Shortcuts

Migration of Schedule Settings

We update schedule settings in MMEdit 1.x. Important modifications are as following.

  • Now we use optim_wrapper field to specify all configuration about the optimization process. And the optimizer is a sub field of optim_wrapper now.

  • The lr_config field is removed and we use new param_scheduler to replace it.

  • The total_iters field is moved to train_cfg as max_iters, val_cfg and test_cfg, which configure the loop in training, validation and test.

Original New
optimizers = dict(generator=dict(type='Adam', lr=1e-4, betas=(0.9, 0.999)))  # Config used to build optimizer, support all the optimizers in PyTorch whose arguments are also the same as those in PyTorch
total_iters = 300000 # Total training iters
lr_config = dict( # Learning rate scheduler config used to register LrUpdater hook
    policy='Step', by_epoch=False, step=[200000], gamma=0.5)  # The policy of scheduler
optim_wrapper = dict(
    dict(
        type='OptimWrapper',
        optimizer=dict(type='Adam', lr=1e-4),
    )
)  # Config used to build optimizer, support all the optimizers in PyTorch whose arguments are also the same as those in PyTorch.
param_scheduler = dict(  # Config of learning policy
    type='MultiStepLR', by_epoch=False, milestones=[200000], gamma=0.5)  # The policy of scheduler
train_cfg = dict(
    type='IterBasedTrainLoop', max_iters=300000, val_interval=5000)  # Config of train loop type
val_cfg = dict(type='ValLoop')  # The name of validation loop type
test_cfg = dict(type='TestLoop')  # The name of test loop type

More details of schedule settings are shown in MMEngine Documents.

Read the Docs v: zyh/api-rendering
Versions
master
latest
stable
zyh-doc-notfound-extend
zyh-api-rendering
Downloads
pdf
epub
On Read the Docs
Project Home
Builds

Free document hosting provided by Read the Docs.