From a5ffa496793dc7c96ea3b1f9852381837e98e3bf Mon Sep 17 00:00:00 2001
From: Oli Larkin <olilarkin@googlemail.com>
Date: Wed, 27 Dec 2023 11:45:08 +0100
Subject: [PATCH 1/3] Pass in const reference to weights

And const_iterator rather than iterator
---
 NAM/convnet.cpp |  8 ++++----
 NAM/convnet.h   |  6 +++---
 NAM/dsp.cpp     |  6 +++---
 NAM/dsp.h       |  6 +++---
 NAM/lstm.cpp    |  6 +++---
 NAM/lstm.h      |  4 ++--
 NAM/wavenet.cpp | 12 ++++++------
 NAM/wavenet.h   | 10 +++++-----
 8 files changed, 29 insertions(+), 29 deletions(-)

diff --git a/NAM/convnet.cpp b/NAM/convnet.cpp
index 1d4e3ba..bc0a392 100644
--- a/NAM/convnet.cpp
+++ b/NAM/convnet.cpp
@@ -10,7 +10,7 @@
 #include "dsp.h"
 #include "convnet.h"
 
-nam::convnet::BatchNorm::BatchNorm(const int dim, std::vector<float>::iterator& weights)
+nam::convnet::BatchNorm::BatchNorm(const int dim, std::vector<float>::const_iterator& weights)
 {
   // Extract from param buffer
   Eigen::VectorXf running_mean(dim);
@@ -48,7 +48,7 @@ void nam::convnet::BatchNorm::process_(Eigen::MatrixXf& x, const long i_start, c
 
 void nam::convnet::ConvNetBlock::set_weights_(const int in_channels, const int out_channels, const int _dilation,
                                               const bool batchnorm, const std::string activation,
-                                              std::vector<float>::iterator& weights)
+                                              std::vector<float>::const_iterator& weights)
 {
   this->_batchnorm = batchnorm;
   // HACK 2 kernel
@@ -74,7 +74,7 @@ long nam::convnet::ConvNetBlock::get_out_channels() const
   return this->conv.get_out_channels();
 }
 
-nam::convnet::_Head::_Head(const int channels, std::vector<float>::iterator& weights)
+nam::convnet::_Head::_Head(const int channels, std::vector<float>::const_iterator& weights)
 {
   this->_weight.resize(channels);
   for (int i = 0; i < channels; i++)
@@ -98,7 +98,7 @@ nam::convnet::ConvNet::ConvNet(const int channels, const std::vector<int>& dilat
 {
   this->_verify_weights(channels, dilations, batchnorm, weights.size());
   this->_blocks.resize(dilations.size());
-  std::vector<float>::iterator it = weights.begin();
+  std::vector<float>::const_iterator it = weights.begin();
   for (size_t i = 0; i < dilations.size(); i++)
     this->_blocks[i].set_weights_(i == 0 ? 1 : channels, channels, dilations[i], batchnorm, activation, it);
   this->_block_vals.resize(this->_blocks.size() + 1);
diff --git a/NAM/convnet.h b/NAM/convnet.h
index 310a1e5..a410506 100644
--- a/NAM/convnet.h
+++ b/NAM/convnet.h
@@ -23,7 +23,7 @@ class BatchNorm
 {
 public:
   BatchNorm(){};
-  BatchNorm(const int dim, std::vector<float>::iterator& weights);
+  BatchNorm(const int dim, std::vector<float>::const_iterator& weights);
   void process_(Eigen::MatrixXf& input, const long i_start, const long i_end) const;
 
 private:
@@ -41,7 +41,7 @@ class ConvNetBlock
 public:
   ConvNetBlock(){};
   void set_weights_(const int in_channels, const int out_channels, const int _dilation, const bool batchnorm,
-                    const std::string activation, std::vector<float>::iterator& weights);
+                    const std::string activation, std::vector<float>::const_iterator& weights);
   void process_(const Eigen::MatrixXf& input, Eigen::MatrixXf& output, const long i_start, const long i_end) const;
   long get_out_channels() const;
   Conv1D conv;
@@ -56,7 +56,7 @@ class _Head
 {
 public:
   _Head(){};
-  _Head(const int channels, std::vector<float>::iterator& weights);
+  _Head(const int channels, std::vector<float>::const_iterator& weights);
   void process_(const Eigen::MatrixXf& input, Eigen::VectorXf& output, const long i_start, const long i_end) const;
 
 private:
diff --git a/NAM/dsp.cpp b/NAM/dsp.cpp
index d068eed..68f5f37 100644
--- a/NAM/dsp.cpp
+++ b/NAM/dsp.cpp
@@ -167,7 +167,7 @@ void nam::Linear::process(NAM_SAMPLE* input, NAM_SAMPLE* output, const int num_f
 
 // NN modules =================================================================
 
-void nam::Conv1D::set_weights_(std::vector<float>::iterator& weights)
+void nam::Conv1D::set_weights_(std::vector<float>::const_iterator& weights)
 {
   if (this->_weight.size() > 0)
   {
@@ -198,7 +198,7 @@ void nam::Conv1D::set_size_(const int in_channels, const int out_channels, const
 }
 
 void nam::Conv1D::set_size_and_weights_(const int in_channels, const int out_channels, const int kernel_size,
-                                        const int _dilation, const bool do_bias, std::vector<float>::iterator& weights)
+                                        const int _dilation, const bool do_bias, std::vector<float>::const_iterator& weights)
 {
   this->set_size_(in_channels, out_channels, kernel_size, do_bias, _dilation);
   this->set_weights_(weights);
@@ -236,7 +236,7 @@ nam::Conv1x1::Conv1x1(const int in_channels, const int out_channels, const bool
     this->_bias.resize(out_channels);
 }
 
