Need some help with fine_tune

Hello,
Unfortunately I wasn’t able to find the answer myself, so I am asking here
I am doing some training and this example I follow.


What does first 0 epoch represents?
Thank you

So, when we use fine_tune, we freeze the model and train for one epoch.
Then, unfeeze and train for however more epochs given with smaller learning rate.
To learn more about it, you can learn the documentation, where you can find a description, example, and source code. However, if you want to learn more in depth, you can check out chapter 5 of fastbook.

2 Likes

Now it makes a sense.
Maybe I didn’t payed attention that part in chapter 05,. I will go back.
Thank you so much

1 Like