Why is 'log' keyword in the return parameter of training_step of LightningModule is not working?

[kochark1]

I am seeing the following message in the Tensorboard:
No dashboards are active for the current data set.

I use the following code

    def training_step(self, batch, batch_idx):
            
            # computes loss
    
            tensorboard_logs = {'train_loss': loss}
    
            return {"loss": loss, 'log': tensorboard_logs}

I do the same in validation_epoch_end() as well.

Is this functionality of logging through the 'log' keyword not supported anymore? Should we log explicitly using self.log() or something similar?

[akihironitta]

@kochark1 No, it isn't supported since 1.0.0 (#3681), and yes, you need to use self.log(...). See details here: https://pytorch-lightning.readthedocs.io/en/stable/extensions/logging.html

[danielyan86129]

saved my day...