-void nam::Conv1x1::set_weights_(std::vector<float>::iterator& weights)
+void nam::Conv1x1::set_weights_(std::vector<float>::const_iterator& weights)
 {
   for (int i = 0; i < this->_weight.rows(); i++)
     for (int j = 0; j < this->_weight.cols(); j++)
diff --git a/NAM/dsp.h b/NAM/dsp.h
index c3e8ec9..737d6c3 100644
--- a/NAM/dsp.h
+++ b/NAM/dsp.h
@@ -124,11 +124,11 @@ class Conv1D
 {
 public:
   Conv1D() { this->_dilation = 1; };
-  void set_weights_(std::vector<float>::iterator& weights);
+  void set_weights_(std::vector<float>::const_iterator& weights);
   void set_size_(const int in_channels, const int out_channels, const int kernel_size, const bool do_bias,
                  const int _dilation);
   void set_size_and_weights_(const int in_channels, const int out_channels, const int kernel_size, const int _dilation,
-                             const bool do_bias, std::vector<float>::iterator& weights);
+                             const bool do_bias, std::vector<float>::const_iterator& weights);
   // Process from input to output
   //  Rightmost indices of input go from i_start to i_end,
   //  Indices on output for from j_start (to j_start + i_end - i_start)
@@ -153,7 +153,7 @@ class Conv1x1
 {
 public:
   Conv1x1(const int in_channels, const int out_channels, const bool _bias);
-  void set_weights_(std::vector<float>::iterator& weights);
+  void set_weights_(std::vector<float>::const_iterator& weights);
   // :param input: (N,Cin) or (Cin,)
   // :return: (N,Cout) or (Cout,), respectively
   Eigen::MatrixXf process(const Eigen::MatrixXf& input) const;
diff --git a/NAM/lstm.cpp b/NAM/lstm.cpp
index 38ff494..bde249f 100644
--- a/NAM/lstm.cpp
+++ b/NAM/lstm.cpp
@@ -4,7 +4,7 @@
 
 #include "lstm.h"
 
-nam::lstm::LSTMCell::LSTMCell(const int input_size, const int hidden_size, std::vector<float>::iterator& weights)
+nam::lstm::LSTMCell::LSTMCell(const int input_size, const int hidden_size, std::vector<float>::const_iterator& weights)
 {
   // Resize arrays
   this->_w.resize(4 * hidden_size, input_size + hidden_size);
@@ -63,12 +63,12 @@ void nam::lstm::LSTMCell::process_(const Eigen::VectorXf& x)
   }
 }
 
-nam::lstm::LSTM::LSTM(const int num_layers, const int input_size, const int hidden_size, std::vector<float>& weights,
+nam::lstm::LSTM::LSTM(const int num_layers, const int input_size, const int hidden_size, const std::vector<float>& weights,
                       const double expected_sample_rate)
 : DSP(expected_sample_rate)
 {
   this->_input.resize(1);
-  std::vector<float>::iterator it = weights.begin();
+  std::vector<float>::const_iterator it = weights.begin();
   for (int i = 0; i < num_layers; i++)
     this->_layers.push_back(LSTMCell(i == 0 ? input_size : hidden_size, hidden_size, it));
   this->_head_weight.resize(hidden_size);
diff --git a/NAM/lstm.h b/NAM/lstm.h
index 6b02b18..cdfada0 100644
--- a/NAM/lstm.h
+++ b/NAM/lstm.h
@@ -22,7 +22,7 @@ namespace lstm
 class LSTMCell
 {
 public:
-  LSTMCell(const int input_size, const int hidden_size, std::vector<float>::iterator& weights);
+  LSTMCell(const int input_size, const int hidden_size, std::vector<float>::const_iterator& weights);
   Eigen::VectorXf get_hidden_state() const { return this->_xh(Eigen::placeholders::lastN(this->_get_hidden_size())); };
   void process_(const Eigen::VectorXf& x);
 
@@ -50,7 +50,7 @@ class LSTMCell
 class LSTM : public DSP
 {
 public:
-  LSTM(const int num_layers, const int input_size, const int hidden_size, std::vector<float>& weights,
+  LSTM(const int num_layers, const int input_size, const int hidden_size, const std::vector<float>& weights,
        const double expected_sample_rate = -1.0);
   ~LSTM() = default;
 
diff --git a/NAM/wavenet.cpp b/NAM/wavenet.cpp
index fc96fa6..111dae0 100644
--- a/NAM/wavenet.cpp
+++ b/NAM/wavenet.cpp
@@ -12,7 +12,7 @@ nam::wavenet::_DilatedConv::_DilatedConv(const int in_channels, const int out_ch
   this->set_size_(in_channels, out_channels, kernel_size, bias, dilation);
 }
 
-void nam::wavenet::_Layer::set_weights_(std::vector<float>::iterator& weights)
+void nam::wavenet::_Layer::set_weights_(std::vector<float>::const_iterator& weights)
 {
   this->_conv.set_weights_(weights);
   this->_input_mixin.set_weights_(weights);
@@ -133,7 +133,7 @@ void nam::wavenet::_LayerArray::set_num_frames_(const long num_frames)
     this->_layers[i].set_num_frames_(num_frames);
 }
 
-void nam::wavenet::_LayerArray::set_weights_(std::vector<float>::iterator& weights)
+void nam::wavenet::_LayerArray::set_weights_(std::vector<float>::const_iterator& weights)
 {
   this->_rechannel.set_weights_(weights);
   for (size_t i = 0; i < this->_layers.size(); i++)
@@ -186,7 +186,7 @@ nam::wavenet::_Head::_Head(const int input_size, const int num_layers, const int
   }
 }
 
