);
  return ans;
}

Info()

std::string Info ( ) const

virtual

Reimplemented from Component.

Definition at line 2805 of file nnet-component.cc.

References Component::Info().

                                         {
  std::stringstream stream;
  std::ostringstream os;
  std::copy(context_.begin(), context_.end(),
            std::ostream_iterator<int32>(os, " "));
  stream << Component::Info() << ", context=" << os.str();
  return stream.str();
}

Init()

void Init	(	int32	dim,
		std::vector< int32 >	context
	)

Definition at line 2814 of file nnet-component.cc.

References KALDI_ASSERT.

Referenced by kaldi::nnet2::BasicDebugTestForSpliceMax(), and SpliceMaxComponent::Copy().

                                                         {
  dim_ = dim;
  context_ = context;
  KALDI_ASSERT(dim_ > 0 && context_.front() <= 0 && context_.back() >= 0);
}

InitFromString()

void InitFromString ( std::string  args )

virtual

Initialize, typically from a line of a config file.

The "args" will contain any parameters that need to be passed to the Component, e.g. dimensions.

Implements Component.

Definition at line 2823 of file nnet-component.cc.

References rnnlm::i, AffineComponentPreconditionedOnline::Init(), KALDI_ASSERT, KALDI_ERR, kaldi::nnet2::ParseFromString(), and AffineComponentPreconditionedOnline::Type().

                                                      {
  std::string orig_args(args);
  int32 dim, left_context, right_context;
  std::vector <int32> context;
  bool dim_ok = ParseFromString("dim", &args, &dim);
  bool context_ok = ParseFromString("context", &args, &context);
  bool left_right_context_ok = ParseFromString("left-context",
                                               &args, &left_context) &&
                               ParseFromString("right-context", &args,
                                               &right_context);

  if (!(dim_ok && (context_ok || left_right_context_ok)) ||
      !args.empty() || dim <= 0)
    KALDI_ERR << "Invalid initializer for layer of type "
              << Type() << ": \"" << orig_args << "\"";
  if (left_right_context_ok)  {
    KALDI_ASSERT(context.size() == 0);
    for (int32 i = -1 * left_context; i <= right_context; i++)
      context.push_back(i);
  }
  Init(dim, context);
}

InputDim()

virtual int32 InputDim ( ) const

inlinevirtual

Get size of input vectors.

Implements Component.

Definition at line 1141 of file nnet-component.h.

{ return dim_; }

KALDI_DISALLOW_COPY_AND_ASSIGN()

KALDI_DISALLOW_COPY_AND_ASSIGN ( SpliceMaxComponent  )

private

OutputDim()

virtual int32 OutputDim ( ) const

inlinevirtual

Get size of output vectors.

Implements Component.

Definition at line 1142 of file nnet-component.h.

Referenced by kaldi::nnet2::BasicDebugTestForSpliceMax().

{ return dim_; }

Propagate()

void Propagate ( const ChunkInfo &  in_info, const ChunkInfo &  out_info, const CuMatrixBase< BaseFloat > &  in, CuMatrixBase< BaseFloat > *  out  ) const

virtual

Perform forward pass propagation Input->Output.

Each row is one frame or training example. Interpreted as "num_chunks" equally sized chunks of frames; this only matters for layers that do things like context splicing. Typically this variable will either be 1 (when we're processing a single contiguous chunk of data) or will be the same as in.NumFrames(), but other values are possible if some layers do splicing.

Implements Component.

Definition at line 2847 of file nnet-component.cc.

References ChunkInfo::Check(), ChunkInfo::CheckSize(), ChunkInfo::ChunkSize(), CuMatrixBase< Real >::CopyFromMat(), CuMatrixBase< Real >::CopyRows(), ChunkInfo::GetIndex(), ChunkInfo::GetOffset(), rnnlm::i, KALDI_ASSERT, CuMatrixBase< Real >::Max(), ChunkInfo::NumChunks(), and ChunkInfo::NumCols().

Referenced by kaldi::nnet2::BasicDebugTestForSpliceMax().

