***** train metrics ***** epoch = 1.33 total_flos = 14124142GF train_loss = 0.7836 train_runtime = 1:27:16.97 train_samples_per_second = 0.382 train_steps_per_second = 0.095 {'train_runtime': 5236.9755, 'train_samples_per_second': 0.382, 'train_steps_per_second': 0.095, 'total_flos': 1.5165682398461952e+16, 'train_loss': 0.7835705888271332, 'epoch': 1.33}