-void nam::wavenet::_Head::set_weights_(std::vector<float>::iterator& weights)
+void nam::wavenet::_Head::set_weights_(std::vector<float>::const_iterator& weights)
 {
   for (size_t i = 0; i < this->_layers.size(); i++)
     this->_layers[i].set_weights_(weights);
@@ -231,7 +231,7 @@ void nam::wavenet::_Head::_apply_activation_(Eigen::MatrixXf& x)
 // WaveNet ====================================================================
 
 nam::wavenet::WaveNet::WaveNet(const std::vector<nam::wavenet::LayerArrayParams>& layer_array_params,
-                               const float head_scale, const bool with_head, std::vector<float> weights,
+                               const float head_scale, const bool with_head, const std::vector<float>& weights,
                                const double expected_sample_rate)
 : DSP(expected_sample_rate)
 , _num_frames(0)
@@ -272,9 +272,9 @@ void nam::wavenet::WaveNet::finalize_(const int num_frames)
   this->_advance_buffers_(num_frames);
 }
 
-void nam::wavenet::WaveNet::set_weights_(std::vector<float>& weights)
+void nam::wavenet::WaveNet::set_weights_(const std::vector<float>& weights)
 {
-  std::vector<float>::iterator it = weights.begin();
+  std::vector<float>::const_iterator it = weights.begin();
   for (size_t i = 0; i < this->_layer_arrays.size(); i++)
     this->_layer_arrays[i].set_weights_(it);
   // this->_head.set_params_(it);
diff --git a/NAM/wavenet.h b/NAM/wavenet.h
index 7ea94f1..b4ab962 100644
--- a/NAM/wavenet.h
+++ b/NAM/wavenet.h
@@ -30,7 +30,7 @@ class _Layer
   , _1x1(channels, channels, true)
   , _activation(activations::Activation::get_activation(activation))
   , _gated(gated){};
-  void set_weights_(std::vector<float>::iterator& weights);
+  void set_weights_(std::vector<float>::const_iterator& weights);
   // :param `input`: from previous layer
   // :param `output`: to next layer
   void process_(const Eigen::MatrixXf& input, const Eigen::MatrixXf& condition, Eigen::MatrixXf& head_input,
@@ -108,7 +108,7 @@ class _LayerArray
                 Eigen::MatrixXf& head_outputs // post head-rechannel
   );
   void set_num_frames_(const long num_frames);
-  void set_weights_(std::vector<float>::iterator& it);
+  void set_weights_(std::vector<float>::const_iterator& it);
 
   // "Zero-indexed" receptive field.
   // E.g. a 1x1 convolution has a z.i.r.f. of zero.
@@ -144,7 +144,7 @@ class _Head
 {
 public:
   _Head(const int input_size, const int num_layers, const int channels, const std::string activation);
-  void set_weights_(std::vector<float>::iterator& weights);
+  void set_weights_(std::vector<float>::const_iterator& weights);
   // NOTE: the head transforms the provided input by applying a nonlinearity
   // to it in-place!
   void process_(Eigen::MatrixXf& inputs, Eigen::MatrixXf& outputs);
@@ -169,11 +169,11 @@ class WaveNet : public DSP
 {
 public:
   WaveNet(const std::vector<LayerArrayParams>& layer_array_params, const float head_scale, const bool with_head,
-          std::vector<float> weights, const double expected_sample_rate = -1.0);
+          const std::vector<float>& weights, const double expected_sample_rate = -1.0);
   ~WaveNet() = default;
 
   void finalize_(const int num_frames) override;
-  void set_weights_(std::vector<float>& weights);
+  void set_weights_(const std::vector<float>& weights);
 
 private:
   long _num_frames;

From 0833ea6760c70c9fde8f48c47d370d6222c8269a Mon Sep 17 00:00:00 2001
From: Oli Larkin <olilarkin@googlemail.com>
Date: Sat, 30 Dec 2023 21:11:29 +0100
Subject: [PATCH 2/3] Introduce weights_it alias

---
 NAM/convnet.cpp | 8 ++++----
 NAM/convnet.h   | 6 +++---
 NAM/dsp.cpp     | 6 +++---
 NAM/dsp.h       | 8 +++++---
 NAM/lstm.cpp    | 5 +++--
 NAM/lstm.h      | 2 +-
 NAM/wavenet.cpp | 8 ++++----
 NAM/wavenet.h   | 6 +++---
 8 files changed, 26 insertions(+), 23 deletions(-)

diff --git a/NAM/convnet.cpp b/NAM/convnet.cpp
index bc0a392..937bf0b 100644
--- a/NAM/convnet.cpp
+++ b/NAM/convnet.cpp
@@ -10,7 +10,7 @@
 #include "dsp.h"
 #include "convnet.h"
 
-nam::convnet::BatchNorm::BatchNorm(const int dim, std::vector<float>::const_iterator& weights)
+nam::convnet::BatchNorm::BatchNorm(const int dim, weights_it& weights)
 {
   // Extract from param buffer
   Eigen::VectorXf running_mean(dim);
@@ -48,7 +48,7 @@ void nam::convnet::BatchNorm::process_(Eigen::MatrixXf& x, const long i_start, c
 
 void nam::convnet::ConvNetBlock::set_weights_(const int in_channels, const int out_channels, const int _dilation,
                                               const bool batchnorm, const std::string activation,
-                                              std::vector<float>::const_iterator& weights)
+                                              weights_it& weights)
 {
   this->_batchnorm = batchnorm;
   // HACK 2 kernel
@@ -74,7 +74,7 @@ long nam::convnet::ConvNetBlock::get_out_channels() const
   return this->conv.get_out_channels();
 }
 
