Retrain a pretrained model from scratch without restarting colab kernel

vision
7

Yes, a proper learning rate is essential to train a deep learning model and the batch size is also a hyperparameter, which influences the training.