Logging metrics with compute()

[ss007k]

Trying to understand compute() for logging. Could someone please tell if the self.log(train_acc..., on_epoch=True) is same as def training_epoch_end() which uses compute - code below

def __init__():
    # ...
    self.train_acc = pl.metrics.Accuracy()
    self.valid_acc = pl.metrics.Accuracy()

def training_step(self, batch, batch_idx):
    # ...
    self.train_acc(pred, target)
    self.log("train_acc", self.train_acc, on_step=False, on_epoch=True)
    self.log('train_loss', loss, on_step=True, on_epoch=False)

    return loss

def training_epoch_end(self, training_step_outputs):
    self.log('train_acc_epoch', self.train_acc.compute())
    

def validation_step(self, batch, batch_idx):
    # ...
    self.valid_acc(pred, target)
    self.log('val_loss', loss, prog_bar=True)
    self.log('val_acc', self.valid_acc, on_step=False, on_epoch=True, prog_bar=True)

    return loss


def validation_epoch_end(self, validation_step_outputs):
    self.log('valid_acc_epoch', self.valid_acc.compute())

[SkafteNicki]

Yes, its the same. Here is the relevant code:
https://github.com/PyTorchLightning/pytorch-lightning/blob/4a8422c2dc870d29d26748e6aa01406d2e74fa48/pytorch_lightning/core/step_result.py#L325-L330
Setting on_epoch=True will essentially call metric.compute at the end of the epoch.

[ss007k]

Thank you for clarfiying