pmlpp/lin_alg/mlpp_tensor3.cpp

/*************************************************************************/
/*  mlpp_tensor3.cpp                                                     */
/*************************************************************************/
/*                         This file is part of:                         */
/*                    PMLPP Machine Learning Library                     */
/*                   https://github.com/Relintai/pmlpp                   */
/*************************************************************************/
/* Copyright (c) 2023-present Péter Magyar.                              */
/* Copyright (c) 2022-2023 Marc Melikyan                                 */
/*                                                                       */
/* Permission is hereby granted, free of charge, to any person obtaining */
/* a copy of this software and associated documentation files (the       */
/* "Software"), to deal in the Software without restriction, including   */
/* without limitation the rights to use, copy, modify, merge, publish,   */
/* distribute, sublicense, and/or sell copies of the Software, and to    */
/* permit persons to whom the Software is furnished to do so, subject to */
/* the following conditions:                                             */
/*                                                                       */
/* The above copyright notice and this permission notice shall be        */
/* included in all copies or substantial portions of the Software.       */
/*                                                                       */
/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,       */
/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF    */
/* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.*/
/* IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY  */
/* CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,  */
/* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE     */
/* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.                */
/*************************************************************************/

#include "mlpp_tensor3.h"

#ifdef USING_SFW
#include "sfw.h"
#else
#include "core/io/image.h"
#endif

Array MLPPTensor3::get_data() {
  PoolRealArray pl;

  int ds = data_size();

  if (ds) {
    pl.resize(ds);
    PoolRealArray::Write w = pl.write();
    real_t *dest = w.ptr();

    for (int i = 0; i < ds; ++i) {
      dest[i] = _data[i];
    }
  }

  Array arr;
  arr.push_back(size());
  arr.push_back(pl);

  return arr;
}
void MLPPTensor3::set_data(const Array &p_from) {
  if (p_from.size() != 2) {
    return;
  }

  Size3i s = p_from[0];
  PoolRealArray pl = p_from[1];

  int ds = s.x * s.y * s.z;

  if (ds != pl.size()) {
    return;
  }

  if (_size != s) {
    resize(s);
  }

  PoolRealArray::Read r = pl.read();
  for (int i = 0; i < ds; ++i) {
    _data[i] = r[i];
  }
}

void MLPPTensor3::z_slice_add(const Vector<real_t> &p_row) {
  if (p_row.size() == 0) {
    return;
  }

  int fms = z_slice_data_size();

  ERR_FAIL_COND(fms != p_row.size());

  int ci = data_size();

  ++_size.z;

  _data = (real_t *)memrealloc(_data, data_size() * sizeof(real_t));
  CRASH_COND_MSG(!_data, "Out of memory");

  const real_t *row_arr = p_row.ptr();

  for (int i = 0; i < p_row.size(); ++i) {
    _data[ci + i] = row_arr[i];
  }
}

void MLPPTensor3::z_slice_add_pool_vector(const PoolRealArray &p_row) {
  if (p_row.size() == 0) {
    return;
  }

  int fms = z_slice_data_size();

  ERR_FAIL_COND(fms != p_row.size());

  int ci = data_size();

  ++_size.z;

  _data = (real_t *)memrealloc(_data, data_size() * sizeof(real_t));
  CRASH_COND_MSG(!_data, "Out of memory");

  PoolRealArray::Read rread = p_row.read();
  const real_t *row_arr = rread.ptr();

  for (int i = 0; i < p_row.size(); ++i) {
    _data[ci + i] = row_arr[i];
  }
}

void MLPPTensor3::z_slice_add_mlpp_vector(const Ref<MLPPVector> &p_row) {
  ERR_FAIL_COND(!p_row.is_valid());

  int p_row_size = p_row->size();

  if (p_row_size == 0) {
    return;
  }

  int fms = z_slice_data_size();

  ERR_FAIL_COND(fms != p_row_size);

  int ci = data_size();

  ++_size.z;

  _data = (real_t *)memrealloc(_data, data_size() * sizeof(real_t));
  CRASH_COND_MSG(!_data, "Out of memory");

  const real_t *row_ptr = p_row->ptr();

  for (int i = 0; i < p_row_size; ++i) {
    _data[ci + i] = row_ptr[i];
  }
}

void MLPPTensor3::z_slice_add_mlpp_matrix(const Ref<MLPPMatrix> &p_matrix) {
  ERR_FAIL_COND(!p_matrix.is_valid());

  int other_data_size = p_matrix->data_size();

  if (other_data_size == 0) {
    return;
  }

  Size2i matrix_size = p_matrix->size();
  Size2i fms = z_slice_size();

  ERR_FAIL_COND(fms != matrix_size);

  int start_offset = data_size();

  ++_size.z;

  _data = (real_t *)memrealloc(_data, data_size() * sizeof(real_t));
  CRASH_COND_MSG(!_data, "Out of memory");

  const real_t *other_ptr = p_matrix->ptr();

  for (int i = 0; i < other_data_size; ++i) {
    _data[start_offset + i] = other_ptr[i];
  }
}

void MLPPTensor3::z_slice_remove(int p_index) {
  ERR_FAIL_INDEX(p_index, _size.z);

  --_size.z;

  int ds = data_size();

  if (ds == 0) {
    memfree(_data);
    _data = NULL;
    return;
  }

  int fmds = z_slice_data_size();

  for (int i = calculate_z_slice_index(p_index); i < ds; ++i) {
    _data[i] = _data[i + fmds];
  }

  _data = (real_t *)memrealloc(_data, data_size() * sizeof(real_t));
  CRASH_COND_MSG(!_data, "Out of memory");
}

// Removes the item copying the last value into the position of the one to
// remove. It's generally faster than `remove`.
void MLPPTensor3::z_slice_remove_unordered(int p_index) {
  ERR_FAIL_INDEX(p_index, _size.z);

  --_size.z;

  int ds = data_size();

  if (ds == 0) {
    memfree(_data);
    _data = NULL;
    return;
  }

  int start_ind = calculate_z_slice_index(p_index);
  int end_ind = calculate_z_slice_index(p_index + 1);

  for (int i = start_ind; i < end_ind; ++i) {
    _data[i] = _data[ds + i];
  }

  _data = (real_t *)memrealloc(_data, data_size() * sizeof(real_t));
  CRASH_COND_MSG(!_data, "Out of memory");
}

void MLPPTensor3::z_slice_swap(int p_index_1, int p_index_2) {
  ERR_FAIL_INDEX(p_index_1, _size.z);
  ERR_FAIL_INDEX(p_index_2, _size.z);

  int ind1_start = calculate_z_slice_index(p_index_1);
  int ind2_start = calculate_z_slice_index(p_index_2);

  int fmds = z_slice_data_size();

  for (int i = 0; i < fmds; ++i) {
    SWAP(_data[ind1_start + i], _data[ind2_start + i]);
  }
}

void MLPPTensor3::resize(const Size3i &p_size) {
  _size = p_size;

  int ds = data_size();

  if (ds == 0) {
    if (_data) {
      memfree(_data);
      _data = NULL;
    }

    return;
  }

  _data = (real_t *)memrealloc(_data, ds * sizeof(real_t));
  CRASH_COND_MSG(!_data, "Out of memory");
}

void MLPPTensor3::shape_set(const Size3i &p_size) {
  int ds = data_size();
  int new_data_size = p_size.x * p_size.y * p_size.z;

  ERR_FAIL_COND_MSG(ds != new_data_size,
                    "The new size has a different volume than the old. If this "
                    "is intended use resize()!");

  _size = p_size;
}

Vector<real_t> MLPPTensor3::row_get_vector(int p_index_z, int p_index_y) const {
  ERR_FAIL_INDEX_V(p_index_y, _size.y, Vector<real_t>());
  ERR_FAIL_INDEX_V(p_index_z, _size.z, Vector<real_t>());

  Vector<real_t> ret;

  if (unlikely(_size.x == 0)) {
    return ret;
  }

  ret.resize(_size.x);

  int ind_start = p_index_y * _size.x;

  real_t *row_ptr = ret.ptrw();

  for (int i = 0; i < _size.x; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }

  return ret;
}

PoolRealArray MLPPTensor3::row_get_pool_vector(int p_index_z,
                                               int p_index_y) const {
  ERR_FAIL_INDEX_V(p_index_y, _size.y, PoolRealArray());
  ERR_FAIL_INDEX_V(p_index_z, _size.z, PoolRealArray());

  PoolRealArray ret;

  if (unlikely(_size.x == 0)) {
    return ret;
  }

  ret.resize(_size.x);

  int ind_start = p_index_y * _size.x + _size.x * _size.y * p_index_z;

  PoolRealArray::Write w = ret.write();
  real_t *row_ptr = w.ptr();

  for (int i = 0; i < _size.x; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }

  return ret;
}

Ref<MLPPVector> MLPPTensor3::row_get_mlpp_vector(int p_index_z,
                                                 int p_index_y) const {
  ERR_FAIL_INDEX_V(p_index_y, _size.y, Ref<MLPPVector>());
  ERR_FAIL_INDEX_V(p_index_z, _size.z, Ref<MLPPVector>());

