# start training, the model will be automatically saved to the hub and the output directory
trainer.train()
 
# save model at the end of training
trainer.save_model()