A general-purpose, lightweight feed-forward neural network library for C++
- Fully cross-platform compilation and targeting
- Customizable input, hidden, and output layer sizes
- Serializing/deserializing trained models, saving/loading, etc.
For the Pong demo, just run Run.sh. This will compile and run main.cpp to show an example of this FNN library playing Pong. Almost always, main.cpp can be modified to fit your use case, and it is a great example to build from.