Shortcuts

pytorch_lightning.utilities.debugging module

class pytorch_lightning.utilities.debugging.InternalDebugger(trainer)[source]

Bases: object

count_events(evt_type, strict=False)[source]
Return type

int

track_checkpointing_history(filepath)[source]
track_early_stopping_history(current)[source]
track_eval_loss_history(test_mode, batch_idx, dataloader_idx, output)[source]
track_event(evt_type, evt_value=None, global_rank=None, local_rank=None, comment='')[source]
Return type

None

track_logged_metrics_history(scalar_metrics)[source]
track_lr_schedulers_update(batch_idx, interval, scheduler_idx, old_lr, new_lr, monitor_key=None)[source]
track_pbar_metrics_history(metrics)[source]
track_train_loss_history(batch_idx, loss)[source]
property num_seen_sanity_check_batches[source]
property num_seen_test_check_batches[source]
property num_seen_val_check_batches[source]
pytorch_lightning.utilities.debugging.enabled_only(fn)[source]

Decorate a logger method to run it only on the process with rank 0.

Parameters

fn (Callable) – Function to decorate