polynomial_decay¶

paddle.fluid.layers. polynomial_decay ( learning_rate, decay_steps, end_learning_rate=0.0001, power=1.0, cycle=False ) [源代码] ¶

对初始学习率使用多项式衰减

if cycle:
    decay_steps = decay_steps * ceil(global_step / decay_steps)
else:
    global_step = min(global_step, decay_steps)
    decayed_learning_rate = (learning_rate - end_learning_rate) *
        (1 - global_step / decay_steps) ^ power + end_learning_rate

参数：

learning_rate (Variable|float) - 训练过程中的初始学习率，数据类型为float的常数或变量。
decay_steps (int) - 衰减步数
end_learning_rate (float) - 训练过程的最终学习率
power (float) - 多项式衰减系数
cycle (bool) - step 超出 decay_steps 后是否继续循环，默认为False

返回：衰减的学习率

返回类型：变量（Variable）

代码示例：

import paddle.fluid as fluid
start_lr = 0.01
total_step = 5000
end_lr = 0
lr = fluid.layers.polynomial_decay(
    start_lr, total_step, end_lr, power=1)

PaddlePaddle

paddle.fluid / layers / polynomial_decay

polynomial_decay¶

PaddlePaddle

paddle

paddle.amp

paddle.callbacks

paddle.compat

paddle.distributed

paddle.distribution

paddle.fluid

paddle.io

paddle.jit

paddle.metric

paddle.nn

paddle.onnx

paddle.optimizer

paddle.regularizer

paddle.static

paddle.sysconfig

paddle.text

paddle.utils

paddle.vision

paddle.fluid / layers / polynomial_decay

polynomial_decay¶