Update core/train_eval.py

core/train_eval.py

@@ -28,7 +28,7 @@ def mean_absolute_percentage_error(y_true, y_pred):
 
 def train_and_evaluate(
     df,
-    future_df,  # Added
+    future_df,
     model_cls,
     horizon=1,
     hidden=64,
@@ -77,7 +77,7 @@ def train_and_evaluate(
     model = model_cls(input_size=input_dim, hidden_size=hidden, num_layers=layers, output_size=horizon, dropout=dropout).to(device)
     optimizer = torch.optim.AdamW(model.parameters(), lr=lr, betas=(beta1, beta2), weight_decay=weight_decay)
     loss_fn = nn.MSELoss()
-    scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode='min', patience=5, factor=0.5, threshold=1e-4, verbose=True)
+    scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, mode='min', patience=5, factor=0.5, threshold=1e-4)
 
     train_losses = []
     val_losses = []
@@ -85,6 +85,7 @@ def train_and_evaluate(
     patience = 5
     counter = 0
     best_model_state = None
+    last_lr = lr  # Track learning rate for manual logging
 
     model.train()
     for epoch in range(epochs):
@@ -110,10 +111,15 @@ def train_and_evaluate(
         val_loss /= len(val_loader)
         val_losses.append(val_loss)
 
+        # Step scheduler and manually log learning rate changes
         scheduler.step(val_loss)
+        current_lr = optimizer.param_groups[0]['lr']
+        if current_lr != last_lr and verbose:
+            print(f"Epoch {epoch+1}: Learning rate reduced to {current_lr:.6f}")
+        last_lr = current_lr
 
         if verbose and (epoch + 1) % 10 == 0:
-            print(f"Epoch {epoch+1}/{epochs} - Train Loss: {train_losses[-1]:.4f}, Val Loss: {val_losses[-1]:.4f}, LR: {optimizer.param_groups[0]['lr']:.6f}")
+            print(f"Epoch {epoch+1}/{epochs} - Train Loss: {train_losses[-1]:.4f}, Val Loss: {val_losses[-1]:.4f}, LR: {current_lr:.6f}")
 
         if val_loss < best_val_loss:
             best_val_loss = val_loss