Adding `safetensors` variant of this model (#1)

tokenizer: add initial config

model: add converted PyTorch weights (from final 100k steps)

model: add original TensorFlow checkpoints

git: extend lfs list

readme: add initial version

initial commit