-nam::convnet::_Head::_Head(const int channels, std::vector<float>::const_iterator& weights)
+nam::convnet::_Head::_Head(const int channels, weights_it& weights)
 {
   this->_weight.resize(channels);
   for (int i = 0; i < channels; i++)
@@ -98,7 +98,7 @@ nam::convnet::ConvNet::ConvNet(const int channels, const std::vector<int>& dilat
 {
   this->_verify_weights(channels, dilations, batchnorm, weights.size());
   this->_blocks.resize(dilations.size());
-  std::vector<float>::const_iterator it = weights.begin();
+  weights_it it = weights.begin();
   for (size_t i = 0; i < dilations.size(); i++)
     this->_blocks[i].set_weights_(i == 0 ? 1 : channels, channels, dilations[i], batchnorm, activation, it);
   this->_block_vals.resize(this->_blocks.size() + 1);
diff --git a/NAM/convnet.h b/NAM/convnet.h
index a410506..02af4c8 100644
--- a/NAM/convnet.h
+++ b/NAM/convnet.h
@@ -23,7 +23,7 @@ class BatchNorm
 {
 public:
   BatchNorm(){};
-  BatchNorm(const int dim, std::vector<float>::const_iterator& weights);
+  BatchNorm(const int dim, weights_it& weights);
   void process_(Eigen::MatrixXf& input, const long i_start, const long i_end) const;
 
 private:
@@ -41,7 +41,7 @@ class ConvNetBlock
 public:
   ConvNetBlock(){};
   void set_weights_(const int in_channels, const int out_channels, const int _dilation, const bool batchnorm,
-                    const std::string activation, std::vector<float>::const_iterator& weights);
+                    const std::string activation, weights_it& weights);
   void process_(const Eigen::MatrixXf& input, Eigen::MatrixXf& output, const long i_start, const long i_end) const;
   long get_out_channels() const;
   Conv1D conv;
@@ -56,7 +56,7 @@ class _Head
 {
 public:
   _Head(){};
-  _Head(const int channels, std::vector<float>::const_iterator& weights);
+  _Head(const int channels, weights_it& weights);
   void process_(const Eigen::MatrixXf& input, Eigen::VectorXf& output, const long i_start, const long i_end) const;
 
 private:
diff --git a/NAM/dsp.cpp b/NAM/dsp.cpp
index 68f5f37..db0730b 100644
--- a/NAM/dsp.cpp
+++ b/NAM/dsp.cpp
@@ -167,7 +167,7 @@ void nam::Linear::process(NAM_SAMPLE* input, NAM_SAMPLE* output, const int num_f
 
 // NN modules =================================================================
 
-void nam::Conv1D::set_weights_(std::vector<float>::const_iterator& weights)
+void nam::Conv1D::set_weights_(weights_it& weights)
 {
   if (this->_weight.size() > 0)
   {
@@ -198,7 +198,7 @@ void nam::Conv1D::set_size_(const int in_channels, const int out_channels, const
 }
 
 void nam::Conv1D::set_size_and_weights_(const int in_channels, const int out_channels, const int kernel_size,
-                                        const int _dilation, const bool do_bias, std::vector<float>::const_iterator& weights)
+                                        const int _dilation, const bool do_bias, weights_it& weights)
 {
   this->set_size_(in_channels, out_channels, kernel_size, do_bias, _dilation);
   this->set_weights_(weights);
@@ -236,7 +236,7 @@ nam::Conv1x1::Conv1x1(const int in_channels, const int out_channels, const bool
     this->_bias.resize(out_channels);
 }
 
-void nam::Conv1x1::set_weights_(std::vector<float>::const_iterator& weights)
+void nam::Conv1x1::set_weights_(weights_it& weights)
 {
   for (int i = 0; i < this->_weight.rows(); i++)
     for (int j = 0; j < this->_weight.cols(); j++)
diff --git a/NAM/dsp.h b/NAM/dsp.h
index 737d6c3..e74a153 100644
--- a/NAM/dsp.h
+++ b/NAM/dsp.h
@@ -23,6 +23,8 @@
 
 namespace nam
 {
+using weights_it = std::vector<float>::const_iterator;
+
 enum EArchitectures
 {
   kLinear = 0,
@@ -124,11 +126,11 @@ class Conv1D
 {
 public:
   Conv1D() { this->_dilation = 1; };
-  void set_weights_(std::vector<float>::const_iterator& weights);
+  void set_weights_(weights_it& weights);
   void set_size_(const int in_channels, const int out_channels, const int kernel_size, const bool do_bias,
                  const int _dilation);
   void set_size_and_weights_(const int in_channels, const int out_channels, const int kernel_size, const int _dilation,
-                             const bool do_bias, std::vector<float>::const_iterator& weights);
+                             const bool do_bias, weights_it& weights);
   // Process from input to output
   //  Rightmost indices of input go from i_start to i_end,
   //  Indices on output for from j_start (to j_start + i_end - i_start)
@@ -153,7 +155,7 @@ class Conv1x1
 {
 public:
   Conv1x1(const int in_channels, const int out_channels, const bool _bias);
-  void set_weights_(std::vector<float>::const_iterator& weights);
+  void set_weights_(weights_it& weights);
   // :param input: (N,Cin) or (Cin,)
   // :return: (N,Cout) or (Cout,), respectively
   Eigen::MatrixXf process(const Eigen::MatrixXf& input) const;
diff --git a/NAM/lstm.cpp b/NAM/lstm.cpp
index bde249f..3b55575 100644
--- a/NAM/lstm.cpp
+++ b/NAM/lstm.cpp
@@ -2,9 +2,10 @@
 #include <string>
 #include <vector>
 
