Loading TOC...

cntk:nesterov-learner

cntk:nesterov-learner(
   $parameters as cntk:parameter*,
   $lr as cntk:schedule,
   $momentum as cntk:schedule,
   [$unit-gain as xs:boolean],
   [$alo as map:map]
) as cntk:schedule

Summary

Constructs an Nesterov learner.

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

Example

  let $shape := (3, 3, 3)
  let $parameter1 := cntk:parameter(cntk:shape($shape), "float",
    cntk:xavier-initializer(5.5, 1, 2), cntk:cpu(), "parameter1")
  let $lr := cntk:learning-rate-schedule-from-constant(0.01, 100)
  let $momentum := cntk:momentum-schedule-from-constant(15.5, 50)
  let $learner := cntk:nesterov-learner($parameter1, $lr, $momentum)
  return fn:replace(xdmp:quote($learner), "0x[0-9A-Za-z]*", "Learner")
  => 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.