MatrixRandomizer Class Reference

Shuffles rows of a matrix according to the indices in the mask,. More...

#include <nnet-randomizer.h>

Collaboration diagram for MatrixRandomizer:

Public Member Functions
	MatrixRandomizer ()
	MatrixRandomizer (const NnetDataRandomizerOptions &conf)
void	Init (const NnetDataRandomizerOptions &conf)
	Set the randomizer parameters (size) More...
void	AddData (const CuMatrixBase< BaseFloat > &m)
	Add data to randomization buffer. More...
bool	IsFull ()
	Returns true, when capacity is full. More...
int32	NumFrames ()
	Number of frames stored inside the Randomizer. More...
void	Randomize (const std::vector< int32 > &mask)
	Randomize matrix row-order using mask. More...
bool	Done ()
	Returns true, if no more data for another mini-batch (after current one) More...
void	Next ()
	Sets cursor to next mini-batch. More...
const CuMatrixBase< BaseFloat > &	Value ()
	Returns matrix-window with next mini-batch. More...

Private Attributes
CuMatrix< BaseFloat >	data_
CuMatrix< BaseFloat >	data_aux_
CuMatrix< BaseFloat >	minibatch_
int32	data_begin_
	A cursor, pointing to the 'row' where the next mini-batch begins,. More...
int32	data_end_
	A cursor, pointing to the 'row' after the end of data,. More...
NnetDataRandomizerOptions	conf_

Detailed Description

Shuffles rows of a matrix according to the indices in the mask,.

Definition at line 87 of file nnet-randomizer.h.

Constructor & Destructor Documentation

MatrixRandomizer() [1/2]

MatrixRandomizer ( )

inline

Definition at line 89 of file nnet-randomizer.h.

                    :
    data_begin_(0),
    data_end_(0)
  { }

MatrixRandomizer() [2/2]

MatrixRandomizer ( const NnetDataRandomizerOptions &  conf )

inlineexplicit

Definition at line 94 of file nnet-randomizer.h.

                                                                  :
    data_begin_(0),
    data_end_(0)
  {
    Init(conf);
  }

Member Function Documentation

AddData()

void AddData

(

const CuMatrixBase< BaseFloat > &

m

)

Add data to randomization buffer.

Definition at line 47 of file nnet-randomizer.cc.

References data_, KALDI_ASSERT, CuMatrixBase< Real >::NumCols(), MatrixBase< Real >::NumRows(), and CuMatrixBase< Real >::NumRows().

Referenced by main(), and UnitTestMatrixRandomizer().

                                                               {
  // pre-allocate before 1st use
  if (data_.NumCols() == 0) {
    data_.Resize(conf_.randomizer_size, m.NumCols());
  }
  // optionally put previous left-over to front
  if (data_begin_ > 0) {
    KALDI_ASSERT(data_begin_ <= data_end_);  // sanity check,
    int32 leftover = data_end_ - data_begin_;
    KALDI_ASSERT(leftover < data_begin_);  // no overlap,
    if (leftover > 0) {
      data_.RowRange(0, leftover).CopyFromMat(data_.RowRange(data_begin_, leftover));
    }
    data_begin_ = 0;
    data_end_ = leftover;
    // set zero to the rest of the buffer,
    data_.RowRange(leftover, data_.NumRows() - leftover).SetZero();
  }
  // extend the buffer if necessary,
  if (data_.NumRows() < data_end_ + m.NumRows()) {
    // CuMatrix -> Matrix -> CuMatrix (needs less GPU memory),
    Matrix<BaseFloat> data_aux(data_);
    // Add extra 3% rows, so we don't reallocate soon:
    int32 extra_rows = 0.03 * data_.NumRows();
    data_.Resize(data_end_ + m.NumRows() + extra_rows, data_.NumCols());
    data_.RowRange(0, data_aux.NumRows()).CopyFromMat(data_aux);
  }
  // copy the data
  data_.RowRange(data_end_, m.NumRows()).CopyFromMat(m);
  data_end_ += m.NumRows();
}

Done()

bool Done ( )

inline

Returns true, if no more data for another mini-batch (after current one)

Definition at line 123 of file nnet-randomizer.h.

