Using log() and log_dict() in pl 1.9.4.

[vgthengane]

I need some clarification on log_dict() and log(). In single GPU mode, in my code, I was using log_dict() but recently I updated the code to support DDP. When I did that I was getting a warning PossibleUserWarning: It is recommended to use 'self.log('some_val', ..., sync_dist=True)' when logging on epoch level in distributed setting to accumulate the metric across devices. I went through some issues on GitHub all off them are using log() but I have to log every metric in this case, also I found that log_dict supports sync_dist=True. Here I want to know if log_dict works in the same way as log() or differently. Also how to check if log_dict is accumulation metrics from all the devices? I am using pytorch_lightning==1.9.4