4-bit quantized weights (q4_0) of OpenAssistant's 12B Pythia model for use with gptneox.cpp (fork of llama.cpp)

https://github.com/byroneverson/gptneox.cpp

The weights bin file should be placed in the models/pythia subdirectory of gptneox.cpp in order to use the chat scripts.

This is a ggjt version of the model, the tensors are 32 byte aligned to allow mmap loading. This will load just fine in gptneox.cpp or my fork or ggml.