  Ref<MLPPVector> ret;
  ret.instance();

  if (unlikely(_size.x == 0)) {
    return ret;
  }

  ret->resize(_size.x);

  int ind_start = p_index_y * _size.x + _size.x * _size.y * p_index_z;

  real_t *row_ptr = ret->ptrw();

  for (int i = 0; i < _size.x; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }

  return ret;
}

void MLPPTensor3::row_get_into_mlpp_vector(int p_index_z, int p_index_y,
                                           Ref<MLPPVector> target) const {
  ERR_FAIL_COND(!target.is_valid());
  ERR_FAIL_INDEX(p_index_y, _size.y);
  ERR_FAIL_INDEX(p_index_z, _size.z);

  if (unlikely(target->size() != _size.x)) {
    target->resize(_size.x);
  }

  int ind_start = p_index_y * _size.x + _size.x * _size.y * p_index_z;

  real_t *row_ptr = target->ptrw();

  for (int i = 0; i < _size.x; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }
}

void MLPPTensor3::row_set_vector(int p_index_z, int p_index_y,
                                 const Vector<real_t> &p_row) {
  ERR_FAIL_COND(p_row.size() != _size.x);
  ERR_FAIL_INDEX(p_index_y, _size.y);
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int ind_start = p_index_y * _size.x + _size.x * _size.y * p_index_z;

  const real_t *row_ptr = p_row.ptr();

  for (int i = 0; i < _size.x; ++i) {
    _data[ind_start + i] = row_ptr[i];
  }
}

void MLPPTensor3::row_set_pool_vector(int p_index_z, int p_index_y,
                                      const PoolRealArray &p_row) {
  ERR_FAIL_COND(p_row.size() != _size.x);
  ERR_FAIL_INDEX(p_index_y, _size.y);
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int ind_start = p_index_y * _size.x + _size.x * _size.y * p_index_z;

  PoolRealArray::Read r = p_row.read();
  const real_t *row_ptr = r.ptr();

  for (int i = 0; i < _size.x; ++i) {
    _data[ind_start + i] = row_ptr[i];
  }
}

void MLPPTensor3::row_set_mlpp_vector(int p_index_z, int p_index_y,
                                      const Ref<MLPPVector> &p_row) {
  ERR_FAIL_COND(!p_row.is_valid());
  ERR_FAIL_COND(p_row->size() != _size.x);
  ERR_FAIL_INDEX(p_index_y, _size.y);
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int ind_start = p_index_y * _size.x + _size.x * _size.y * p_index_z;

  const real_t *row_ptr = p_row->ptr();

  for (int i = 0; i < _size.x; ++i) {
    _data[ind_start + i] = row_ptr[i];
  }
}

Vector<real_t> MLPPTensor3::z_slice_get_vector(int p_index_z) const {
  ERR_FAIL_INDEX_V(p_index_z, _size.z, Vector<real_t>());

  Vector<real_t> ret;

  int fmds = z_slice_data_size();

  if (unlikely(fmds == 0)) {
    return ret;
  }

  ret.resize(fmds);

  int ind_start = calculate_z_slice_index(p_index_z);

  real_t *row_ptr = ret.ptrw();

  for (int i = 0; i < fmds; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }

  return ret;
}

PoolRealArray MLPPTensor3::z_slice_get_pool_vector(int p_index_z) const {
  ERR_FAIL_INDEX_V(p_index_z, _size.z, PoolRealArray());

  PoolRealArray ret;

  int fmds = z_slice_data_size();

  if (unlikely(fmds == 0)) {
    return ret;
  }

  ret.resize(fmds);

  int ind_start = calculate_z_slice_index(p_index_z);

  PoolRealArray::Write w = ret.write();
  real_t *row_ptr = w.ptr();

  for (int i = 0; i < fmds; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }

  return ret;
}

Ref<MLPPVector> MLPPTensor3::z_slice_get_mlpp_vector(int p_index_z) const {
  ERR_FAIL_INDEX_V(p_index_z, _size.z, Ref<MLPPVector>());

  Ref<MLPPVector> ret;
  ret.instance();

  int fmds = z_slice_data_size();

  if (unlikely(fmds == 0)) {
    return ret;
  }

  ret->resize(fmds);

  int ind_start = calculate_z_slice_index(p_index_z);

  real_t *row_ptr = ret->ptrw();

  for (int i = 0; i < fmds; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }

  return ret;
}

void MLPPTensor3::z_slice_get_into_mlpp_vector(int p_index_z,
                                               Ref<MLPPVector> target) const {
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int fmds = z_slice_data_size();

  if (unlikely(target->size() != fmds)) {
    target->resize(fmds);
  }

  int ind_start = calculate_z_slice_index(p_index_z);

  real_t *row_ptr = target->ptrw();

  for (int i = 0; i < fmds; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }
}

Ref<MLPPMatrix> MLPPTensor3::z_slice_get_mlpp_matrix(int p_index_z) const {
  ERR_FAIL_INDEX_V(p_index_z, _size.z, Ref<MLPPMatrix>());

  Ref<MLPPMatrix> ret;
  ret.instance();

  int fmds = z_slice_data_size();

  if (unlikely(fmds == 0)) {
    return ret;
  }

  ret->resize(z_slice_size());

  int ind_start = calculate_z_slice_index(p_index_z);

  real_t *row_ptr = ret->ptrw();

  for (int i = 0; i < fmds; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }

  return ret;
}

void MLPPTensor3::z_slice_get_into_mlpp_matrix(int p_index_z,
                                               Ref<MLPPMatrix> target) const {
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int fmds = z_slice_data_size();
  Size2i fms = z_slice_size();

  if (unlikely(target->size() != fms)) {
    target->resize(fms);
  }

  int ind_start = calculate_z_slice_index(p_index_z);

  real_t *row_ptr = target->ptrw();

  for (int i = 0; i < fmds; ++i) {
    row_ptr[i] = _data[ind_start + i];
  }
}

void MLPPTensor3::z_slice_set_vector(int p_index_z,
                                     const Vector<real_t> &p_row) {
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int fmds = z_slice_data_size();

  ERR_FAIL_COND(p_row.size() != fmds);

  int ind_start = calculate_z_slice_index(p_index_z);

  const real_t *row_ptr = p_row.ptr();

  for (int i = 0; i < fmds; ++i) {
    _data[ind_start + i] = row_ptr[i];
  }
}

void MLPPTensor3::z_slice_set_pool_vector(int p_index_z,
                                          const PoolRealArray &p_row) {
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int fmds = z_slice_data_size();

  ERR_FAIL_COND(p_row.size() != fmds);

  int ind_start = calculate_z_slice_index(p_index_z);

  PoolRealArray::Read r = p_row.read();
  const real_t *row_ptr = r.ptr();

  for (int i = 0; i < fmds; ++i) {
    _data[ind_start + i] = row_ptr[i];
  }
}

void MLPPTensor3::z_slice_set_mlpp_vector(int p_index_z,
                                          const Ref<MLPPVector> &p_row) {
  ERR_FAIL_INDEX(p_index_z, _size.z);
  ERR_FAIL_COND(!p_row.is_valid());

  int fmds = z_slice_data_size();

  ERR_FAIL_COND(p_row->size() != fmds);

  int ind_start = calculate_z_slice_index(p_index_z);

  const real_t *row_ptr = p_row->ptr();

  for (int i = 0; i < fmds; ++i) {
    _data[ind_start + i] = row_ptr[i];
  }
}

void MLPPTensor3::z_slice_set_mlpp_matrix(int p_index_z,
                                          const Ref<MLPPMatrix> &p_mat) {
  ERR_FAIL_INDEX(p_index_z, _size.z);
  ERR_FAIL_COND(!p_mat.is_valid());

  int fmds = z_slice_data_size();

  ERR_FAIL_COND(p_mat->size() != z_slice_size());

  int ind_start = calculate_z_slice_index(p_index_z);

  const real_t *row_ptr = p_mat->ptr();

  for (int i = 0; i < fmds; ++i) {
    _data[ind_start + i] = row_ptr[i];
  }
}

void MLPPTensor3::x_slice_get_into(int p_index_x,
                                   Ref<MLPPMatrix> target) const {
  ERR_FAIL_INDEX(p_index_x, _size.x);
  ERR_FAIL_COND(!target.is_valid());

  if (unlikely(target->size() != Size2i(_size.y, _size.z))) {
    target->resize(Size2i(_size.y, _size.z));
  }

  for (int z = 0; z < _size.z; ++z) {
    for (int y = 0; y < _size.y; ++y) {
      target->element_set(z, y, element_get(z, p_index_x, y));
    }
  }
}
Ref<MLPPMatrix> MLPPTensor3::x_slice_get(int p_index_x) const {
  ERR_FAIL_INDEX_V(p_index_x, _size.x, Ref<MLPPMatrix>());

  Ref<MLPPMatrix> m;
  m.instance();

  x_slice_get_into(p_index_x, m);

  return m;
}
void MLPPTensor3::x_slice_set(int p_index_x, const Ref<MLPPMatrix> &p_mat) {
  ERR_FAIL_INDEX(p_index_x, _size.x);
  ERR_FAIL_COND(!p_mat.is_valid());
  ERR_FAIL_COND(p_mat->size() != Size2i(_size.y, _size.z));

  for (int z = 0; z < _size.z; ++z) {
    for (int y = 0; y < _size.y; ++y) {
      element_set(z, p_index_x, y, p_mat->element_get(z, y));
    }
  }
}

void MLPPTensor3::y_slice_get_into(int p_index_y,
                                   Ref<MLPPMatrix> target) const {
  ERR_FAIL_INDEX(p_index_y, _size.y);
  ERR_FAIL_COND(!target.is_valid());

  if (unlikely(target->size() != Size2i(_size.y, _size.z))) {
    target->resize(Size2i(_size.x, _size.z));
  }

  for (int z = 0; z < _size.z; ++z) {
    for (int x = 0; x < _size.x; ++x) {
      target->element_set(z, x, element_get(z, x, p_index_y));
    }
  }
}
Ref<MLPPMatrix> MLPPTensor3::y_slice_get(int p_index_y) const {
  ERR_FAIL_INDEX_V(p_index_y, _size.y, Ref<MLPPMatrix>());

