fix printing

mines-opt-ml · Aug 11, 2023 · aad6213 · aad6213
1 parent 0c629ad
commit aad6213
Showing 1 changed file with 11 additions and 10 deletions.
diff --git a/src/trainer.py b/src/trainer.py
@@ -204,13 +204,14 @@ def trainer_warcraft(net, train_dataset, val_dataset, test_dataset,
     epoch_time=0
 
     # print initial test loss in :7.3e format
-    print('initial_val_loss: ', "{:5.2e}".format(val_loss), 
-          ' | initial_val_acc: ', "{:<4f}".format(val_acc), 
-          ' | initial_val_cost_pred: ', "{:5.2e}".format(val_cost_pred),
+    print('INITIAL VALUES:')
+    print('val_loss: ', "{:5.2e}".format(val_loss), 
+          ' | val_acc: ', "{:<4.3g}".format(val_acc), 
+          ' | : ', "{:5.2e}".format(val_cost_pred),
           ' | true val_cost: ', "{:5.2e}".format(val_cost_true),
-          'initial_test_loss: ', "{:5.2e}".format(test_loss), 
-          ' | initial_test_acc: ', "{:<4f}".format(test_acc),
-          ' | initial_test_cost_pred: ', "{:5.2e}".format(test_cost_pred),
+          'test_loss: ', "{:5.2e}".format(test_loss), 
+          ' | test_acc: ', "{:<4.3g}".format(test_acc),
+          ' | test_cost_pred: ', "{:5.2e}".format(test_cost_pred),
           ' | true test_cost: ', "{:5.2e}".format(test_cost_true))
 
     while epoch <= max_epochs:
@@ -256,7 +257,7 @@ def trainer_warcraft(net, train_dataset, val_dataset, test_dataset,
 
         print('epoch: ', epoch, '| ave_tr_loss: ', "{:5.2e}".format(train_loss_ave), 
               '| val_loss: ', "{:5.2e}".format(val_loss), 
-              '| val_acc.: ', "{:<4f}".format(val_acc), 
+              '| val_acc.: ', "{:<4.3g}".format(val_acc), 
               '| val_cost_pred: ', "{:5.2e}".format(val_cost_pred), 
               '| lr: ', "{:5.2e}".format(optimizer.param_groups[0]['lr']), 
               '| time: ', "{:<15f}".format(epoch_time))
@@ -269,8 +270,8 @@ def trainer_warcraft(net, train_dataset, val_dataset, test_dataset,
             print('path_pred edge = ', torch.nonzero(path_pred[2,:]))
             print(edge_to_node(path_pred[2,:], edge_list, grid_size, device))
             print('\n True Path \n')
-            print(edge_to_node(path_batch[2,:], edge_list, grid_size, device))
-            print('path_batch edge = ', torch.nonzero(path_batch[2,:]))
+            print(edge_to_node(path_batch_edge[2,:], edge_list, grid_size, device))
+            print('path_batch edge = ', torch.nonzero(path_batch_edge[2,:]))
             print('\n ------------------------ \n')
 
 
@@ -294,7 +295,7 @@ def trainer_warcraft(net, train_dataset, val_dataset, test_dataset,
             # pred_batch_edge_form=True means path_pred_edge is in edge form.
             test_acc, test_cost_pred, test_cost_true = compute_accuracy(path_pred_edge, path_batch_vertex, costs_batch, edge_list, grid_size, device, pred_batch_edge_form=True)
 
-    print('final test loss is ', "{:5.2e}".format(test_loss), ' | final test acc. is ', "{:<4f}".format(test_acc), ' | final test cost pred is ', "{:5.2e}".format(test_cost_pred), ' | final test cost true is ', "{:5.2e}".format(test_cost_true))
+    print('final test loss is ', "{:5.2e}".format(test_loss), ' | final test acc. is ', "{:<4.3g}".format(test_acc), ' | final test cost pred is ', "{:5.2e}".format(test_cost_pred), ' | final test cost true is ', "{:5.2e}".format(test_cost_true))
 
     return best_params, val_loss_hist, val_acc_hist, test_loss, test_acc, train_time