funasr/bin/train.py
@@ -241,6 +241,8 @@ f"estimated to finish {trainer.max_epoch} " f"epoch: {(trainer.max_epoch - epoch) * time_escaped:.3f} hours\n" ) trainer.train_acc_avg = 0.0 trainer.train_loss_avg = 0.0 if trainer.rank == 0: average_checkpoints(trainer.output_dir, trainer.avg_nbest_model)