  Ref<MLPPMatrix> m;
  m.instance();

  y_slice_get_into(p_index_y, m);

  return m;
}
void MLPPTensor3::y_slice_set(int p_index_y, const Ref<MLPPMatrix> &p_mat) {
  ERR_FAIL_INDEX(p_index_y, _size.y);
  ERR_FAIL_COND(!p_mat.is_valid());
  ERR_FAIL_COND(p_mat->size() != Size2i(_size.y, _size.z));

  for (int z = 0; z < _size.z; ++z) {
    for (int x = 0; x < _size.x; ++x) {
      element_set(z, x, p_index_y, p_mat->element_get(z, x));
    }
  }
}

void MLPPTensor3::z_slices_add_image(const Ref<Image> &p_img,
                                     const int p_channels) {
  ERR_FAIL_COND(!p_img.is_valid());

  Size2i img_size = Size2i(p_img->get_width(), p_img->get_height());

  int channel_count = 0;
  int channels[4];

  if (p_channels & IMAGE_CHANNEL_FLAG_R) {
    channels[channel_count] = 0;
    ++channel_count;
  }

  if (p_channels & IMAGE_CHANNEL_FLAG_G) {
    channels[channel_count] = 1;
    ++channel_count;
  }

  if (p_channels & IMAGE_CHANNEL_FLAG_B) {
    channels[channel_count] = 2;
    ++channel_count;
  }

  if (p_channels & IMAGE_CHANNEL_FLAG_A) {
    channels[channel_count] = 3;
    ++channel_count;
  }

  ERR_FAIL_COND(channel_count == 0);

  if (unlikely(_size == Size3i())) {
    resize(Size3i(img_size.x, img_size.y, channel_count));
  }

  Size2i fms = z_slice_size();

  ERR_FAIL_COND(img_size != fms);

  int start_channel = _size.y;

  _size.y += channel_count;

  resize(_size);

  Ref<Image> img = p_img;

  img->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int x = 0; x < fms.x; ++x) {
      Color c = img->get_pixel(x, y);

      for (int i = 0; i < channel_count; ++i) {
        element_set(start_channel + i, y, x, c[channels[i]]);
      }
    }
  }

  img->unlock();
}

Ref<Image> MLPPTensor3::z_slice_get_image(const int p_index_z) const {
  ERR_FAIL_INDEX_V(p_index_z, _size.z, Ref<Image>());

  Ref<Image> image;
  image.instance();

  if (data_size() == 0) {
    return image;
  }

  PoolByteArray arr;

  int fmsi = calculate_z_slice_index(p_index_z);
  int fms = z_slice_data_size();

  arr.resize(fms);

  PoolByteArray::Write w = arr.write();
  uint8_t *wptr = w.ptr();

  for (int i = 0; i < fms; ++i) {
    wptr[i] = static_cast<uint8_t>(_data[fmsi + i] * 255.0);
  }

  image->create(_size.x, _size.y, false, Image::FORMAT_L8, arr);

  return image;
}
Ref<Image> MLPPTensor3::z_slices_get_image(const int p_index_r,
                                           const int p_index_g,
                                           const int p_index_b,
                                           const int p_index_a) const {
  if (p_index_r != -1) {
    ERR_FAIL_INDEX_V(p_index_r, _size.z, Ref<Image>());
  }

  if (p_index_g != -1) {
    ERR_FAIL_INDEX_V(p_index_g, _size.z, Ref<Image>());
  }

  if (p_index_b != -1) {
    ERR_FAIL_INDEX_V(p_index_b, _size.z, Ref<Image>());
  }

  if (p_index_a != -1) {
    ERR_FAIL_INDEX_V(p_index_a, _size.z, Ref<Image>());
  }

  Ref<Image> image;
  image.instance();

  if (data_size() == 0) {
    return image;
  }

  Size2i fms = z_slice_size();

  image->create(_size.x, _size.y, false, Image::FORMAT_RGBA8);

  image->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int x = 0; x < fms.x; ++x) {
      Color c;

      if (p_index_r != -1) {
        c.r = element_get(p_index_r, y, x);
      }

      if (p_index_g != -1) {
        c.g = element_get(p_index_g, y, x);
      }

      if (p_index_b != -1) {
        c.b = element_get(p_index_b, y, x);
      }

      if (p_index_a != -1) {
        c.a = element_get(p_index_a, y, x);
      }

      image->set_pixel(x, y, c);
    }
  }

  image->unlock();

  return image;
}

void MLPPTensor3::z_slice_get_into_image(Ref<Image> p_target,
                                         const int p_index_z,
                                         const int p_target_channels) const {
  ERR_FAIL_INDEX(p_index_z, _size.z);
  ERR_FAIL_COND(!p_target.is_valid());

  int channel_count = 0;
  int channels[4];

  if (p_target_channels & IMAGE_CHANNEL_FLAG_R) {
    channels[channel_count] = 0;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_G) {
    channels[channel_count] = 1;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_B) {
    channels[channel_count] = 2;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_A) {
    channels[channel_count] = 3;
    ++channel_count;
  }

  ERR_FAIL_COND(channel_count == 0);

  if (data_size() == 0) {
    p_target->clear();
    return;
  }

  Size2i img_size = Size2i(p_target->get_width(), p_target->get_height());
  Size2i fms = z_slice_size();
  if (img_size != fms) {
    bool mip_maps = p_target->has_mipmaps();
    p_target->resize(fms.x, fms.y, Image::INTERPOLATE_NEAREST);

    if (p_target->has_mipmaps() != mip_maps) {
      if (mip_maps) {
        p_target->generate_mipmaps();
      } else {
        p_target->clear_mipmaps();
      }
    }
  }

  p_target->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int x = 0; x < fms.x; ++x) {
      Color c;

      float e = element_get(p_index_z, y, x);

      for (int i = 0; i < channel_count; ++i) {
        c[channels[i]] = e;
      }

      p_target->set_pixel(x, y, c);
    }
  }

  p_target->unlock();
}
void MLPPTensor3::z_slices_get_into_image(Ref<Image> p_target,
                                          const int p_index_r,
                                          const int p_index_g,
                                          const int p_index_b,
                                          const int p_index_a) const {
  ERR_FAIL_COND(!p_target.is_valid());

  if (p_index_r != -1) {
    ERR_FAIL_INDEX(p_index_r, _size.z);
  }

  if (p_index_g != -1) {
    ERR_FAIL_INDEX(p_index_g, _size.z);
  }

  if (p_index_b != -1) {
    ERR_FAIL_INDEX(p_index_b, _size.z);
  }

  if (p_index_a != -1) {
    ERR_FAIL_INDEX(p_index_a, _size.z);
  }

  if (data_size() == 0) {
    p_target->clear();
    return;
  }

  Size2i img_size = Size2i(p_target->get_width(), p_target->get_height());
  Size2i fms = z_slice_size();
  if (img_size != fms) {
    bool mip_maps = p_target->has_mipmaps();
    p_target->resize(fms.x, fms.y, Image::INTERPOLATE_NEAREST);

    if (p_target->has_mipmaps() != mip_maps) {
      if (mip_maps) {
        p_target->generate_mipmaps();
      } else {
        p_target->clear_mipmaps();
      }
    }
  }

  p_target->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int x = 0; x < fms.x; ++x) {
      Color c;

      if (p_index_r != -1) {
        c.r = element_get(p_index_r, y, x);
      }

      if (p_index_g != -1) {
        c.g = element_get(p_index_g, y, x);
      }

      if (p_index_b != -1) {
        c.b = element_get(p_index_b, y, x);
      }

      if (p_index_a != -1) {
        c.a = element_get(p_index_a, y, x);
      }

      p_target->set_pixel(x, y, c);
    }
  }

  p_target->unlock();
}

void MLPPTensor3::z_slice_set_image(const Ref<Image> &p_img,
                                    const int p_index_z,
                                    const int p_image_channel_flag) {
  ERR_FAIL_COND(!p_img.is_valid());
  ERR_FAIL_INDEX(p_index_z, _size.z);

  int channel_index = -1;

  for (int i = 0; i < 4; ++i) {
    if (((p_image_channel_flag & (1 << i)) != 0)) {
      channel_index = i;
      break;
    }
  }

  ERR_FAIL_INDEX(channel_index, 4);

  Size2i img_size = Size2i(p_img->get_width(), p_img->get_height());
  Size2i fms = z_slice_size();

  ERR_FAIL_COND(img_size != fms);

  Ref<Image> img = p_img;

  img->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int x = 0; x < fms.x; ++x) {
      Color c = img->get_pixel(x, y);

      element_set(p_index_z, y, x, c[channel_index]);
    }
  }

  img->unlock();
}
void MLPPTensor3::z_slices_set_image(const Ref<Image> &p_img,
                                     const int p_index_r, const int p_index_g,
                                     const int p_index_b, const int p_index_a) {
  ERR_FAIL_COND(!p_img.is_valid());

  if (p_index_r != -1) {
    ERR_FAIL_INDEX(p_index_r, _size.z);
  }

  if (p_index_g != -1) {
    ERR_FAIL_INDEX(p_index_g, _size.z);
  }

  if (p_index_b != -1) {
    ERR_FAIL_INDEX(p_index_b, _size.z);
  }

  if (p_index_a != -1) {
    ERR_FAIL_INDEX(p_index_a, _size.z);
  }

  Size2i img_size = Size2i(p_img->get_width(), p_img->get_height());
  Size2i fms = z_slice_size();

