Different module metrics for train/val

[Rohit Gupta]

Module metrics stores internal states computed over each call on different batches. So using the same instance for both train and val might not lead to correct results when computed over epoch with (on_epoch=True) in step hooks. I'd suggest creating separate ones for each task (train & val).

ref: https://torchmetrics.readthedocs.io/en/latest/pages/quickstart.html#module-metrics

MLOps-Basics/week_9_monitoring/model.py

Lines 24 to 33 in d4ad5a9

 self.accuracy_metric = torchmetrics.Accuracy() 
 self.f1_metric = torchmetrics.F1(num_classes=self.num_classes) 
 self.precision_macro_metric = torchmetrics.Precision( 
     average="macro", num_classes=self.num_classes 
 ) 
 self.recall_macro_metric = torchmetrics.Recall( 
     average="macro", num_classes=self.num_classes 
 ) 
 self.precision_micro_metric = torchmetrics.Precision(average="micro") 
 self.recall_micro_metric = torchmetrics.Recall(average="micro") 

[Rohit Gupta]

cc @graviraja

[raviraja]

@rohitgr7 fixed it. Thanks for pointing it out :)