Doubt, please help

34 views
Skip to first unread message

Liset Vázquez Romaguera

unread,
Sep 8, 2016, 2:32:01 PM9/8/16
to Caffe Users
I can define type: "Adam" in my solver.protoxt file, but I can also choose caffe.SGDSolver. I even saw a caffe.pycaffe.Adamsolver. What is the correct way to define optimizer: from prototxt file, or by running SGDsolver or Adamsolver from pycaffe? 

Thanks for help!

mprl

unread,
Sep 9, 2016, 11:16:09 AM9/9/16
to Caffe Users
I'm not sure if i understood the question well, but i think you are making a little confusion. The solver.prototxt file is used only when you are training your model from the command line interface. caffe.SGDSolver and other solver can be used if you are training your model from a python script using Pycaffe.

Evan Shelhamer

unread,
Sep 9, 2016, 6:37:49 PM9/9/16
to Liset Vázquez Romaguera, Caffe Users
`caffe.get_solver('path/to/solver.prototxt')` will instantiate the solver type given in the prototxt.
`caffe.SGDSolver('path/to/solver.prototxt')` will instantiate an `SGDSolver`.

You could prefer `caffe.get_solver()` for generality, but there isn't a strictly correct or incorrect way.

Evan Shelhamer





On Thu, Sep 8, 2016 at 11:32 AM, Liset Vázquez Romaguera <lise...@gmail.com> wrote:
I can define type: "Adam" in my solver.protoxt file, but I can also choose caffe.SGDSolver. I even saw a caffe.pycaffe.Adamsolver. What is the correct way to define optimizer: from prototxt file, or by running SGDsolver or Adamsolver from pycaffe? 

Thanks for help!

--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users+unsubscribe@googlegroups.com.
To post to this group, send email to caffe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/caffe-users/dc4d740f-235d-4e98-9802-14efa9360a97%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply all
Reply to author
Forward
0 new messages