update API PL v1.6 - Trainer dataloaders (#150)

* test_dataloaders
* torch-sparse<0.6.13

Author: Borda

course_UvA-DL/04-inception-resnet-densenet/Inception_ResNet_DenseNet.py

@@ -377,8 +377,8 @@ def train_model(model_name, save_name=None, **kwargs):
         )  # Load best checkpoint after training
 
     # Test best model on validation and test set
-    val_result = trainer.test(model, test_dataloaders=val_loader, verbose=False)
-    test_result = trainer.test(model, test_dataloaders=test_loader, verbose=False)
+    val_result = trainer.test(model, dataloaders=val_loader, verbose=False)
+    test_result = trainer.test(model, dataloaders=test_loader, verbose=False)
     result = {"test": test_result[0]["test_acc"], "val": val_result[0]["test_acc"]}
 
     return model, result

course_UvA-DL/05-transformers-and-MH-attention/Transformers_MHAttention.py

@@ -996,8 +996,8 @@ def train_reverse(**kwargs):
         trainer.fit(model, train_loader, val_loader)
 
     # Test best model on validation and test set
-    val_result = trainer.test(model, test_dataloaders=val_loader, verbose=False)
-    test_result = trainer.test(model, test_dataloaders=test_loader, verbose=False)
+    val_result = trainer.test(model, dataloaders=val_loader, verbose=False)
+    test_result = trainer.test(model, dataloaders=test_loader, verbose=False)
     result = {"test_acc": test_result[0]["test_acc"], "val_acc": val_result[0]["test_acc"]}
 
     model = model.to(device)
@@ -1457,9 +1457,9 @@ def train_anomaly(**kwargs):
         model = AnomalyPredictor.load_from_checkpoint(trainer.checkpoint_callback.best_model_path)
 
     # Test best model on validation and test set
-    train_result = trainer.test(model, test_dataloaders=train_anom_loader, verbose=False)
-    val_result = trainer.test(model, test_dataloaders=val_anom_loader, verbose=False)
-    test_result = trainer.test(model, test_dataloaders=test_anom_loader, verbose=False)
+    train_result = trainer.test(model, dataloaders=train_anom_loader, verbose=False)
+    val_result = trainer.test(model, dataloaders=val_anom_loader, verbose=False)
+    test_result = trainer.test(model, dataloaders=test_anom_loader, verbose=False)
     result = {
         "test_acc": test_result[0]["test_acc"],
         "val_acc": val_result[0]["test_acc"],

course_UvA-DL/06-graph-neural-networks/.meta.yml

@@ -19,7 +19,7 @@ description: |
   The full list of tutorials can be found at https://uvadlc-notebooks.rtfd.io.
 requirements:
   - torch-scatter
-  - torch-sparse
+  - torch-sparse<0.6.13
   - torch-cluster
   - torch-spline-conv
   - torch-geometric==2.0.2

course_UvA-DL/06-graph-neural-networks/GNN_overview.py

@@ -670,7 +670,7 @@ def train_node_classifier(model_name, dataset, **model_kwargs):
         model = NodeLevelGNN.load_from_checkpoint(trainer.checkpoint_callback.best_model_path)
 
     # Test best model on the test set
-    test_result = trainer.test(model, test_dataloaders=node_data_loader, verbose=False)
+    test_result = trainer.test(model, dataloaders=node_data_loader, verbose=False)
     batch = next(iter(node_data_loader))
     batch = batch.to(model.device)
     _, train_acc = model.forward(batch, mode="train")
@@ -953,8 +953,8 @@ def train_graph_classifier(model_name, **model_kwargs):
         model = GraphLevelGNN.load_from_checkpoint(trainer.checkpoint_callback.best_model_path)
 
