NAME

AI::MXNet::LRScheduler - The adaptive scheduler of the learning rate.

DESCRIPTION

Learning rate scheduler, which adaptively changes the learning rate based on the progress.

new

base_lr : float (optional, default 0.01) the initial learning rate

call

Call to schedule current learning rate

The training progress is presented by num_update, which can be roughly viewed as the number of minibatches executed so far. Its value is non-decreasing, and increases at most by one.

The exact value is the upper bound of the number of updates applied to a weight/index

See more details in https://github.com/dmlc/mxnet/issues/625

Parameters ---------- num_update: int the maximal number of updates applied to a weight.

NAME

AI::MXNet::FactorScheduler - Reduces the learning rate by a factor.

DESCRIPTION

Reduces the learning rate by a factor each step. Assume the weight has been updated by n times, then the learning rate will be base_lr * factor^(floor(n/step))

Parameters ---------- step: int schedule the learning rate update after n updates factor: float the factor by which to reduce the learning rate.

NAME

AI::MXNet::MultiFactorScheduler - Reduces the learning rate by an array ref of factors.

DESCRIPTION

Reduce learning rate in factor at steps specified in an array ref. Assume the weight has been updated by n times, then the learning rate will be base_lr * factor^(sum((step/n)<=1)) # step is an array

Parameters ---------- step: array ref of int schedule learning rate after n updates factor: float the factor for reducing the learning rate