monitor display

nnet/xvector.py

@@ -377,7 +377,7 @@ class TrainingMonitor():
         self.logger.critical(f"Best accuracy {self.best_accuracy * 100.} obtained at epoch {self.best_accuracy_epoch}")
 
     def update(self,
-               epoch,
+               epoch=None,
                training_acc=None,
                training_loss=None,
                test_eer=None,
@@ -385,20 +385,21 @@ class TrainingMonitor():
                val_loss=None,
                val_acc=None):
 
+        if epoch is not None:
             self.current_epoch = epoch
-        if training_acc:
+        if training_acc is not None:
             self.training_acc.append(training_acc)
-        if training_loss:
+        if training_loss is not None:
             self.training_loss.append(training_loss)
-        if val_eer:
+        if val_eer is not None:
             self.val_eer.append(val_eer)
-        if val_loss:
+        if val_loss is not None:
             self.val_loss.append(val_loss)
-        if val_acc:
+        if val_acc is not None:
             self.val_acc.append(val_acc)
 
         # remember best accuracy and save checkpoint
-        if self.compute_test_eer and test_eer:
+        if self.compute_test_eer and test_eer is not None:
             self.test_eer.append(test_eer)
             self.is_best = test_eer < self.best_eer
             self.best_eer = min(test_eer, self.best_eer)
@@ -407,7 +408,7 @@ class TrainingMonitor():
                 self.current_patience = self.init_patience
             else:
                 self.current_patience -= 1
-        elif val_eer:
+        elif val_eer is not None:
             self.is_best = val_eer < self.best_eer
             self.best_eer = min(val_eer, self.best_eer)
             if self.is_best:
@@ -1375,7 +1376,7 @@ def new_xtrain(dataset_description,
 
     for epoch in range(1, training_opts["epochs"] + 1):
 
-        monitor.update(epoch)
+        monitor.update(epoch=epoch)
 
         # Process one epoch and return the current model
         if monitor.current_patience == 0:
@@ -1405,15 +1406,13 @@ def new_xtrain(dataset_description,
             if training_opts["compute_test_eer"]:
                 test_eer = new_test_metrics(model, device, training_opts["mixed_precision"])
 
-            monitor.update(epoch,
-                           test_eer=test_eer,
+            monitor.update(test_eer=test_eer,
                            val_eer=val_eer,
                            val_loss=val_loss,
                            val_acc=val_acc)
 
             monitor.display()
 
-
             # Save the current model and if needed update the best one
             # TODO ajouter une option qui garde les modèles à certaines époques (par exemple avant le changement de LR
             save_model(model, monitor, model_opts, training_opts, optimizer, scheduler)