tflogger 1.0.1

TFLogger is a standalone library to read and write events for TensorBoard.

TFLogger

This is a simple library that has been carved out of TensorFlow's codebase for being able to generate the required output format for TensorBoard.

Building

In order to build from source the following packages are required:

• cmake
• protobuf

To build the library simply run the following commands:

git clone https://github.com/shogun-toolbox/tflogger.git
cd tflogger
mkdir build
cd build
cmake ..
make

Testing

In order to enable the unit tests add the -DBUILD-TESTS=ON cmake option to the cmake command. This will automatically create a target for building the unit tests and adds it to the ctest targets.

In other words to run the unit tests:

cmake -DBUILD-TESTS=ON [other cmake flags] ..
make
ctest

Python interface

Currently only the python wrapper is available, but further targets could be specified using SWIG.

For building the python wrapper you need to add the -DINTERFACE_PYTHON=ON cmake flag to the cmake command:

cmake -DINTERFACE_PYTHON=ON ..

Using

Examples of how to use the library are available under example directory.