  ERR_FAIL_COND(img_size != fms);

  Ref<Image> img = p_img;

  img->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int x = 0; x < fms.x; ++x) {
      Color c = img->get_pixel(x, y);

      if (p_index_r != -1) {
        element_set(p_index_r, y, x, c.r);
      }

      if (p_index_g != -1) {
        element_set(p_index_g, y, x, c.g);
      }

      if (p_index_b != -1) {
        element_set(p_index_b, y, x, c.b);
      }

      if (p_index_a != -1) {
        element_set(p_index_a, y, x, c.a);
      }
    }
  }

  img->unlock();
}

void MLPPTensor3::set_from_image(const Ref<Image> &p_img,
                                 const int p_channels) {
  ERR_FAIL_COND(!p_img.is_valid());

  int channel_count = 0;
  int channels[4];

  if (p_channels & IMAGE_CHANNEL_FLAG_R) {
    channels[channel_count] = 0;
    ++channel_count;
  }

  if (p_channels & IMAGE_CHANNEL_FLAG_G) {
    channels[channel_count] = 1;
    ++channel_count;
  }

  if (p_channels & IMAGE_CHANNEL_FLAG_B) {
    channels[channel_count] = 2;
    ++channel_count;
  }

  if (p_channels & IMAGE_CHANNEL_FLAG_A) {
    channels[channel_count] = 3;
    ++channel_count;
  }

  ERR_FAIL_COND(channel_count == 0);

  Size2i img_size = Size2i(p_img->get_width(), p_img->get_height());

  resize(Size3i(img_size.x, img_size.y, channel_count));

  Size2i fms = z_slice_size();

  Ref<Image> img = p_img;

  img->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int x = 0; x < fms.x; ++x) {
      Color c = img->get_pixel(x, y);

      for (int i = 0; i < channel_count; ++i) {
        element_set(i, y, x, c[channels[i]]);
      }
    }
  }

  img->unlock();
}

Ref<Image> MLPPTensor3::x_slice_get_image(const int p_index_x) const {
  ERR_FAIL_INDEX_V(p_index_x, _size.x, Ref<Image>());

  Ref<Image> image;
  image.instance();

  if (data_size() == 0) {
    return image;
  }

  PoolByteArray arr;
  arr.resize(_size.y * _size.z);

  PoolByteArray::Write w = arr.write();
  uint8_t *wptr = w.ptr();
  int i = 0;

  for (int z = 0; z < _size.z; ++z) {
    for (int y = 0; y < _size.y; ++y) {
      wptr[i] = static_cast<uint8_t>(element_get(z, p_index_x, y) * 255.0);

      ++i;
    }
  }

  image->create(_size.y, _size.z, false, Image::FORMAT_L8, arr);

  return image;
}
void MLPPTensor3::x_slice_get_into_image(Ref<Image> p_target,
                                         const int p_index_x,
                                         const int p_target_channels) const {
  ERR_FAIL_INDEX(p_index_x, _size.x);
  ERR_FAIL_COND(!p_target.is_valid());

  int channel_count = 0;
  int channels[4];

  if (p_target_channels & IMAGE_CHANNEL_FLAG_R) {
    channels[channel_count] = 0;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_G) {
    channels[channel_count] = 1;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_B) {
    channels[channel_count] = 2;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_A) {
    channels[channel_count] = 3;
    ++channel_count;
  }

  ERR_FAIL_COND(channel_count == 0);

  if (data_size() == 0) {
    p_target->clear();
    return;
  }

  Size2i img_size = Size2i(p_target->get_width(), p_target->get_height());
  Size2i fms = Size2i(_size.y, _size.z);
  if (img_size != fms) {
    bool mip_maps = p_target->has_mipmaps();
    p_target->resize(fms.x, fms.y, Image::INTERPOLATE_NEAREST);

    if (p_target->has_mipmaps() != mip_maps) {
      if (mip_maps) {
        p_target->generate_mipmaps();
      } else {
        p_target->clear_mipmaps();
      }
    }
  }

  p_target->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int z = 0; z < fms.x; ++z) {
      Color c;

      float e = element_get(z, y, p_index_x);

      for (int i = 0; i < channel_count; ++i) {
        c[channels[i]] = e;
      }

      p_target->set_pixel(z, y, c);
    }
  }

  p_target->unlock();
}
void MLPPTensor3::x_slice_set_image(const Ref<Image> &p_img,
                                    const int p_index_x,
                                    const int p_image_channel_flag) {
  ERR_FAIL_COND(!p_img.is_valid());
  ERR_FAIL_INDEX(p_index_x, _size.x);

  int channel_index = -1;

  for (int i = 0; i < 4; ++i) {
    if (((p_image_channel_flag & (1 << i)) != 0)) {
      channel_index = i;
      break;
    }
  }

  ERR_FAIL_INDEX(channel_index, 4);

  Size2i img_size = Size2i(p_img->get_width(), p_img->get_height());
  Size2i fms = Size2i(_size.y, _size.z);

  ERR_FAIL_COND(img_size != fms);

  Ref<Image> img = p_img;

  img->lock();

  for (int y = 0; y < fms.y; ++y) {
    for (int z = 0; z < fms.x; ++z) {
      Color c = img->get_pixel(z, y);

      element_set(z, y, p_index_x, c[channel_index]);
    }
  }

  img->unlock();
}

Ref<Image> MLPPTensor3::y_slice_get_image(const int p_index_y) const {
  ERR_FAIL_INDEX_V(p_index_y, _size.y, Ref<Image>());

  Ref<Image> image;
  image.instance();

  if (data_size() == 0) {
    return image;
  }

  PoolByteArray arr;
  arr.resize(_size.x * _size.z);

  PoolByteArray::Write w = arr.write();
  uint8_t *wptr = w.ptr();
  int i = 0;

  for (int z = 0; z < _size.z; ++z) {
    for (int x = 0; x < _size.x; ++x) {
      wptr[i] = static_cast<uint8_t>(element_get(z, x, p_index_y) * 255.0);

      ++i;
    }
  }

  image->create(_size.x, _size.z, false, Image::FORMAT_L8, arr);

  return image;
}
void MLPPTensor3::y_slice_get_into_image(Ref<Image> p_target,
                                         const int p_index_y,
                                         const int p_target_channels) const {
  ERR_FAIL_INDEX(p_index_y, _size.y);
  ERR_FAIL_COND(!p_target.is_valid());

  int channel_count = 0;
  int channels[4];

  if (p_target_channels & IMAGE_CHANNEL_FLAG_R) {
    channels[channel_count] = 0;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_G) {
    channels[channel_count] = 1;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_B) {
    channels[channel_count] = 2;
    ++channel_count;
  }

  if (p_target_channels & IMAGE_CHANNEL_FLAG_A) {
    channels[channel_count] = 3;
    ++channel_count;
  }

  ERR_FAIL_COND(channel_count == 0);

  if (data_size() == 0) {
    p_target->clear();
    return;
  }

  Size2i img_size = Size2i(p_target->get_width(), p_target->get_height());
  Size2i fms = Size2i(_size.x, _size.z);
  if (img_size != fms) {
    bool mip_maps = p_target->has_mipmaps();
    p_target->resize(fms.x, fms.y, Image::INTERPOLATE_NEAREST);

    if (p_target->has_mipmaps() != mip_maps) {
      if (mip_maps) {
        p_target->generate_mipmaps();
      } else {
        p_target->clear_mipmaps();
      }
    }
  }

  p_target->lock();

  for (int x = 0; x < fms.y; ++x) {
    for (int z = 0; z < fms.x; ++z) {
      Color c;

      float e = element_get(z, p_index_y, x);

      for (int i = 0; i < channel_count; ++i) {
        c[channels[i]] = e;
      }

      p_target->set_pixel(z, x, c);
    }
  }

  p_target->unlock();
}
void MLPPTensor3::y_slice_set_image(const Ref<Image> &p_img,
                                    const int p_index_y,
                                    const int p_image_channel_flag) {
  ERR_FAIL_COND(!p_img.is_valid());
  ERR_FAIL_INDEX(p_index_y, _size.y);

  int channel_index = -1;

  for (int i = 0; i < 4; ++i) {
    if (((p_image_channel_flag & (1 << i)) != 0)) {
      channel_index = i;
      break;
    }
  }

  ERR_FAIL_INDEX(channel_index, 4);

  Size2i img_size = Size2i(p_img->get_width(), p_img->get_height());
  Size2i fms = Size2i(_size.x, _size.z);

  ERR_FAIL_COND(img_size != fms);

  Ref<Image> img = p_img;

  img->lock();

  for (int z = 0; z < fms.y; ++z) {
    for (int x = 0; x < fms.x; ++x) {
      Color c = img->get_pixel(x, z);

      element_set(z, p_index_y, x, c[channel_index]);
    }
  }

  img->unlock();
}

void MLPPTensor3::add(const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!B.is_valid());
  ERR_FAIL_COND(_size != B->size());