+#include "dsp.h"
 #include "lstm.h"
 
-nam::lstm::LSTMCell::LSTMCell(const int input_size, const int hidden_size, std::vector<float>::const_iterator& weights)
+nam::lstm::LSTMCell::LSTMCell(const int input_size, const int hidden_size, weights_it& weights)
 {
   // Resize arrays
   this->_w.resize(4 * hidden_size, input_size + hidden_size);
@@ -68,7 +69,7 @@ nam::lstm::LSTM::LSTM(const int num_layers, const int input_size, const int hidd
 : DSP(expected_sample_rate)
 {
   this->_input.resize(1);
-  std::vector<float>::const_iterator it = weights.begin();
+  auto it = weights.begin();
   for (int i = 0; i < num_layers; i++)
     this->_layers.push_back(LSTMCell(i == 0 ? input_size : hidden_size, hidden_size, it));
   this->_head_weight.resize(hidden_size);
diff --git a/NAM/lstm.h b/NAM/lstm.h
index cdfada0..a432169 100644
--- a/NAM/lstm.h
+++ b/NAM/lstm.h
@@ -22,7 +22,7 @@ namespace lstm
 class LSTMCell
 {
 public:
-  LSTMCell(const int input_size, const int hidden_size, std::vector<float>::const_iterator& weights);
+  LSTMCell(const int input_size, const int hidden_size, weights_it& weights);
   Eigen::VectorXf get_hidden_state() const { return this->_xh(Eigen::placeholders::lastN(this->_get_hidden_size())); };
   void process_(const Eigen::VectorXf& x);
 
diff --git a/NAM/wavenet.cpp b/NAM/wavenet.cpp
index 111dae0..42dd299 100644
--- a/NAM/wavenet.cpp
+++ b/NAM/wavenet.cpp
@@ -12,7 +12,7 @@ nam::wavenet::_DilatedConv::_DilatedConv(const int in_channels, const int out_ch
   this->set_size_(in_channels, out_channels, kernel_size, bias, dilation);
 }
 
-void nam::wavenet::_Layer::set_weights_(std::vector<float>::const_iterator& weights)
+void nam::wavenet::_Layer::set_weights_(weights_it& weights)
 {
   this->_conv.set_weights_(weights);
   this->_input_mixin.set_weights_(weights);
@@ -133,7 +133,7 @@ void nam::wavenet::_LayerArray::set_num_frames_(const long num_frames)
     this->_layers[i].set_num_frames_(num_frames);
 }
 
-void nam::wavenet::_LayerArray::set_weights_(std::vector<float>::const_iterator& weights)
+void nam::wavenet::_LayerArray::set_weights_(weights_it& weights)
 {
   this->_rechannel.set_weights_(weights);
   for (size_t i = 0; i < this->_layers.size(); i++)
@@ -186,7 +186,7 @@ nam::wavenet::_Head::_Head(const int input_size, const int num_layers, const int
   }
 }
 
-void nam::wavenet::_Head::set_weights_(std::vector<float>::const_iterator& weights)
+void nam::wavenet::_Head::set_weights_(weights_it& weights)
 {
   for (size_t i = 0; i < this->_layers.size(); i++)
     this->_layers[i].set_weights_(weights);
@@ -274,7 +274,7 @@ void nam::wavenet::WaveNet::finalize_(const int num_frames)
 
 void nam::wavenet::WaveNet::set_weights_(const std::vector<float>& weights)
 {
-  std::vector<float>::const_iterator it = weights.begin();
+  weights_it it = weights.begin();
   for (size_t i = 0; i < this->_layer_arrays.size(); i++)
     this->_layer_arrays[i].set_weights_(it);
   // this->_head.set_params_(it);
diff --git a/NAM/wavenet.h b/NAM/wavenet.h
index b4ab962..bd30f03 100644
--- a/NAM/wavenet.h
+++ b/NAM/wavenet.h
@@ -30,7 +30,7 @@ class _Layer
   , _1x1(channels, channels, true)
   , _activation(activations::Activation::get_activation(activation))
   , _gated(gated){};
-  void set_weights_(std::vector<float>::const_iterator& weights);
+  void set_weights_(weights_it& weights);
   // :param `input`: from previous layer
   // :param `output`: to next layer
   void process_(const Eigen::MatrixXf& input, const Eigen::MatrixXf& condition, Eigen::MatrixXf& head_input,
@@ -108,7 +108,7 @@ class _LayerArray
                 Eigen::MatrixXf& head_outputs // post head-rechannel
   );
   void set_num_frames_(const long num_frames);
-  void set_weights_(std::vector<float>::const_iterator& it);
+  void set_weights_(weights_it& it);
 
   // "Zero-indexed" receptive field.
   // E.g. a 1x1 convolution has a z.i.r.f. of zero.
@@ -144,7 +144,7 @@ class _Head
 {
 public:
   _Head(const int input_size, const int num_layers, const int channels, const std::string activation);
-  void set_weights_(std::vector<float>::const_iterator& weights);
+  void set_weights_(weights_it& weights);
   // NOTE: the head transforms the provided input by applying a nonlinearity
   // to it in-place!
   void process_(Eigen::MatrixXf& inputs, Eigen::MatrixXf& outputs);