Referenced by main(), and UnitTestMatrixRandomizer().

              {
    return (data_end_ - data_begin_ < conf_.minibatch_size);
  }

Init()

void Init ( const NnetDataRandomizerOptions &  conf )

inline

Set the randomizer parameters (size)

Definition at line 102 of file nnet-randomizer.h.

Referenced by UnitTestMatrixRandomizer().

                                                   {
    conf_ = conf;
  }

IsFull()

bool IsFull ( )

inline

Returns true, when capacity is full.

Definition at line 110 of file nnet-randomizer.h.

Referenced by main(), and UnitTestMatrixRandomizer().

                {
    return ((data_begin_ == 0) && (data_end_ > conf_.randomizer_size ));
  }

Next()

void Next

(

)

Sets cursor to next mini-batch.

Definition at line 98 of file nnet-randomizer.cc.

Referenced by main(), and UnitTestMatrixRandomizer().

                            {
  data_begin_ += conf_.minibatch_size;
}

NumFrames()

int32 NumFrames ( )

inline

Number of frames stored inside the Randomizer.

Definition at line 115 of file nnet-randomizer.h.

References kaldi::cu::Randomize().

Referenced by main(), and UnitTestMatrixRandomizer().

                    {
    return data_end_;
  }

Randomize()

void Randomize

(

const std::vector< int32 > &

mask

)

Randomize matrix row-order using mask.

Definition at line 79 of file nnet-randomizer.cc.

References CuArray< T >::CopyFromVec(), data_, KALDI_ASSERT, and kaldi::cu::Randomize().

Referenced by main(), and UnitTestMatrixRandomizer().

                                                             {
  KALDI_ASSERT(data_begin_ == 0);
  KALDI_ASSERT(data_end_ > 0);
  KALDI_ASSERT(data_end_ == mask.size());
  // Copy to auxiliary buffer for unshuffled data
  data_aux_ = data_;
  // Put the mask to GPU
  CuArray<int32> mask_in_gpu(mask.size());
  mask_in_gpu.CopyFromVec(mask);
  // Randomize the data, mask is used to index rows in source matrix:
  // (Here the vector 'mask_in_gpu' is typically shorter than number
  //  of rows in 'data_aux_', because the buffer 'data_aux_'
  //  is larger than capacity 'randomizer_size'.
  //  The extra rows in 'data_aux_' do not contain speech frames and
  //  are not copied from 'data_aux_', the extra rows in 'data_' are
  //  unchanged by cu::Randomize.)
  cu::Randomize(data_aux_, mask_in_gpu, &data_);
}

Value()

const CuMatrixBase< BaseFloat > & Value

(

)

Returns matrix-window with next mini-batch.

Definition at line 102 of file nnet-randomizer.cc.

References data_, KALDI_ASSERT, and kaldi::kUndefined.

Referenced by main(), and UnitTestMatrixRandomizer().

                                                       {
  // make sure we have data for next minibatch,
  KALDI_ASSERT(data_end_ - data_begin_ >= conf_.minibatch_size);
  // prepare the mini-batch buffer,
  minibatch_.Resize(conf_.minibatch_size, data_.NumCols(), kUndefined);
  minibatch_.CopyFromMat(data_.RowRange(data_begin_, conf_.minibatch_size));
  return minibatch_;
}

Member Data Documentation

conf_

NnetDataRandomizerOptions conf_

private

Definition at line 143 of file nnet-randomizer.h.

data_

CuMatrix<BaseFloat> data_

private

Definition at line 134 of file nnet-randomizer.h.

data_aux_

CuMatrix<BaseFloat> data_aux_

private

Definition at line 135 of file nnet-randomizer.h.

data_begin_

int32 data_begin_

private

A cursor, pointing to the 'row' where the next mini-batch begins,.

Definition at line 139 of file nnet-randomizer.h.

data_end_

int32 data_end_

private

A cursor, pointing to the 'row' after the end of data,.

Definition at line 141 of file nnet-randomizer.h.

minibatch_

CuMatrix<BaseFloat> minibatch_

private

Definition at line 136 of file nnet-randomizer.h.

---

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

• nnet/nnet-randomizer.h
• nnet/nnet-randomizer.cc