  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    c_ptr[i] += b_ptr[i];
  }
}
Ref<MLPPTensor3> MLPPTensor3::addn(const Ref<MLPPTensor3> &B) const {
  ERR_FAIL_COND_V(!B.is_valid(), Ref<MLPPTensor3>());
  ERR_FAIL_COND_V(_size != B->size(), Ref<MLPPTensor3>());

  Ref<MLPPTensor3> C;
  C.instance();
  C->resize(_size);

  const real_t *a_ptr = ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = C->ptrw();

  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    c_ptr[i] = a_ptr[i] + b_ptr[i];
  }

  return C;
}
void MLPPTensor3::addb(const Ref<MLPPTensor3> &A, const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!A.is_valid() || !B.is_valid());
  Size3i a_size = A->size();
  ERR_FAIL_COND(a_size != B->size());

  if (_size != a_size) {
    resize(a_size);
  }

  const real_t *a_ptr = A->ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  int data_size = A->data_size();

  for (int i = 0; i < data_size; ++i) {
    c_ptr[i] = a_ptr[i] + b_ptr[i];
  }
}

void MLPPTensor3::sub(const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!B.is_valid());
  ERR_FAIL_COND(_size != B->size());

  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    c_ptr[i] -= b_ptr[i];
  }
}
Ref<MLPPTensor3> MLPPTensor3::subn(const Ref<MLPPTensor3> &B) const {
  ERR_FAIL_COND_V(!B.is_valid(), Ref<MLPPTensor3>());
  ERR_FAIL_COND_V(_size != B->size(), Ref<MLPPTensor3>());

  Ref<MLPPTensor3> C;
  C.instance();
  C->resize(_size);

  const real_t *a_ptr = ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = C->ptrw();

  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    c_ptr[i] = a_ptr[i] - b_ptr[i];
  }

  return C;
}
void MLPPTensor3::subb(const Ref<MLPPTensor3> &A, const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!A.is_valid() || !B.is_valid());
  Size3i a_size = A->size();
  ERR_FAIL_COND(a_size != B->size());

  if (_size != a_size) {
    resize(a_size);
  }

  const real_t *a_ptr = A->ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  int data_size = A->data_size();

  for (int i = 0; i < data_size; ++i) {
    c_ptr[i] = a_ptr[i] - b_ptr[i];
  }
}

void MLPPTensor3::division_element_wise(const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!B.is_valid());
  ERR_FAIL_COND(_size != B->size());

  int ds = data_size();

  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  for (int i = 0; i < ds; i++) {
    c_ptr[i] /= b_ptr[i];
  }
}
Ref<MLPPTensor3>
MLPPTensor3::division_element_wisen(const Ref<MLPPTensor3> &B) const {
  ERR_FAIL_COND_V(!B.is_valid(), Ref<MLPPTensor3>());
  ERR_FAIL_COND_V(_size != B->size(), Ref<MLPPTensor3>());

  int ds = data_size();

  Ref<MLPPTensor3> C;
  C.instance();
  C->resize(_size);

  const real_t *a_ptr = ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = C->ptrw();

  for (int i = 0; i < ds; i++) {
    c_ptr[i] = a_ptr[i] / b_ptr[i];
  }

  return C;
}
void MLPPTensor3::division_element_wiseb(const Ref<MLPPTensor3> &A,
                                         const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!A.is_valid() || !B.is_valid());
  Size3i a_size = A->size();
  ERR_FAIL_COND(a_size != B->size());

  if (a_size != _size) {
    resize(a_size);
  }

  int ds = data_size();

  const real_t *a_ptr = A->ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  for (int i = 0; i < ds; i++) {
    c_ptr[i] = a_ptr[i] / b_ptr[i];
  }
}

void MLPPTensor3::sqrt() {
  int ds = data_size();

  real_t *out_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = Math::sqrt(out_ptr[i]);
  }
}
Ref<MLPPTensor3> MLPPTensor3::sqrtn() const {
  Ref<MLPPTensor3> out;
  out.instance();
  out->resize(size());

  int ds = data_size();

  const real_t *a_ptr = ptr();
  real_t *out_ptr = out->ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = Math::sqrt(a_ptr[i]);
  }

  return out;
}
void MLPPTensor3::sqrtb(const Ref<MLPPTensor3> &A) {
  ERR_FAIL_COND(!A.is_valid());

  Size3i a_size = A->size();

  if (a_size != size()) {
    resize(a_size);
  }

  int ds = data_size();

  const real_t *a_ptr = A->ptr();
  real_t *out_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = Math::sqrt(a_ptr[i]);
  }
}

void MLPPTensor3::exponentiate(real_t p) {
  int ds = data_size();

  real_t *out_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = Math::pow(out_ptr[i], p);
  }
}
Ref<MLPPTensor3> MLPPTensor3::exponentiaten(real_t p) const {
  Ref<MLPPTensor3> out;
  out.instance();
  out->resize(size());

  int ds = data_size();

  const real_t *a_ptr = ptr();
  real_t *out_ptr = out->ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = Math::pow(a_ptr[i], p);
  }

  return out;
}
void MLPPTensor3::exponentiateb(const Ref<MLPPTensor3> &A, real_t p) {
  ERR_FAIL_COND(!A.is_valid());

  Size3i a_size = A->size();

  if (a_size != size()) {
    resize(a_size);
  }

  int ds = data_size();

  const real_t *a_ptr = A->ptr();
  real_t *out_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = Math::pow(a_ptr[i], p);
  }
}

void MLPPTensor3::scalar_multiply(const real_t scalar) {
  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    _data[i] *= scalar;
  }
}
Ref<MLPPTensor3> MLPPTensor3::scalar_multiplyn(const real_t scalar) const {
  Ref<MLPPTensor3> AN = duplicate_fast();
  int ds = AN->data_size();
  real_t *an_ptr = AN->ptrw();

  for (int i = 0; i < ds; ++i) {
    an_ptr[i] *= scalar;
  }

  return AN;
}
void MLPPTensor3::scalar_multiplyb(const real_t scalar,
                                   const Ref<MLPPTensor3> &A) {
  ERR_FAIL_COND(!A.is_valid());

  if (A->size() != _size) {
    resize(A->size());
  }

  int ds = data_size();
  real_t *an_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    _data[i] = an_ptr[i] * scalar;
  }
}

void MLPPTensor3::scalar_add(const real_t scalar) {
  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    _data[i] += scalar;
  }
}
Ref<MLPPTensor3> MLPPTensor3::scalar_addn(const real_t scalar) const {
  Ref<MLPPTensor3> AN = duplicate_fast();
  int ds = AN->data_size();
  real_t *an_ptr = AN->ptrw();

  for (int i = 0; i < ds; ++i) {
    an_ptr[i] += scalar;
  }

  return AN;
}
void MLPPTensor3::scalar_addb(const real_t scalar, const Ref<MLPPTensor3> &A) {
  ERR_FAIL_COND(!A.is_valid());

  if (A->size() != _size) {
    resize(A->size());
  }

  int ds = data_size();
  real_t *an_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    _data[i] = an_ptr[i] + scalar;
  }
}

void MLPPTensor3::hadamard_product(const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!B.is_valid());
  ERR_FAIL_COND(_size != B->size());

  int ds = data_size();

  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  for (int i = 0; i < ds; i++) {
    c_ptr[i] = c_ptr[i] * b_ptr[i];
  }
}
Ref<MLPPTensor3>
MLPPTensor3::hadamard_productn(const Ref<MLPPTensor3> &B) const {
  ERR_FAIL_COND_V(!B.is_valid(), Ref<MLPPTensor3>());
  ERR_FAIL_COND_V(_size != B->size(), Ref<MLPPTensor3>());

  int ds = data_size();

  Ref<MLPPTensor3> C;
  C.instance();
  C->resize(_size);

  const real_t *a_ptr = ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = C->ptrw();

  for (int i = 0; i < ds; i++) {
    c_ptr[i] = a_ptr[i] * b_ptr[i];
  }

  return C;
}
void MLPPTensor3::hadamard_productb(const Ref<MLPPTensor3> &A,
                                    const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!A.is_valid() || !B.is_valid());
  Size3i a_size = A->size();
  ERR_FAIL_COND(a_size != B->size());

  if (a_size != _size) {
    resize(a_size);
  }

  int ds = data_size();

  const real_t *a_ptr = A->ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  for (int i = 0; i < ds; i++) {
    c_ptr[i] = a_ptr[i] * b_ptr[i];
  }
}

void MLPPTensor3::max(const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!B.is_valid());
  ERR_FAIL_COND(_size != B->size());

  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    c_ptr[i] = MAX(c_ptr[i], b_ptr[i]);
  }
}
Ref<MLPPTensor3> MLPPTensor3::maxn(const Ref<MLPPTensor3> &B) const {
  ERR_FAIL_COND_V(!B.is_valid(), Ref<MLPPTensor3>());
  ERR_FAIL_COND_V(_size != B->size(), Ref<MLPPTensor3>());

  Ref<MLPPTensor3> C;
  C.instance();
  C->resize(_size);

  const real_t *a_ptr = ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = C->ptrw();

