deep learning - How to pretrain networks using Imagenet -


i want pretrain residual net using imagenet in caffe. wonder hyper-parameters should set when run training. specific,i want know probablyvalue parameters below (1)learning rate (2)learning policy (3)drop out (4)l1 or l2 regulation besides want know how many neurons of inner-product layers before 1000 output layer should set. sincerely thanks.


Comments

Popular posts from this blog

asp.net - How to correctly use QUERY_STRING in ISAPI rewrite? -

jsf - "PropertyNotWritableException: Illegal Syntax for Set Operation" error when setting value in bean -

laravel - Undefined property: Illuminate\Pagination\LengthAwarePaginator::$id (View: F:\project\resources\views\admin\carousels\index.blade.php) -