Hook

class mmengine.hooks.Hook

Base hook class.

All hooks should inherit from this class.

after_load_checkpoint(runner, checkpoint)

All subclasses should override this method, if they need any operations after loading the checkpoint.

Parameters:

• runner (Runner) – The runner of the training, validation or testing process.

• checkpoint (dict) – Model’s checkpoint.

Return type:

None

after_run(runner)

All subclasses should override this method, if they need any operations before the training validation or testing process.

Parameters:

runner (Runner) – The runner of the training, validation or testing process.

Return type:

None

after_test(runner)

All subclasses should override this method, if they need any operations after testing.

Parameters:

runner (Runner) – The runner of the testing process.

Return type:

None

after_test_epoch(runner, metrics=None)

All subclasses should override this method, if they need any operations after each test epoch.

Parameters:

• runner (Runner) – The runner of the testing process.

• metrics (Dict[str, float], optional) – Evaluation results of all metrics on test dataset. The keys are the names of the metrics, and the values are corresponding results.

Return type:

None

after_test_iter(runner, batch_idx, data_batch=None, outputs=None)

All subclasses should override this method, if they need any operations after each test iteration.

Parameters:

• runner (Runner) – The runner of the training process.

• batch_idx (int) – The index of the current batch in the test loop.

• data_batch (dict or tuple or list, optional) – Data from dataloader.

• outputs (Sequence, optional) – Outputs from model.

Return type:

None

after_train(runner)

All subclasses should override this method, if they need any operations after train.

Parameters:

runner (Runner) – The runner of the training process.

Return type:

None

after_train_epoch(runner)

All subclasses should override this method, if they need any operations after each training epoch.

Parameters:

runner (Runner) – The runner of the training process.

Return type:

None

after_train_iter(runner, batch_idx, data_batch=None, outputs=None)

All subclasses should override this method, if they need any operations after each training iteration.

Parameters:

• runner (Runner) – The runner of the training process.

• batch_idx (int) – The index of the current batch in the train loop.

• data_batch (dict tuple or list, optional) – Data from dataloader.

• outputs (dict, optional) – Outputs from model.

Return type:

None

after_val(runner)

All subclasses should override this method, if they need any operations after validation.

Parameters:

runner (Runner) – The runner of the validation process.

Return type:

None

after_val_epoch(runner, metrics=None)

All subclasses should override this method, if they need any operations after each validation epoch.

Parameters:

• runner (Runner) – The runner of the validation process.

• metrics (Dict[str, float], optional) – Evaluation results of all metrics on validation dataset. The keys are the names of the metrics, and the values are corresponding results.

Return type:

None

after_val_iter(runner, batch_idx, data_batch=None, outputs=None)

All subclasses should override this method, if they need any operations after each validation iteration.

Parameters:

• runner (Runner) – The runner of the validation process.

• batch_idx (int) – The index of the current batch in the val loop.

• data_batch (dict or tuple or list, optional) – Data from dataloader.

• outputs (Sequence, optional) – Outputs from model.

Return type:

None

before_run(runner)

All subclasses should override this method, if they need any operations before the training validation or testing process.

Parameters:

runner (Runner) – The runner of the training, validation or testing process.

Return type:

None

before_save_checkpoint(runner, checkpoint)

All subclasses should override this method, if they need any operations before saving the checkpoint.

Parameters:

• runner (Runner) – The runner of the training, validation or testing process.

• checkpoint (dict) – Model’s checkpoint.

Return type:

None

before_test(runner)

All subclasses should override this method, if they need any operations before testing.

Parameters:

runner (Runner) – The runner of the testing process.

Return type:

None

before_test_epoch(runner)

All subclasses should override this method, if they need any operations before each test epoch.

Parameters:

runner (Runner) – The runner of the testing process.

Return type:

None

before_test_iter(runner, batch_idx, data_batch=None)

All subclasses should override this method, if they need any operations before each test iteration.

Parameters:

• runner (Runner) – The runner of the testing process.

• batch_idx (int) – The index of the current batch in the test loop.

• data_batch (dict or tuple or list, optional) – Data from dataloader. Defaults to None.

Return type:

None

before_train(runner)

All subclasses should override this method, if they need any operations before train.

Parameters:

runner (Runner) – The runner of the training process.

Return type:

None

before_train_epoch(runner)

All subclasses should override this method, if they need any operations before each training epoch.

Parameters:

runner (Runner) – The runner of the training process.

Return type:

None

before_train_iter(runner, batch_idx, data_batch=None)

All subclasses should override this method, if they need any operations before each training iteration.

Parameters:

• runner (Runner) – The runner of the training process.

• batch_idx (int) – The index of the current batch in the train loop.

• data_batch (dict or tuple or list, optional) – Data from dataloader.

Return type:

None

before_val(runner)

All subclasses should override this method, if they need any operations before validation.

Parameters:

runner (Runner) – The runner of the validation process.

Return type:

None

before_val_epoch(runner)

All subclasses should override this method, if they need any operations before each validation epoch.

Parameters:

runner (Runner) – The runner of the validation process.

Return type:

None

before_val_iter(runner, batch_idx, data_batch=None)

All subclasses should override this method, if they need any operations before each validation iteration.

Parameters:

• runner (Runner) – The runner of the validation process.

• batch_idx (int) – The index of the current batch in the val loop.

• data_batch (dict, optional) – Data from dataloader. Defaults to None.

Return type:

None

end_of_epoch(dataloader, batch_idx)

Check whether the current iteration reaches the last iteration of the dataloader.

Parameters:

• dataloader (Dataloader) – The dataloader of the training, validation or testing process.

• batch_idx (int) – The index of the current batch in the loop.

Returns:

Whether reaches the end of current epoch or not.

Return type:

bool

every_n_epochs(runner, n, start=0)

Test whether current epoch can be evenly divided by n.

Parameters:

• runner (Runner) – The runner of the training, validation or testing process.

• n (int) – Whether current epoch can be evenly divided by n.

• start (int) – Starting from start to check the logic for every n epochs. Defaults to 0.

Returns:

Whether current epoch can be evenly divided by n.

Return type:

bool

every_n_inner_iters(batch_idx, n)

Test whether current inner iteration can be evenly divided by n.

Parameters:

• batch_idx (int) – Current batch index of the training, validation or testing loop.

• n (int) – Whether current inner iteration can be evenly divided by n.

Returns:

Whether current inner iteration can be evenly divided by n.

Return type:

bool

every_n_train_iters(runner, n, start=0)

Test whether current training iteration can be evenly divided by n.

Parameters:

• runner (Runner) – The runner of the training, validation or testing process.

• n (int) – Whether current iteration can be evenly divided by n.

• start (int) – Starting from start to check the logic for every n iterations. Defaults to 0.

Returns:

Return True if the current iteration can be evenly divided by n, otherwise False.

Return type:

bool

get_triggered_stages()

Get all triggered stages with method name of the hook.

Returns:

List of triggered stages.

Return type:

list

is_last_train_epoch(runner)

Test whether current epoch is the last train epoch.

Parameters:

runner (Runner) – The runner of the training process.

Returns:

Whether reaches the end of training epoch.

Return type:

bool

is_last_train_iter(runner)

Test whether current iteration is the last train iteration.

Parameters:

runner (Runner) – The runner of the training process.

Returns:

Whether current iteration is the last train iteration.

Return type:

bool