     # Test best model on validation and test set
-    train_result = trainer.test(model, test_dataloaders=graph_train_loader, verbose=False)
-    test_result = trainer.test(model, test_dataloaders=graph_test_loader, verbose=False)
+    train_result = trainer.test(model, dataloaders=graph_train_loader, verbose=False)
+    test_result = trainer.test(model, dataloaders=graph_test_loader, verbose=False)
     result = {"test": test_result[0]["test_acc"], "train": train_result[0]["test_acc"]}
     return model, result
 

course_UvA-DL/08-deep-autoencoders/Deep_Autoencoders.py

@@ -405,8 +405,8 @@ def train_cifar(latent_dim):
         model = Autoencoder(base_channel_size=32, latent_dim=latent_dim)
         trainer.fit(model, train_loader, val_loader)
     # Test best model on validation and test set
-    val_result = trainer.test(model, test_dataloaders=val_loader, verbose=False)
-    test_result = trainer.test(model, test_dataloaders=test_loader, verbose=False)
+    val_result = trainer.test(model, dataloaders=val_loader, verbose=False)
+    test_result = trainer.test(model, dataloaders=test_loader, verbose=False)
     result = {"test": test_result, "val": val_result}
     return model, result
 

course_UvA-DL/09-normalizing-flows/NF_image_modeling.py

@@ -947,9 +947,9 @@ def train_flow(flow, model_name="MNISTFlow"):
     # Test best model on validation and test set if no result has been found
     # Testing can be expensive due to the importance sampling.
     if result is None:
-        val_result = trainer.test(flow, test_dataloaders=val_loader, verbose=False)
+        val_result = trainer.test(flow, dataloaders=val_loader, verbose=False)
         start_time = time.time()
-        test_result = trainer.test(flow, test_dataloaders=test_loader, verbose=False)
+        test_result = trainer.test(flow, dataloaders=test_loader, verbose=False)
         duration = time.time() - start_time
         result = {"test": test_result, "val": val_result, "time": duration / len(test_loader) / flow.import_samples}
 

course_UvA-DL/10-autoregressive-image-modeling/Autoregressive_Image_Modeling.py

@@ -698,8 +698,8 @@ def train_model(**kwargs):
 
     if result is None:
         # Test best model on validation and test set
-        val_result = trainer.test(model, test_dataloaders=val_loader, verbose=False)
-        test_result = trainer.test(model, test_dataloaders=test_loader, verbose=False)
+        val_result = trainer.test(model, dataloaders=val_loader, verbose=False)
+        test_result = trainer.test(model, dataloaders=test_loader, verbose=False)
         result = {"test": test_result, "val": val_result}
     return model, result
 

course_UvA-DL/13-contrastive-learning/SimCLR.py

@@ -571,8 +571,8 @@ def train_logreg(batch_size, train_feats_data, test_feats_data, model_suffix, ma
         model = LogisticRegression.load_from_checkpoint(trainer.checkpoint_callback.best_model_path)
 
     # Test best model on train and validation set
-    train_result = trainer.test(model, test_dataloaders=train_loader, verbose=False)
-    test_result = trainer.test(model, test_dataloaders=test_loader, verbose=False)
+    train_result = trainer.test(model, dataloaders=train_loader, verbose=False)
+    test_result = trainer.test(model, dataloaders=test_loader, verbose=False)
     result = {"train": train_result[0]["test_acc"], "test": test_result[0]["test_acc"]}
 
     return model, result
@@ -767,8 +767,8 @@ def train_resnet(batch_size, max_epochs=100, **kwargs):
         model = ResNet.load_from_checkpoint(trainer.checkpoint_callback.best_model_path)
 
     # Test best model on validation set
-    train_result = trainer.test(model, test_dataloaders=train_loader, verbose=False)
-    val_result = trainer.test(model, test_dataloaders=test_loader, verbose=False)
+    train_result = trainer.test(model, dataloaders=train_loader, verbose=False)
+    val_result = trainer.test(model, dataloaders=test_loader, verbose=False)
     result = {"train": train_result[0]["test_acc"], "test": val_result[0]["test_acc"]}
 
     return model, result