From 78f42d048aaa9757616792427ba8cdeb712584e7 Mon Sep 17 00:00:00 2001
From: Oli Larkin <olilarkin@googlemail.com>
Date: Sat, 30 Dec 2023 21:55:33 +0100
Subject: [PATCH 3/3] Use Eigen::Ref<> rather than normal references

https://eigen.tuxfamily.org/dox/TopicFunctionTakingEigenTypes.html#TopicUsingRefClass
---
 NAM/activations.h |  2 +-
 NAM/convnet.cpp   | 10 +++++-----
 NAM/convnet.h     |  6 +++---
 NAM/dsp.cpp       |  8 ++++----
 NAM/dsp.h         |  4 ++--
 NAM/lstm.cpp      |  2 +-
 NAM/wavenet.cpp   | 30 ++++++++++++++++++++----------
 NAM/wavenet.h     | 18 +++++++++---------
 8 files changed, 45 insertions(+), 35 deletions(-)

diff --git a/NAM/activations.h b/NAM/activations.h
index e9afc33..51b03d9 100644
--- a/NAM/activations.h
+++ b/NAM/activations.h
@@ -45,7 +45,7 @@ class Activation
 public:
   Activation() = default;
   virtual ~Activation() = default;
-  virtual void apply(Eigen::MatrixXf& matrix) { apply(matrix.data(), matrix.rows() * matrix.cols()); }
+  virtual void apply(Eigen::Ref<Eigen::MatrixXf> matrix) { apply(matrix.data(), matrix.rows() * matrix.cols()); }
   virtual void apply(Eigen::Block<Eigen::MatrixXf> block) { apply(block.data(), block.rows() * block.cols()); }
   virtual void apply(Eigen::Block<Eigen::MatrixXf, -1, -1, true> block)
   {
diff --git a/NAM/convnet.cpp b/NAM/convnet.cpp
index 937bf0b..1122871 100644
--- a/NAM/convnet.cpp
+++ b/NAM/convnet.cpp
@@ -35,7 +35,7 @@ nam::convnet::BatchNorm::BatchNorm(const int dim, weights_it& weights)
   this->loc = _bias - this->scale.cwiseProduct(running_mean);
 }
 
-void nam::convnet::BatchNorm::process_(Eigen::MatrixXf& x, const long i_start, const long i_end) const
+void nam::convnet::BatchNorm::process_(Eigen::Ref<Eigen::MatrixXf> x, const long i_start, const long i_end) const
 {
   // todo using colwise?
   // #speed but conv probably dominates
@@ -58,7 +58,7 @@ void nam::convnet::ConvNetBlock::set_weights_(const int in_channels, const int o
   this->activation = activations::Activation::get_activation(activation);
 }
 
-void nam::convnet::ConvNetBlock::process_(const Eigen::MatrixXf& input, Eigen::MatrixXf& output, const long i_start,
+void nam::convnet::ConvNetBlock::process_(const Eigen::Ref<const Eigen::MatrixXf> input, Eigen::Ref<Eigen::MatrixXf> output, const long i_start,
                                           const long i_end) const
 {
   const long ncols = i_end - i_start;
@@ -82,7 +82,7 @@ nam::convnet::_Head::_Head(const int channels, weights_it& weights)
   this->_bias = *(weights++);
 }
 
-void nam::convnet::_Head::process_(const Eigen::MatrixXf& input, Eigen::VectorXf& output, const long i_start,
+void nam::convnet::_Head::process_(const Eigen::Ref<const Eigen::MatrixXf> input, Eigen::VectorXf& output, const long i_start,
                                    const long i_end) const
 {
   const long length = i_end - i_start;
@@ -146,7 +146,7 @@ void nam::convnet::ConvNet::_update_buffers_(NAM_SAMPLE* input, const int num_fr
 
   const size_t buffer_size = this->_input_buffer.size();
 
-  if (this->_block_vals[0].rows() != 1 || this->_block_vals[0].cols() != buffer_size)
+  if (this->_block_vals[0].rows() != Eigen::Index(1) || this->_block_vals[0].cols() != Eigen::Index(buffer_size))
   {
     this->_block_vals[0].resize(1, buffer_size);
     this->_block_vals[0].setZero();
@@ -155,7 +155,7 @@ void nam::convnet::ConvNet::_update_buffers_(NAM_SAMPLE* input, const int num_fr
   for (size_t i = 1; i < this->_block_vals.size(); i++)
   {
     if (this->_block_vals[i].rows() == this->_blocks[i - 1].get_out_channels()
-        && this->_block_vals[i].cols() == buffer_size)
+        && this->_block_vals[i].cols() == Eigen::Index(buffer_size))
       continue; // Already has correct size
     this->_block_vals[i].resize(this->_blocks[i - 1].get_out_channels(), buffer_size);
     this->_block_vals[i].setZero();
diff --git a/NAM/convnet.h b/NAM/convnet.h
index 02af4c8..a117b43 100644
--- a/NAM/convnet.h
+++ b/NAM/convnet.h
@@ -24,7 +24,7 @@ class BatchNorm
 public:
   BatchNorm(){};
   BatchNorm(const int dim, weights_it& weights);
-  void process_(Eigen::MatrixXf& input, const long i_start, const long i_end) const;
+  void process_(Eigen::Ref<Eigen::MatrixXf> input, const long i_start, const long i_end) const;
 
 private:
   // TODO simplify to just ax+b
@@ -42,7 +42,7 @@ class ConvNetBlock
   ConvNetBlock(){};
   void set_weights_(const int in_channels, const int out_channels, const int _dilation, const bool batchnorm,
                     const std::string activation, weights_it& weights);
-  void process_(const Eigen::MatrixXf& input, Eigen::MatrixXf& output, const long i_start, const long i_end) const;
+  void process_(const Eigen::Ref<const Eigen::MatrixXf> input, Eigen::Ref<Eigen::MatrixXf> output, const long i_start, const long i_end) const;
   long get_out_channels() const;
   Conv1D conv;
 
