Continue training?

Hi, 

Since I'm not familiar with RL, I would like to ask a simple question.

Can I train my policy based on the result of last training? How can I realize it?

I would be really appreciated for your reply!
