Appropriate number of epochs when fine-tuning an existing LM

[AliHaiderAhmad001]

Hi, I want to fine-tuning flair on a domain specific dataset, what I want to know is how many epochs are suitable? And how much time would you estimate to complete the training if the number of sentences was about 7 million (about 150 million words)?

I'm using this code you provided here, but it seems to be using a default number of epochs, what is their number? Is it appropriate in your opinion for the size of my data set?

Thanks @alanakbik