@@ -57,7 +57,7 @@ class _Head
 public:
   _Head(){};
   _Head(const int channels, weights_it& weights);
-  void process_(const Eigen::MatrixXf& input, Eigen::VectorXf& output, const long i_start, const long i_end) const;
+  void process_(const Eigen::Ref<const Eigen::MatrixXf> input, Eigen::VectorXf& output, const long i_start, const long i_end) const;
 
 private:
   Eigen::VectorXf _weight;
diff --git a/NAM/dsp.cpp b/NAM/dsp.cpp
index db0730b..9b27ec9 100644
--- a/NAM/dsp.cpp
+++ b/NAM/dsp.cpp
@@ -39,7 +39,7 @@ void nam::DSP::prewarm()
 void nam::DSP::process(NAM_SAMPLE* input, NAM_SAMPLE* output, const int num_frames)
 {
   // Default implementation is the null operation
-  for (size_t i = 0; i < num_frames; i++)
+  for (auto i = 0; i < num_frames; i++)
     output[i] = input[i];
 }
 
@@ -157,7 +157,7 @@ void nam::Linear::process(NAM_SAMPLE* input, NAM_SAMPLE* output, const int num_f
   this->nam::Buffer::_update_buffers_(input, num_frames);
 
   // Main computation!
-  for (size_t i = 0; i < num_frames; i++)
+  for (auto i = 0; i < num_frames; i++)
   {
     const size_t offset = this->_input_buffer_offset - this->_weight.size() + i + 1;
     auto input = Eigen::Map<const Eigen::VectorXf>(&this->_input_buffer[offset], this->_receptive_field);
@@ -204,7 +204,7 @@ void nam::Conv1D::set_size_and_weights_(const int in_channels, const int out_cha
   this->set_weights_(weights);
 }
 
-void nam::Conv1D::process_(const Eigen::MatrixXf& input, Eigen::MatrixXf& output, const long i_start, const long ncols,
+void nam::Conv1D::process_(const Eigen::Ref<const Eigen::MatrixXf> input, Eigen::Ref<Eigen::MatrixXf> output, const long i_start, const long ncols,
                            const long j_start) const
 {
   // This is the clever part ;)
@@ -246,7 +246,7 @@ void nam::Conv1x1::set_weights_(weights_it& weights)
       this->_bias(i) = *(weights++);
 }
 
-Eigen::MatrixXf nam::Conv1x1::process(const Eigen::MatrixXf& input) const
+Eigen::MatrixXf nam::Conv1x1::process(const Eigen::Ref<const Eigen::MatrixXf> input) const
 {
   if (this->_do_bias)
     return (this->_weight * input).colwise() + this->_bias;
diff --git a/NAM/dsp.h b/NAM/dsp.h
index e74a153..118e782 100644
--- a/NAM/dsp.h
+++ b/NAM/dsp.h
@@ -134,7 +134,7 @@ class Conv1D
   // Process from input to output
   //  Rightmost indices of input go from i_start to i_end,
   //  Indices on output for from j_start (to j_start + i_end - i_start)
-  void process_(const Eigen::MatrixXf& input, Eigen::MatrixXf& output, const long i_start, const long i_end,
+  void process_(const Eigen::Ref<const Eigen::MatrixXf> input, Eigen::Ref<Eigen::MatrixXf> output, const long i_start, const long i_end,
                 const long j_start) const;
   long get_in_channels() const { return this->_weight.size() > 0 ? this->_weight[0].cols() : 0; };
   long get_kernel_size() const { return this->_weight.size(); };
@@ -158,7 +158,7 @@ class Conv1x1
   void set_weights_(weights_it& weights);
   // :param input: (N,Cin) or (Cin,)
   // :return: (N,Cout) or (Cout,), respectively
-  Eigen::MatrixXf process(const Eigen::MatrixXf& input) const;
+  Eigen::MatrixXf process(const Eigen::Ref<const Eigen::MatrixXf> input) const;
 
   long get_out_channels() const { return this->_weight.rows(); };
 
diff --git a/NAM/lstm.cpp b/NAM/lstm.cpp
index 3b55575..aa0f6ad 100644
--- a/NAM/lstm.cpp
+++ b/NAM/lstm.cpp
@@ -81,7 +81,7 @@ nam::lstm::LSTM::LSTM(const int num_layers, const int input_size, const int hidd
 
 void nam::lstm::LSTM::process(NAM_SAMPLE* input, NAM_SAMPLE* output, const int num_frames)
 {
-  for (size_t i = 0; i < num_frames; i++)
+  for (auto i = 0; i < num_frames; i++)
     output[i] = this->_process_sample(input[i]);
 }
 
diff --git a/NAM/wavenet.cpp b/NAM/wavenet.cpp
index 42dd299..7899c3b 100644
--- a/NAM/wavenet.cpp
+++ b/NAM/wavenet.cpp
@@ -19,8 +19,8 @@ void nam::wavenet::_Layer::set_weights_(weights_it& weights)
   this->_1x1.set_weights_(weights);
 }
 
-void nam::wavenet::_Layer::process_(const Eigen::MatrixXf& input, const Eigen::MatrixXf& condition,
-                                    Eigen::MatrixXf& head_input, Eigen::MatrixXf& output, const long i_start,
+void nam::wavenet::_Layer::process_(const Eigen::Ref<const Eigen::MatrixXf> input, const Eigen::Ref<const Eigen::MatrixXf> condition,
+                                    Eigen::Ref<Eigen::MatrixXf> head_input, Eigen::Ref<Eigen::MatrixXf> output, const long i_start,
                                     const long j_start)
 {
   const long ncols = condition.cols();
@@ -102,17 +102,27 @@ void nam::wavenet::_LayerArray::prepare_for_frames_(const long num_frames)
     this->_rewind_buffers_();
 }
 
