Loading TOC...

cntk.fsAdaGradLearner

cntk.fsAdaGradLearner(
   $parameters as cntk.parameter[],
   $lr as cntk.schedule,
   $momentum as cntk.schedule,
   [$unit-gain as Boolean],
   [$variance-momentum-schedule as cntk.schedule],
   [$alo as Object]
) as cntk.schedule

Summary

Constructs an FSAdaGrad learner.

Parameters
$parameters
$lr Learning rate schedule.
$momentum Momentum schedule.
$unit-gain Unit gain.
$variance-momentum-schedule Variance momentum schedule.
$alo Additional learning options.

Example

  var shape = [3, 3, 3]
  var parameter1 = cntk.parameter(cntk.shape(shape), "float",
    cntk.xavierInitializer(5.5, 1, 2), cntk.cpu(), "parameter1")
  var lr = cntk.learningRateScheduleFromConstant(0.01, 100)
  var momentum = cntk.momentumScheduleFromConstant(15.5, 50)
  cntk.fsAdaGradLearner(parameter1, lr, momentum)
  => cntk:learner(Name(cntk:learner))

Stack Overflow iconStack Overflow: Get the most useful answers to questions from the MarkLogic community, or ask your own question.