learningrate

Query and control PyTorch optimizer learning rate.

torchutils.learningrate.get_lr(optimizer)

Get learning rate.

Parameters

optimizer (optim.Optimizer) – PyTorch optimizer.

Returns

Learning rate of the optimizer.

Return type

float

Example:

import torchvision
import torchutils as tu
import torch.optim as optim

model = torchvision.models.alexnet()
optimizer = optim.Adam(model.parameters())
current_lr = tu.get_lr(optimizer)
print('Current learning rate:', current_lr)

Out:

Current learning rate: 0.001

torchutils.learningrate.set_lr(optimizer, lr)

Set learning rate.

Parameters

• optimizer (optim.Optimizer) – PyTorch optimizer.

• lr (float) – New learning rate value.

Returns

PyTorch optimizer.

Return type

optim.Optimizer

Example:

import torchvision
import torchutils as tu
import torch.optim as optim

model = torchvision.models.alexnet()
optimizer = optim.Adam(model.parameters())
current_lr = tu.get_lr(optimizer)
print('Current learning rate:', current_lr)

optimizer = tu.set_lr(optimizer, current_lr*0.1)
revised_lr = tu.get_lr(optimizer)
print('Revised learning rate:', revised_lr)

Out:

Current learning rate: 0.001
Revised learning rate: 0.0001