-void nam::wavenet::_LayerArray::process_(const Eigen::MatrixXf& layer_inputs, const Eigen::MatrixXf& condition,
-                                         Eigen::MatrixXf& head_inputs, Eigen::MatrixXf& layer_outputs,
-                                         Eigen::MatrixXf& head_outputs)
+void nam::wavenet::_LayerArray::process_(const Eigen::Ref<const Eigen::MatrixXf> layer_inputs, const Eigen::Ref<const Eigen::MatrixXf> condition,
+                                         Eigen::Ref<Eigen::MatrixXf> head_inputs, Eigen::Ref<Eigen::MatrixXf> layer_outputs,
+                                         Eigen::Ref<Eigen::MatrixXf> head_outputs)
 {
   this->_layer_buffers[0].middleCols(this->_buffer_start, layer_inputs.cols()) = this->_rechannel.process(layer_inputs);
   const size_t last_layer = this->_layers.size() - 1;
   for (size_t i = 0; i < this->_layers.size(); i++)
   {
-    this->_layers[i].process_(this->_layer_buffers[i], condition, head_inputs,
-                              i == last_layer ? layer_outputs : this->_layer_buffers[i + 1], this->_buffer_start,
-                              i == last_layer ? 0 : this->_buffer_start);
+    if (i == last_layer)
+    {
+      this->_layers[i].process_(this->_layer_buffers[i], condition, head_inputs,
+                                layer_outputs, this->_buffer_start,
+                                0);
+    }
+    else
+    {
+      this->_layers[i].process_(this->_layer_buffers[i], condition, head_inputs,
+                                this->_layer_buffers[i + 1], this->_buffer_start,
+                                this->_buffer_start);
+    }
+
   }
   head_outputs = this->_head_rechannel.process(head_inputs);
 }
@@ -192,7 +202,7 @@ void nam::wavenet::_Head::set_weights_(weights_it& weights)
     this->_layers[i].set_weights_(weights);
 }
 
-void nam::wavenet::_Head::process_(Eigen::MatrixXf& inputs, Eigen::MatrixXf& outputs)
+void nam::wavenet::_Head::process_(Eigen::Ref<Eigen::MatrixXf> inputs, Eigen::Ref<Eigen::MatrixXf> outputs)
 {
   const size_t num_layers = this->_layers.size();
   this->_apply_activation_(inputs);
@@ -223,7 +233,7 @@ void nam::wavenet::_Head::set_num_frames_(const long num_frames)
   }
 }
 
-void nam::wavenet::_Head::_apply_activation_(Eigen::MatrixXf& x)
+void nam::wavenet::_Head::_apply_activation_(Eigen::Ref<Eigen::MatrixXf> x)
 {
   this->_activation->apply(x);
 }
diff --git a/NAM/wavenet.h b/NAM/wavenet.h
index bd30f03..fbaab54 100644
--- a/NAM/wavenet.h
+++ b/NAM/wavenet.h
@@ -33,8 +33,8 @@ class _Layer
   void set_weights_(weights_it& weights);
   // :param `input`: from previous layer
   // :param `output`: to next layer
-  void process_(const Eigen::MatrixXf& input, const Eigen::MatrixXf& condition, Eigen::MatrixXf& head_input,
-                Eigen::MatrixXf& output, const long i_start, const long j_start);
+  void process_(const Eigen::Ref<const Eigen::MatrixXf> input, const Eigen::Ref<const Eigen::MatrixXf> condition, Eigen::Ref<Eigen::MatrixXf> head_input,
+                Eigen::Ref<Eigen::MatrixXf> output, const long i_start, const long j_start);
   void set_num_frames_(const long num_frames);
   long get_channels() const { return this->_conv.get_in_channels(); };
   int get_dilation() const { return this->_conv.get_dilation(); };
@@ -101,11 +101,11 @@ class _LayerArray
   void prepare_for_frames_(const long num_frames);
 
   // All arrays are "short".
-  void process_(const Eigen::MatrixXf& layer_inputs, // Short
-                const Eigen::MatrixXf& condition, // Short
-                Eigen::MatrixXf& layer_outputs, // Short
-                Eigen::MatrixXf& head_inputs, // Sum up on this.
-                Eigen::MatrixXf& head_outputs // post head-rechannel
+  void process_(const Eigen::Ref<const Eigen::MatrixXf> layer_inputs, // Short
+                const Eigen::Ref<const Eigen::MatrixXf> condition, // Short
+                Eigen::Ref<Eigen::MatrixXf> layer_outputs, // Short
+                Eigen::Ref<Eigen::MatrixXf> head_inputs, // Sum up on this.
+                Eigen::Ref<Eigen::MatrixXf> head_outputs // post head-rechannel
   );
   void set_num_frames_(const long num_frames);
   void set_weights_(weights_it& it);
@@ -147,7 +147,7 @@ class _Head
   void set_weights_(weights_it& weights);
   // NOTE: the head transforms the provided input by applying a nonlinearity
   // to it in-place!
-  void process_(Eigen::MatrixXf& inputs, Eigen::MatrixXf& outputs);
+  void process_(Eigen::Ref<Eigen::MatrixXf> inputs, Eigen::Ref<Eigen::MatrixXf> outputs);
   void set_num_frames_(const long num_frames);
 
 private:
@@ -161,7 +161,7 @@ class _Head
   std::vector<Eigen::MatrixXf> _buffers;
 
   // Apply the activation to the provided array, in-place
-  void _apply_activation_(Eigen::MatrixXf& x);
+  void _apply_activation_(Eigen::Ref<Eigen::MatrixXf> x);
 };
 
 // The main WaveNet model