set `renormalize_logits=True` for better performance (see `transformers` issue #4290)

Training in progress, epoch 0

update model card README.md

End of training

Training in progress, epoch 1

Training in progress, epoch 1

initial commit