  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    c_ptr[i] = MAX(a_ptr[i], b_ptr[i]);
  }

  return C;
}
void MLPPTensor3::maxb(const Ref<MLPPTensor3> &A, const Ref<MLPPTensor3> &B) {
  ERR_FAIL_COND(!A.is_valid() || !B.is_valid());
  Size3i a_size = A->size();
  ERR_FAIL_COND(a_size != B->size());

  if (_size != a_size) {
    resize(a_size);
  }

  const real_t *a_ptr = A->ptr();
  const real_t *b_ptr = B->ptr();
  real_t *c_ptr = ptrw();

  int data_size = A->data_size();

  for (int i = 0; i < data_size; ++i) {
    c_ptr[i] = MAX(a_ptr[i], b_ptr[i]);
  }
}

void MLPPTensor3::abs() {
  int ds = data_size();

  real_t *out_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = ABS(out_ptr[i]);
  }
}
Ref<MLPPTensor3> MLPPTensor3::absn() const {
  Ref<MLPPTensor3> out;
  out.instance();
  out->resize(size());

  int ds = data_size();

  const real_t *a_ptr = ptr();
  real_t *out_ptr = out->ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = ABS(a_ptr[i]);
  }

  return out;
}
void MLPPTensor3::absb(const Ref<MLPPTensor3> &A) {
  ERR_FAIL_COND(!A.is_valid());

  Size3i a_size = A->size();

  if (a_size != size()) {
    resize(a_size);
  }

  int ds = data_size();

  const real_t *a_ptr = A->ptr();
  real_t *out_ptr = ptrw();

  for (int i = 0; i < ds; ++i) {
    out_ptr[i] = ABS(a_ptr[i]);
  }
}

Ref<MLPPVector> MLPPTensor3::flatten() const {
  int ds = data_size();

  Ref<MLPPVector> res;
  res.instance();
  res->resize(ds);

  real_t *res_ptr = res->ptrw();
  const real_t *a_ptr = ptr();

  for (int i = 0; i < ds; ++i) {
    res_ptr[i] = a_ptr[i];
  }

  return res;
}
void MLPPTensor3::flatteno(Ref<MLPPVector> out) const {
  ERR_FAIL_COND(!out.is_valid());

  int ds = data_size();

  if (unlikely(out->size() != ds)) {
    out->resize(ds);
  }

  real_t *res_ptr = out->ptrw();
  const real_t *a_ptr = ptr();

  for (int i = 0; i < ds; ++i) {
    res_ptr[i] = a_ptr[i];
  }
}

/*
real_t MLPPTensor3::norm_2(std::vector<std::vector<std::vector<real_t>>> A) {
        real_t sum = 0;
        for (uint32_t i = 0; i < A.size(); i++) {
                for (uint32_t j = 0; j < A[i].size(); j++) {
                        for (uint32_t k = 0; k < A[i][j].size(); k++) {
                                sum += A[i][j][k] * A[i][j][k];
                        }
                }
        }
        return Math::sqrt(sum);
}
*/

Ref<MLPPMatrix> MLPPTensor3::tensor_vec_mult(const Ref<MLPPVector> &b) {
  Ref<MLPPMatrix> C;
  C.instance();
  C->resize(Size2i(_size.y, _size.z));

  Ref<MLPPVector> row_tmp;
  row_tmp.instance();
  row_tmp->resize(_size.x);

  for (int i = 0; i < _size.z; i++) {
    for (int j = 0; j < _size.y; j++) {
      row_get_into_mlpp_vector(i, j, row_tmp);

      C->element_set(i, j, row_tmp->dot(b));
    }
  }

  return C;
}

/*
// Bad implementation. Change this later.
std::vector<std::vector<std::vector<real_t>>>
MLPPTensor3::vector_wise_tensor_product(std::vector<std::vector<std::vector<real_t>>>
A, std::vector<std::vector<real_t>> B) {
        std::vector<std::vector<std::vector<real_t>>> C;
        C = resize(C, A);
        for (uint32_t i = 0; i < A[0].size(); i++) {
                for (uint32_t j = 0; j < A[0][i].size(); j++) {
                        std::vector<real_t> currentVector;
                        currentVector.resize(A.size());

                        for (uint32_t k = 0; k < C.size(); k++) {
                                currentVector[k] = A[k][i][j];
                        }

                        currentVector = mat_vec_mult(B, currentVector);

                        for (uint32_t k = 0; k < C.size(); k++) {
                                C[k][i][j] = currentVector[k];
                        }
                }
        }
        return C;
}
*/

void MLPPTensor3::fill(real_t p_val) {
  if (!_data) {
    return;
  }

  int ds = data_size();
  for (int i = 0; i < ds; ++i) {
    _data[i] = p_val;
  }
}

Vector<real_t> MLPPTensor3::to_flat_vector() const {
  Vector<real_t> ret;
  ret.resize(data_size());
  real_t *w = ret.ptrw();
  memcpy(w, _data, sizeof(real_t) * data_size());
  return ret;
}

PoolRealArray MLPPTensor3::to_flat_pool_vector() const {
  PoolRealArray pl;
  if (data_size()) {
    pl.resize(data_size());
    typename PoolRealArray::Write w = pl.write();
    real_t *dest = w.ptr();

    for (int i = 0; i < data_size(); ++i) {
      dest[i] = static_cast<real_t>(_data[i]);
    }
  }
  return pl;
}

Vector<uint8_t> MLPPTensor3::to_flat_byte_array() const {
  Vector<uint8_t> ret;
  ret.resize(data_size() * sizeof(real_t));
  uint8_t *w = ret.ptrw();
  memcpy(w, _data, sizeof(real_t) * data_size());
  return ret;
}

Ref<MLPPTensor3> MLPPTensor3::duplicate_fast() const {
  Ref<MLPPTensor3> ret;
  ret.instance();

  ret->set_from_mlpp_tensor3r(*this);

  return ret;
}

void MLPPTensor3::set_from_mlpp_tensor3(const Ref<MLPPTensor3> &p_from) {
  ERR_FAIL_COND(!p_from.is_valid());

  resize(p_from->size());

  int ds = p_from->data_size();
  const real_t *ptr = p_from->ptr();

  for (int i = 0; i < ds; ++i) {
    _data[i] = ptr[i];
  }
}

void MLPPTensor3::set_from_mlpp_tensor3r(const MLPPTensor3 &p_from) {
  resize(p_from.size());

  int ds = p_from.data_size();
  const real_t *ptr = p_from.ptr();

  for (int i = 0; i < ds; ++i) {
    _data[i] = ptr[i];
  }
}

void MLPPTensor3::set_from_mlpp_matrix(const Ref<MLPPMatrix> &p_from) {
  ERR_FAIL_COND(!p_from.is_valid());

  Size2i mat_size = p_from->size();
  resize(Size3i(mat_size.x, mat_size.y, 1));

  int ds = p_from->data_size();
  const real_t *ptr = p_from->ptr();

  for (int i = 0; i < ds; ++i) {
    _data[i] = ptr[i];
  }
}

void MLPPTensor3::set_from_mlpp_matrixr(const MLPPMatrix &p_from) {
  Size2i mat_size = p_from.size();
  resize(Size3i(mat_size.x, mat_size.y, 1));

  int ds = p_from.data_size();
  const real_t *ptr = p_from.ptr();

  for (int i = 0; i < ds; ++i) {
    _data[i] = ptr[i];
  }
}

void MLPPTensor3::set_from_mlpp_vectors(const Vector<Ref<MLPPVector>> &p_from) {
  if (p_from.size() == 0) {
    reset();
    return;
  }

  if (!p_from[0].is_valid()) {
    reset();
    return;
  }

  resize(Size3i(p_from[0]->size(), p_from.size(), 1));

  if (data_size() == 0) {
    reset();
    return;
  }

  for (int i = 0; i < p_from.size(); ++i) {
    const Ref<MLPPVector> &r = p_from[i];

    ERR_CONTINUE(!r.is_valid());
    ERR_CONTINUE(r->size() != _size.x);

    int start_index = i * _size.x;

    const real_t *from_ptr = r->ptr();
    for (int j = 0; j < _size.x; j++) {
      _data[start_index + j] = from_ptr[j];
    }
  }
}

void MLPPTensor3::set_from_mlpp_matricess(
    const Vector<Ref<MLPPMatrix>> &p_from) {
  if (p_from.size() == 0) {
    reset();
    return;
  }

  if (!p_from[0].is_valid()) {
    reset();
    return;
  }

  resize(Size3i(p_from[0]->size().x, p_from[0]->size().y, p_from.size()));

  if (data_size() == 0) {
    reset();
    return;
  }

  Size2i fms = z_slice_size();
  int fmds = z_slice_data_size();

  for (int i = 0; i < p_from.size(); ++i) {
    const Ref<MLPPMatrix> &r = p_from[i];

    ERR_CONTINUE(!r.is_valid());
    ERR_CONTINUE(r->size() != fms);

    int start_index = calculate_z_slice_index(i);

    const real_t *from_ptr = r->ptr();
    for (int j = 0; j < fmds; j++) {
      _data[start_index + j] = from_ptr[j];
    }
  }
}

void MLPPTensor3::set_from_mlpp_vectors_array(const Array &p_from) {
  if (p_from.size() == 0) {
    reset();
    return;
  }

  Ref<MLPPVector> v0 = p_from[0];

  if (!v0.is_valid()) {
    reset();
    return;
  }

  resize(Size3i(v0->size(), p_from.size(), 1));

  if (data_size() == 0) {
    reset();
    return;
  }

  for (int i = 0; i < p_from.size(); ++i) {
    Ref<MLPPVector> r = p_from[i];

    ERR_CONTINUE(!r.is_valid());
    ERR_CONTINUE(r->size() != _size.x);

    int start_index = i * _size.x;

    const real_t *from_ptr = r->ptr();
    for (int j = 0; j < _size.x; j++) {
      _data[start_index + j] = from_ptr[j];
    }
  }
}

void MLPPTensor3::set_from_mlpp_matrices_array(const Array &p_from) {
  if (p_from.size() == 0) {
    reset();
    return;
  }

