This repo is a reproduction of Jianling's C++ port of Andrej Karpathy's phenomenal nanoGPT using jank's seamless C++ interop.
You will need some Python dependencies for the data.
requestsnumpy
Then run:
python data/shakespeare_char/prepare.py
To get the data for training.
Note:
- Install instructions for libtorch can be found here.
- Update the
LIBTORCH_INSTALL_DIRvariable in theMakefileto point to the libtorch installation directory.
make test-setupmake run