Training/Cross validation/Test sets

I am new to machine learning and I am a little lost on these concepts, I trained a model with my training data set(60%) and get the optimized parameters(theta values), do i have to train the cross validation sets? if the answer is yes what then do I do with the new theta values for my cross validation data? If not why do I still need a cross validation sets?

Since I can test my model accuracy with just the test sets. I will appreciate it if someone takes me through.

Also how to assess model accuracy with my test sets. I appreciate any help in advance. Even a tutorial video/codes