jacobian

commit 52c389b017298f07a363362afd73ecf54be816af
parent 3902178663cfd0c5de3727adad3a417370099217
Author: David Freifeld <freifeld.david@gmail.com>
Date:   Sat, 27 Mar 2021 23:46:04 -0700

Futher progress with Python binds

Diffstat:
M
src/bpnn.hpp
 | 
6
++----
M
src/pybind.cpp
 | 
60
++++++++++++++++++++++++++++++++++++------------------------

2 files changed, 38 insertions(+), 28 deletions(-)
diff --git a/src/bpnn.hpp b/src/bpnn.hpp
@@ -53,6 +53,7 @@ protected:
     std::function<void(void)> decay;
     std::function<void(std::vector<Eigen::MatrixXf>, int, int)> grad_calc;
     std::function<void(std::vector<Eigen::MatrixXf>, int)> update;
+	void next_batch(int fd);
 public:
     int data;
     int val_data;
@@ -77,8 +78,6 @@ public:
             float l, float ratio, bool early_exit=true, float cutoff=0);
     ~Network();
     void add_layer(int nodes, const char* name, std::function<float(float)> activation, std::function<float(float)> activation_deriv);
-    void init_decay(const char* type, ...);
-    void init_optimizer(const char* name, ...);
     void initialize();
     void set_activation(int index, std::function<float(float)> custom, std::function<float(float)> custom_deriv);
     void feedforward();
@@ -86,8 +85,7 @@ public:
     void list_net();
     float cost();
     float accuracy();
-    Eigen::MatrixXf backpropagate();
-    void next_batch(int fd);
+    Eigen::MatrixXf backpropagate();    
     void validate(const char* path);
     void train();
     float get_acc() {return epoch_acc;}
diff --git a/src/pybind.cpp b/src/pybind.cpp
@@ -8,30 +8,42 @@
 #include <pybind11/pybind11.h>
 #include <pybind11/functional.h>
 #include "bpnn.hpp"
+#include "utils.hpp"
 namespace py = pybind11;
 
-PYBIND11_MODULE(_jacobian, m) {
-  m.doc() = "Fast machine learning in C++"; // optional module docstring
-  py::enum_<Regularization>(m, "Regularization")
-	  .value("L1", Regularization::L1)
-	  .value("L2", Regularization::L2)	  
-	  .export_values();
-  py::class_<Network>(m, "Network")
-	  .def(py::init<char*, int, float, float, Regularization, float, float, bool, float>())
-	  .def("add_layer", &Network::add_layer, py::arg("nodes"), py::arg("name"), py::arg("activation"), py::arg("activation_deriv"))
-	  .def("initialize", &Network::initialize)
-      //.def("init_decay", &Network::init_decay, py::arg("type"), py::arg("a_0"), py::arg("k"))
-      //.def("set_activation", &Network::set_activation)
-	  .def("feedforward", &Network::feedforward)
-	  .def("backpropagate", &Network::backpropagate)
-	  .def("list_net", &Network::list_net)
-	  .def("cost", &Network::cost)
-	  .def("accuracy", &Network::accuracy)
-      //.def("update_layer", &Network::update_layer, py::arg("vals"), py::arg("len"), py::arg("index"))
-	  .def("next_batch", &Network::next_batch)
-	  .def("train", &Network::train)
-	  .def("get_acc", &Network::get_acc)
-	  .def("get_cost", &Network::get_cost)
-	  .def("get_val_acc", &Network::get_val_acc)
-	  .def("get_val_cost", &Network::get_val_cost);
+PYBIND11_MODULE(_jacobian, m)
+{
+	m.doc() = "Fast machine learning in C++"; // optional module docstring
+	py::enum_<Regularization>(m, "Regularization")
+		.value("L1", Regularization::L1)
+		.value("L2", Regularization::L2)
+		.export_values();
+	py::class_<Network>(m, "Network")
+		.def(py::init<char *, int, float, float, Regularization, float,
+			      float, bool, float>(),
+		     py::arg("path"), py::arg("batch"), py::arg("learn_rate"),
+		     py::arg("bias_rate"), py::arg("regularization"),
+		     py::arg("lambda"), py::arg("ratio"),
+		     py::arg("early_exit") = true, py::arg("cutoff") = 0)
+		.def("add_layer", &Network::add_layer, py::arg("nodes"),
+		     py::arg("name"), py::arg("activation"),
+		     py::arg("activation_deriv"))
+		.def("initialize", &Network::initialize)
+		.def("set_activation", &Network::set_activation,
+		     py::arg("index"), py::arg("custom"),
+		     py::arg("custom_deriv"))
+		.def("feedforward", &Network::feedforward)
+		.def("backpropagate", &Network::backpropagate)
+		.def("list_net", &Network::list_net)
+		.def("cost", &Network::cost)
+		.def("accuracy", &Network::accuracy)
+		.def("train", &Network::train)
+		.def("get_acc", &Network::get_acc)
+		.def("get_cost", &Network::get_cost)
+		.def("get_val_acc", &Network::get_val_acc)
+		.def("get_val_cost", &Network::get_val_cost);
+	m.def("linear", &linear, py::arg("x"));
+	m.def("linear_deriv", &linear_deriv, py::arg("x"));
+	m.def("sigmoid", &sigmoid, py::arg("x"));
+	m.def("sigmoid_deriv", &sigmoid_deriv, py::arg("x"));
 }