  Ref<MLPPMatrix> v0 = p_from[0];

  if (!v0.is_valid()) {
    reset();
    return;
  }

  resize(Size3i(v0->size().x, v0->size().y, p_from.size()));

  if (data_size() == 0) {
    reset();
    return;
  }

  Size2i fms = z_slice_size();
  int fmds = z_slice_data_size();

  for (int i = 0; i < p_from.size(); ++i) {
    Ref<MLPPMatrix> r = p_from[i];

    ERR_CONTINUE(!r.is_valid());
    ERR_CONTINUE(r->size() != fms);

    int start_index = calculate_z_slice_index(i);

    const real_t *from_ptr = r->ptr();
    for (int j = 0; j < fmds; j++) {
      _data[start_index + j] = from_ptr[j];
    }
  }
}

bool MLPPTensor3::is_equal_approx(const Ref<MLPPTensor3> &p_with,
                                  real_t tolerance) const {
  ERR_FAIL_COND_V(!p_with.is_valid(), false);

  if (unlikely(this == p_with.ptr())) {
    return true;
  }

  if (_size != p_with->size()) {
    return false;
  }

  int ds = data_size();

  for (int i = 0; i < ds; ++i) {
    if (!Math::is_equal_approx(_data[i], p_with->_data[i], tolerance)) {
      return false;
    }
  }

  return true;
}

String MLPPTensor3::to_string() {
  String str;

  str += "[MLPPTensor3: \n";

  for (int z = 0; z < _size.z; ++z) {
    int z_ofs = _size.x * _size.y * z;

    str += "  [ ";

    for (int y = 0; y < _size.y; ++y) {
      str += "    [ ";

      for (int x = 0; x < _size.x; ++x) {
        str += String::num(_data[_size.x * y + x + z_ofs]);
        str += " ";
      }

      str += "  ]\n";
    }

    str += "],\n";
  }

  str += "]\n";

  return str;
}

MLPPTensor3::MLPPTensor3() { _data = NULL; }

MLPPTensor3::MLPPTensor3(const MLPPMatrix &p_from) {
  _data = NULL;

  Size2i mat_size = p_from.size();
  resize(Size3i(mat_size.x, mat_size.y, 1));

  int ds = p_from.data_size();
  const real_t *ptr = p_from.ptr();

  for (int i = 0; i < ds; ++i) {
    _data[i] = ptr[i];
  }
}

MLPPTensor3::MLPPTensor3(const Array &p_from) {
  _data = NULL;

  set_from_mlpp_matrices_array(p_from);
}

MLPPTensor3::~MLPPTensor3() {
  if (_data) {
    reset();
  }
}

std::vector<real_t> MLPPTensor3::to_flat_std_vector() const {
  std::vector<real_t> ret;
  ret.resize(data_size());
  real_t *w = &ret[0];
  memcpy(w, _data, sizeof(real_t) * data_size());
  return ret;
}

void MLPPTensor3::set_from_std_vectors(
    const std::vector<std::vector<std::vector<real_t>>> &p_from) {
  if (p_from.size() == 0) {
    reset();
    return;
  }

  resize(Size3i(p_from[0][0].size(), p_from[0].size(), p_from.size()));

  if (data_size() == 0) {
    reset();
    return;
  }

  for (uint32_t z = 0; z < p_from.size(); ++z) {
    const std::vector<std::vector<real_t>> &vxy = p_from[z];

    for (uint32_t y = 0; y < vxy.size(); ++y) {
      ERR_CONTINUE(vxy.size() != static_cast<uint32_t>(_size.y));

      const std::vector<real_t> &vx = vxy[y];

      ERR_CONTINUE(vx.size() != static_cast<uint32_t>(_size.x));

      for (uint32_t x = 0; x < vx.size(); ++x) {
        element_set(z, y, x, vx[x]);
      }
    }
  }
}

std::vector<std::vector<std::vector<real_t>>> MLPPTensor3::to_std_vector() {
  std::vector<std::vector<std::vector<real_t>>> ret;

  ret.resize(_size.z);

  for (int k = 0; k < _size.z; ++k) {
    ret[k].resize(_size.y);

    for (int i = 0; i < _size.y; ++i) {
      std::vector<real_t> row;

      for (int j = 0; j < _size.x; ++j) {
        row.push_back(_data[calculate_index(k, i, j)]);
      }

      ret[k][i] = row;
    }
  }

  return ret;
}

MLPPTensor3::MLPPTensor3(
    const std::vector<std::vector<std::vector<real_t>>> &p_from) {
  _data = NULL;

  set_from_std_vectors(p_from);
}

void MLPPTensor3::_bind_methods() {
  ClassDB::bind_method(D_METHOD("get_data"), &MLPPTensor3::get_data);
  ClassDB::bind_method(D_METHOD("set_data", "data"), &MLPPTensor3::set_data);
  ADD_PROPERTY(PropertyInfo(Variant::ARRAY, "data"), "set_data", "get_data");

  ClassDB::bind_method(D_METHOD("z_slice_add_pool_vector", "row"),
                       &MLPPTensor3::z_slice_add_pool_vector);
  ClassDB::bind_method(D_METHOD("z_slice_add_mlpp_vector", "row"),
                       &MLPPTensor3::z_slice_add_mlpp_vector);
  ClassDB::bind_method(D_METHOD("z_slice_add_mlpp_matrix", "matrix"),
                       &MLPPTensor3::z_slice_add_mlpp_matrix);

  ClassDB::bind_method(D_METHOD("z_slice_remove", "index"),
                       &MLPPTensor3::z_slice_remove);
  ClassDB::bind_method(D_METHOD("z_slice_remove_unordered", "index"),
                       &MLPPTensor3::z_slice_remove_unordered);

  ClassDB::bind_method(D_METHOD("z_slice_swap", "index_1", "index_2"),
                       &MLPPTensor3::z_slice_swap);

  ClassDB::bind_method(D_METHOD("clear"), &MLPPTensor3::clear);
  ClassDB::bind_method(D_METHOD("reset"), &MLPPTensor3::reset);
  ClassDB::bind_method(D_METHOD("empty"), &MLPPTensor3::empty);

  ClassDB::bind_method(D_METHOD("z_slice_data_size"),
                       &MLPPTensor3::z_slice_data_size);
  ClassDB::bind_method(D_METHOD("z_slice_size"), &MLPPTensor3::z_slice_size);

  ClassDB::bind_method(D_METHOD("data_size"), &MLPPTensor3::data_size);
  ClassDB::bind_method(D_METHOD("size"), &MLPPTensor3::size);

  ClassDB::bind_method(D_METHOD("resize", "size"), &MLPPTensor3::resize);

  ClassDB::bind_method(D_METHOD("shape_set", "size"), &MLPPTensor3::shape_set);
  ClassDB::bind_method(
      D_METHOD("calculate_index", "index_y", "index_x", "index_z"),
      &MLPPTensor3::calculate_index);
  ClassDB::bind_method(D_METHOD("calculate_z_slice_index", "index_z"),
                       &MLPPTensor3::calculate_z_slice_index);

  ClassDB::bind_method(D_METHOD("element_get_index", "index"),
                       &MLPPTensor3::element_get_index);
  ClassDB::bind_method(D_METHOD("element_set_index", "index", "val"),
                       &MLPPTensor3::element_set_index);

  ClassDB::bind_method(D_METHOD("element_get", "index_y", "index_x", "index_z"),
                       &MLPPTensor3::element_get);
  ClassDB::bind_method(
      D_METHOD("element_set", "index_y", "index_x", "index_z", "val"),
      &MLPPTensor3::element_set);

  ClassDB::bind_method(D_METHOD("row_get_pool_vector", "index_y", "index_z"),
                       &MLPPTensor3::row_get_pool_vector);
  ClassDB::bind_method(D_METHOD("row_get_mlpp_vector", "index_y", "index_z"),
                       &MLPPTensor3::row_get_mlpp_vector);
  ClassDB::bind_method(
      D_METHOD("row_get_into_mlpp_vector", "index_y", "index_z", "target"),
      &MLPPTensor3::row_get_into_mlpp_vector);

  ClassDB::bind_method(
      D_METHOD("row_set_pool_vector", "index_y", "index_z", "row"),
      &MLPPTensor3::row_set_pool_vector);
  ClassDB::bind_method(
      D_METHOD("row_set_mlpp_vector", "index_y", "index_z", "row"),
      &MLPPTensor3::row_set_mlpp_vector);

  ClassDB::bind_method(D_METHOD("z_slice_get_pool_vector", "index_z"),
                       &MLPPTensor3::z_slice_get_pool_vector);
  ClassDB::bind_method(D_METHOD("z_slice_get_mlpp_vector", "index_z"),
                       &MLPPTensor3::z_slice_get_mlpp_vector);
  ClassDB::bind_method(
      D_METHOD("z_slice_get_into_mlpp_vector", "index_z", "target"),
      &MLPPTensor3::z_slice_get_into_mlpp_vector);

  ClassDB::bind_method(D_METHOD("z_slice_get_mlpp_matrix", "index_z"),
                       &MLPPTensor3::z_slice_get_mlpp_matrix);
  ClassDB::bind_method(
      D_METHOD("z_slice_get_into_mlpp_matrix", "index_z", "target"),
      &MLPPTensor3::z_slice_get_into_mlpp_matrix);

