update readme

cpp/README.md

+# C++ part
+
+## Description
+In this part the network is read from json, pruned, and then written back. The paths need to be adjusted for that in main.cpp.
+
+
+## Setup
+The dependencies for the interval_network library are boost, Eigen3, nlohmann_json and NAG::dco_cpp. To install the first three ones (sudo previledges required):
+
+    sudo apt-get install build-essential libboost-all-dev libeigen3-dev nlohmann-json3-dev
+
+To set up the interval_network library, create a folder *extern* and clone the repository into it
+
+    mkdir extern && cd extern && git clone https://git.rwth-aachen.de/sisc_lab_2022/interval_network.git
+
+The dco library should also go into the extern folder. dco will be found if it is in the extern folder and is called dco (without the version number). Alternatively, dco can also be installed properly on the computer. In this case it will be located outomatically
\ No newline at end of file

cpp/main.cpp

@@ -31,7 +31,7 @@ int main(int, char**)
     using dco_ttype = dco_mode::tape_t;
 
     // original network
-    std::string networkPath = "/home/jonas/siscLab/example/python/simple_network.json";
+    std::string networkPath = "/path/to/simple_network.json";
 
     std::cout << "read json file" << std::endl;
     std::ifstream f(networkPath);
@@ -103,7 +103,7 @@ int main(int, char**)
 
     // save network
     auto js = network.to_json();
-    std::ofstream o("/home/jonas/siscLab/example/python/pruned_network.json");
+    std::ofstream o("/path/to/pruned_network.json");
     o << std::setw(2) << js << std::endl;
     o.close();
 }

python/README.md

+# Python part
+
+## Description
+the file train_simple_network.py is meant to serve as an example for training a network and the converting it to a json file which can then be further processed from the c++ side. The file network_evaluation.py can be used to check the accuracy of the pruned network
+
+## Setup
+To install the required libraries for cpu-only (maybe after creating a virtual environment):
+
+    pip install numpy tensorflow-cpu
+