TrainOutput( global_step=598, training_loss=0.3596907002371689, metrics={'train_runtime': 24697.2911, 'train_samples_per_second': 1.162, 'train_steps_per_second': 0.024, 'total_flos': 0.0, 'train_loss': 0.3596907002371689, 'epoch': 2.724373576309795 })
accu: 91%