  ClassDB::bind_method(D_METHOD("z_slice_set_pool_vector", "index_z", "row"),
                       &MLPPTensor3::z_slice_set_pool_vector);
  ClassDB::bind_method(D_METHOD("z_slice_set_mlpp_vector", "index_z", "row"),
                       &MLPPTensor3::z_slice_set_mlpp_vector);
  ClassDB::bind_method(D_METHOD("z_slice_set_mlpp_matrix", "index_z", "mat"),
                       &MLPPTensor3::z_slice_set_mlpp_matrix);

  ClassDB::bind_method(D_METHOD("x_slice_get_into", "index_x", "target"),
                       &MLPPTensor3::x_slice_get_into);
  ClassDB::bind_method(D_METHOD("x_slice_get", "index_x"),
                       &MLPPTensor3::x_slice_get);
  ClassDB::bind_method(D_METHOD("x_slice_set", "index_x", "mat"),
                       &MLPPTensor3::x_slice_set);

  ClassDB::bind_method(D_METHOD("y_slice_get_into", "index_y", "target"),
                       &MLPPTensor3::y_slice_get_into);
  ClassDB::bind_method(D_METHOD("y_slice_get", "index_y"),
                       &MLPPTensor3::y_slice_get);
  ClassDB::bind_method(D_METHOD("y_slice_set", "index_y", "mat"),
                       &MLPPTensor3::y_slice_set);

  ClassDB::bind_method(D_METHOD("z_slices_add_image", "img", "channels"),
                       &MLPPTensor3::z_slices_add_image,
                       IMAGE_CHANNEL_FLAG_RGBA);

  ClassDB::bind_method(D_METHOD("z_slice_get_image", "index_z"),
                       &MLPPTensor3::z_slice_get_image);
  ClassDB::bind_method(D_METHOD("z_slices_get_image", "index_r", "index_g",
                                "index_b", "index_a"),
                       &MLPPTensor3::z_slices_get_image, -1, -1, -1, -1);

  ClassDB::bind_method(D_METHOD("z_slice_get_into_image", "target", "index_z",
                                "target_channels"),
                       &MLPPTensor3::z_slice_get_into_image,
                       IMAGE_CHANNEL_FLAG_RGB);
  ClassDB::bind_method(D_METHOD("z_slices_get_into_image", "target", "index_r",
                                "index_g", "index_b", "index_a"),
                       &MLPPTensor3::z_slices_get_into_image, -1, -1, -1, -1);

  ClassDB::bind_method(
      D_METHOD("z_slice_set_image", "img", "index_z", "image_channel_flag"),
      &MLPPTensor3::z_slice_set_image, IMAGE_CHANNEL_FLAG_R);
  ClassDB::bind_method(D_METHOD("z_slices_set_image", "img", "index_r",
                                "index_g", "index_b", "index_a"),
                       &MLPPTensor3::z_slices_set_image);

  ClassDB::bind_method(D_METHOD("set_from_image", "img", "channels"),
                       &MLPPTensor3::set_from_image, IMAGE_CHANNEL_FLAG_RGBA);

  ClassDB::bind_method(D_METHOD("x_slice_get_image", "index_x"),
                       &MLPPTensor3::x_slice_get_image);
  ClassDB::bind_method(D_METHOD("x_slice_get_into_image", "target", "index_x",
                                "target_channels"),
                       &MLPPTensor3::x_slice_get_into_image,
                       IMAGE_CHANNEL_FLAG_RGB);
  ClassDB::bind_method(
      D_METHOD("x_slice_set_image", "img", "index_x", "image_channel_flag"),
      &MLPPTensor3::x_slice_set_image, IMAGE_CHANNEL_FLAG_R);

  ClassDB::bind_method(D_METHOD("y_slice_get_image", "index_x"),
                       &MLPPTensor3::y_slice_get_image);
  ClassDB::bind_method(D_METHOD("y_slice_get_into_image", "target", "index_x",
                                "target_channels"),
                       &MLPPTensor3::y_slice_get_into_image,
                       IMAGE_CHANNEL_FLAG_RGB);
  ClassDB::bind_method(
      D_METHOD("y_slice_set_image", "img", "index_x", "image_channel_flag"),
      &MLPPTensor3::y_slice_set_image, IMAGE_CHANNEL_FLAG_R);

  ClassDB::bind_method(D_METHOD("fill", "val"), &MLPPTensor3::fill);

  ClassDB::bind_method(D_METHOD("to_flat_pool_vector"),
                       &MLPPTensor3::to_flat_pool_vector);
  ClassDB::bind_method(D_METHOD("to_flat_byte_array"),
                       &MLPPTensor3::to_flat_byte_array);

  ClassDB::bind_method(D_METHOD("duplicate_fast"),
                       &MLPPTensor3::duplicate_fast);

  ClassDB::bind_method(D_METHOD("set_from_mlpp_tensor3", "from"),
                       &MLPPTensor3::set_from_mlpp_tensor3);
  ClassDB::bind_method(D_METHOD("set_from_mlpp_matrix", "from"),
                       &MLPPTensor3::set_from_mlpp_matrix);
  ClassDB::bind_method(D_METHOD("set_from_mlpp_vectors_array", "from"),
                       &MLPPTensor3::set_from_mlpp_vectors_array);
  ClassDB::bind_method(D_METHOD("set_from_mlpp_matrices_array", "from"),
                       &MLPPTensor3::set_from_mlpp_matrices_array);

  ClassDB::bind_method(D_METHOD("is_equal_approx", "with", "tolerance"),
                       &MLPPTensor3::is_equal_approx, CMP_EPSILON);

  ClassDB::bind_method(D_METHOD("add", "B"), &MLPPTensor3::add);
  ClassDB::bind_method(D_METHOD("addn", "B"), &MLPPTensor3::addn);
  ClassDB::bind_method(D_METHOD("addb", "A", "B"), &MLPPTensor3::addb);

  ClassDB::bind_method(D_METHOD("sub", "B"), &MLPPTensor3::sub);
  ClassDB::bind_method(D_METHOD("subn", "B"), &MLPPTensor3::subn);
  ClassDB::bind_method(D_METHOD("subb", "A", "B"), &MLPPTensor3::subb);

  ClassDB::bind_method(D_METHOD("hadamard_product", "B"),
                       &MLPPTensor3::hadamard_product);
  ClassDB::bind_method(D_METHOD("hadamard_productn", "B"),
                       &MLPPTensor3::hadamard_productn);
  ClassDB::bind_method(D_METHOD("hadamard_productb", "A", "B"),
                       &MLPPTensor3::hadamard_productb);

  ClassDB::bind_method(D_METHOD("division_element_wise", "B"),
                       &MLPPTensor3::division_element_wise);
  ClassDB::bind_method(D_METHOD("division_element_wisen", "B"),
                       &MLPPTensor3::division_element_wisen);
  ClassDB::bind_method(D_METHOD("division_element_wiseb", "A", "B"),
                       &MLPPTensor3::division_element_wiseb);

  ClassDB::bind_method(D_METHOD("scalar_multiply", "scalar"),
                       &MLPPTensor3::scalar_multiply);
  ClassDB::bind_method(D_METHOD("scalar_multiplyn", "scalar"),
                       &MLPPTensor3::scalar_multiplyn);
  ClassDB::bind_method(D_METHOD("scalar_multiplyb", "scalar", "A"),
                       &MLPPTensor3::scalar_multiplyb);

  ClassDB::bind_method(D_METHOD("scalar_add", "scalar"),
                       &MLPPTensor3::scalar_add);
  ClassDB::bind_method(D_METHOD("scalar_addn", "scalar"),
                       &MLPPTensor3::scalar_addn);
  ClassDB::bind_method(D_METHOD("scalar_addb", "scalar", "A"),
                       &MLPPTensor3::scalar_addb);

  ClassDB::bind_method(D_METHOD("exponentiate", "p"),
                       &MLPPTensor3::exponentiate);
  ClassDB::bind_method(D_METHOD("exponentiaten", "p"),
                       &MLPPTensor3::exponentiaten);
  ClassDB::bind_method(D_METHOD("exponentiateb", "A", "p"),
                       &MLPPTensor3::exponentiateb);

  ClassDB::bind_method(D_METHOD("sqrt"), &MLPPTensor3::sqrt);
  ClassDB::bind_method(D_METHOD("sqrtn"), &MLPPTensor3::sqrtn);
  ClassDB::bind_method(D_METHOD("sqrtb", "A"), &MLPPTensor3::sqrtb);

  ClassDB::bind_method(D_METHOD("abs"), &MLPPTensor3::abs);
  ClassDB::bind_method(D_METHOD("absn"), &MLPPTensor3::absn);
  ClassDB::bind_method(D_METHOD("absb", "A"), &MLPPTensor3::absb);

  ClassDB::bind_method(D_METHOD("max", "B"), &MLPPTensor3::max);
  ClassDB::bind_method(D_METHOD("maxn", "B"), &MLPPTensor3::maxn);
  ClassDB::bind_method(D_METHOD("maxb", "A", "B"), &MLPPTensor3::maxb);

  ClassDB::bind_method(D_METHOD("flatten"), &MLPPTensor3::flatten);
  ClassDB::bind_method(D_METHOD("flatteno", "out"), &MLPPTensor3::flatteno);

  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_R);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_G);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_B);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_A);

  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_NONE);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_RG);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_RGB);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_GB);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_GBA);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_BA);
  BIND_ENUM_CONSTANT(IMAGE_CHANNEL_FLAG_RGBA);
}