Event

class composer.Event(value)

Enum to represent training loop events.

Events mark specific points in the training loop where an Algorithm and Callback can run.

The following pseudocode shows where each event fires in the training loop:

# <INIT>
# <AFTER_LOAD>
# <FIT_START>
for epoch in range(NUM_EPOCHS):
    # <EPOCH_START>
    while True:
        # <BEFORE_DATALOADER>
        batch = next(dataloader)
        if batch is None:
            break
        # <AFTER_DATALOADER>

        # <BATCH_START>

        # <BEFORE_TRAIN_BATCH>

        for microbatch in batch.split(grad_accum):

            # <BEFORE_FORWARD>
            outputs = model(batch)
            # <AFTER_FORWARD>

            # <BEFORE_LOSS>
            loss = model.loss(outputs, batch)
            # <AFTER_LOSS>

            # <BEFORE_BACKWARD>
            loss.backward()
            # <AFTER_BACKWARD>

        # Un-scale and clip gradients

        # <AFTER_TRAIN_BATCH>
        optimizer.step()

        # <BATCH_END>

        if should_eval(batch=True):
            for eval_dataloader in eval_dataloaders:
                # <EVAL_START>
                for batch in eval_dataloader:
                    # <EVAL_BATCH_START>
                    # <EVAL_BEFORE_FORWARD>
                    outputs, targets = model(batch)
                    # <EVAL_AFTER_FORWARD>
                    metrics.update(outputs, targets)
                    # <EVAL_BATCH_END>
                # <EVAL_END>

        # <BATCH_CHECKPOINT>
    # <EPOCH_END>

    if should_eval(batch=False):
        for eval_dataloader in eval_dataloaders:
            # <EVAL_START>
            for batch in eval_dataloader:
                # <EVAL_BATCH_START>
                # <EVAL_BEFORE_FORWARD>
                outputs, targets = model(batch)
                # <EVAL_AFTER_FORWARD>
                metrics.update(outputs, targets)
                # <EVAL_BATCH_END>
            # <EVAL_END>

    # <EPOCH_CHECKPOINT>
# <FIT_END>

INIT

Invoked in the constructor of Trainer. Model surgery (see module_surgery) typically occurs here.

AFTER_LOAD

Immediately after checkpoint is loaded in constructor of Trainer.

FIT_START

Invoked at the beginning of each call to Trainer.fit(). Dataset transformations typically occur here.

EPOCH_START

Start of an epoch.

BEFORE_DATALOADER

Immediately before the dataloader is called.

AFTER_DATALOADER

Immediately after the dataloader is called. Typically used for on-GPU dataloader transforms.

BATCH_START

Start of a batch.

BEFORE_TRAIN_BATCH

Before the forward-loss-backward computation for a training batch. When using gradient accumulation, this is still called only once.

BEFORE_FORWARD

Before the call to model.forward(). This is called multiple times per batch when using gradient accumulation.

AFTER_FORWARD

After the call to model.forward(). This is called multiple times per batch when using gradient accumulation.

BEFORE_LOSS

Before the call to model.loss(). This is called multiple times per batch when using gradient accumulation.

AFTER_LOSS

After the call to model.loss(). This is called multiple times per batch when using gradient accumulation.

BEFORE_BACKWARD

Before the call to loss.backward(). This is called multiple times per batch when using gradient accumulation.

AFTER_BACKWARD

After the call to loss.backward(). This is called multiple times per batch when using gradient accumulation.

AFTER_TRAIN_BATCH

After the forward-loss-backward computation for a training batch. When using gradient accumulation, this event still fires only once.

BATCH_END

End of a batch, which occurs after the optimizer step and any gradient scaling.

BATCH_CHECKPOINT

After Event.BATCH_END and any batch-wise evaluation. Saving checkpoints at this event allows the checkpoint saver to use the results from any batch-wise evaluation to determine whether a checkpoint should be saved.

EPOCH_END

End of an epoch.

EPOCH_CHECKPOINT

After Event.EPOCH_END and any epoch-wise evaluation. Saving checkpoints at this event allows the checkpoint saver to use the results from any epoch-wise evaluation to determine whether a checkpointshould be saved.

FIT_END

Invoked at the end of each call to Trainer.fit(). This event exists primarily for logging information and flushing callbacks. Algorithms should not transform the training state on this event, as any changes will not be preserved in checkpoints.

EVAL_START

Start of evaluation through the validation dataset.

EVAL_BATCH_START

Before the call to model.eval_forward(batch)

EVAL_BEFORE_FORWARD

Before the call to model.eval_forward(batch)

EVAL_AFTER_FORWARD

After the call to model.eval_forward(batch)

EVAL_BATCH_END

After the call to model.eval_forward(batch)

EVAL_END

End of evaluation through the validation dataset.

property canonical_name

The name of the event, without before/after markers.

Events that have a corresponding “before” or “after” event share the same canonical name.

Example

>>> Event.EPOCH_START.canonical_name
'epoch'
>>> Event.EPOCH_END.canonical_name
'epoch'

Returns

str – The canonical name of the event.

property is_after_event

Whether the event is an “after” event.

An “after” event (e.g., AFTER_LOSS) has a corresponding “before” event (.e.g., BEFORE_LOSS).

property is_before_event

Whether the event is an “before” event.

An “before” event (e.g., BEFORE_LOSS) has a corresponding “after” event (.e.g., AFTER_LOSS).

property is_eval

Whether the event is during the eval loop.

property is_predict

Whether the event is during the predict loop.