Caffe LeNet: Difference between `solver.step (1)` and `solver.net.forward ()`

I was looking through the Caffe LeNet tutorial here , and the question came to mind:

What is the difference between these two codes:

self.solver.step(1)

and

self.solver.net.forward()  # train net

They seem to train the network at least in accordance with the comment.

Personally, I think that the first trains the network in training data and updates the weight of both net, and so test_net, but the second seems to only transmit the data packet and apply the lessons learned from the previous step.

If what I think is right, then what is the meaning of the second code in the textbook? why did the code do net.forward? can't solver.step(1)do it myself?

thank you for your time

+4
1

, : , , , forward . ( ).

+7

All Articles