IterBasedTrainLoop — mmengine 0.10.7 documentation (original) (raw)

class mmengine.runner.IterBasedTrainLoop(runner, dataloader, max_iters, val_begin=1, val_interval=1000, dynamic_intervals=None)[source]¶

Loop for iter-based training.

Parameters:

• runner (Runner) – A reference of runner.
• dataloader (Dataloader or dict) – A dataloader object or a dict to build a dataloader.
• max_iters (int) – Total training iterations.
• val_begin (int) – The iteration that begins validating. Defaults to 1.
• val_interval (int) – Validation interval. Defaults to 1000.
• dynamic_intervals (List [_ _Tuple_ _[_int,_ int] ] , optional) – The first element in the tuple is a milestone and the second element is a interval. The interval is used after the corresponding milestone. Defaults to None.

property epoch¶

Current epoch.

Type:

int

property iter¶

Current iteration.

Type:

int

property max_epochs¶

Total epochs to train model.

Type:

int

property max_iters¶

Total iterations to train model.

Type:

int

run()[source]¶

Launch training.

Return type:

None

run_iter(data_batch)[source]¶

Iterate one mini-batch.

Parameters:

data_batch (Sequence _[_dict]) – Batch of data from dataloader.

Return type:

None