Learner starts training from scratch while training more epocs

While training a Pets classification model, I fine-tuned a model and tried training more epochs using fit_one_cycle in the next cycle. But it started training from scratch. Any way around it?


Notebook Link: Google Colab