                                                                        {
  in_info.Check();
  out_info.Check();
  in_info.CheckSize(in);
  out_info.CheckSize(*out);
  KALDI_ASSERT(in_info.NumChunks() == out_info.NumChunks());
  int32 in_chunk_size  = in_info.ChunkSize(),
        out_chunk_size = out_info.ChunkSize(),
        dim = in_info.NumCols();

  CuMatrix<BaseFloat> input_chunk_part(out_chunk_size, dim);
  for (int32 chunk = 0; chunk < in_info.NumChunks(); chunk++) {
    CuSubMatrix<BaseFloat> input_chunk(in,
                                     chunk * in_chunk_size, in_chunk_size,
                                     0, dim),
                        output_chunk(*out,
                                     chunk * out_chunk_size,
                                     out_chunk_size, 0, dim);
    for (int32 offset = 0; offset < context_.size(); offset++) {
      // computing the indices to copy into input_chunk_part from input_chunk
      // copy the rows of the input matrix which correspond to the current
      // context index
      std::vector<int32> input_chunk_inds(out_chunk_size);
      for (int32 i = 0; i < out_chunk_size; i++) {
        int32 out_chunk_ind  = i;
        int32 out_chunk_offset =
            out_info.GetOffset(out_chunk_ind);
        input_chunk_inds[i] =
            in_info.GetIndex(out_chunk_offset + context_[offset]);
      }
      CuArray<int32> cu_chunk_inds(input_chunk_inds);
      input_chunk_part.CopyRows(input_chunk, cu_chunk_inds);
      if (offset == 0)  {
        output_chunk.CopyFromMat(input_chunk_part);
      } else {
        output_chunk.Max(input_chunk_part);
      }
    }
  }
}

Read()

void Read ( std::istream &  is, bool  binary  )

virtual

Implements Component.

Definition at line 2955 of file nnet-component.cc.

References kaldi::nnet2::ExpectOneOrTwoTokens(), kaldi::ExpectToken(), rnnlm::i, KALDI_ERR, kaldi::ReadBasicType(), kaldi::ReadIntegerVector(), and kaldi::ReadToken().

                                                         {
  ExpectOneOrTwoTokens(is, binary, "<SpliceMaxComponent>", "<Dim>");
  ReadBasicType(is, binary, &dim_);
  std::string token;
  ReadToken(is, false, &token);
  if (token == "<LeftContext>") {
    int32 left_context = 0, right_context = 0;
    std::vector<int32> context;
    ReadBasicType(is, binary, &left_context);
    ExpectToken(is, binary, "<RightContext>");
    ReadBasicType(is, binary, &right_context);
    for (int32 i = -1 * left_context; i <= right_context; i++)
      context.push_back(i);
    context_ = context;
  } else  if (token == "<Context>") {
    ReadIntegerVector(is, binary, &context_);
  } else  {
    KALDI_ERR << "Unknown token" << token << ", the model might be corrupted";
  }
  ExpectToken(is, binary, "</SpliceMaxComponent>");
}

Type()

virtual std::string Type ( ) const

inlinevirtual

Implements Component.

Definition at line 1138 of file nnet-component.h.

{ return "SpliceMaxComponent"; }

Write()

void Write ( std::ostream &  os, bool  binary  ) const

virtual

Write component to stream.

Implements Component.

Definition at line 2977 of file nnet-component.cc.

References kaldi::WriteBasicType(), kaldi::WriteIntegerVector(), and kaldi::WriteToken().

                                                                {
  WriteToken(os, binary, "<SpliceMaxComponent>");
  WriteToken(os, binary, "<Dim>");
  WriteBasicType(os, binary, dim_);
  WriteToken(os, binary, "<Context>");
  WriteIntegerVector(os, binary, context_);
  WriteToken(os, binary, "</SpliceMaxComponent>");
}

Member Data Documentation

context_

std::vector<int32> context_

private

Definition at line 1164 of file nnet-component.h.

dim_

int32 dim_

private

Definition at line 1163 of file nnet-component.h.

---

The documentation for this class was generated from the following files:

• nnet2/nnet-component.h
• nnet2/nnet-component.cc