add logging messages and edit requirements.txt

gnn_toolbox/custom_components/utils.py

@@ -38,20 +38,12 @@ def grad_with_checkpoint(outputs: Union[torch.Tensor, Sequence[torch.Tensor]],
     inputs = (inputs,) if isinstance(inputs, torch.Tensor) else tuple(inputs)
 
     for input in inputs:
-        # input.retain_grad()
         if not input.is_leaf:
             input.retain_grad()
 
     torch.autograd.backward(outputs)
 
     grad_outputs = []
-    # for input in inputs:
-    #     if input.grad is not None:
-    #         grad_outputs.append(input.grad.clone())
-    #         input.grad.zero_()
-    #     else:
-    #         # Append zeros in the same shape as the input if no gradient was computed
-    #         grad_outputs.append(torch.ones_like(input))
 
     for input in inputs:
         grad_outputs.append(input.grad.clone())

gnn_toolbox/experiment_handler/artifact_manager.py

@@ -40,9 +40,9 @@ def save_model(self, model, params, result, is_unattacked_model):
             with result_path.open('w') as file:
                 json.dump(result, file, indent=4)
         except Exception as e:
-            logging.error(f"Failed to save model or results: {e}")
+            logging.error(f"Failed to save model or results to {model_dir}: {e}")
         else:
-            logging.info(f'Saved the model {model_suffix} at {model_dir}')
+            logging.info(f'Saved the model {model_suffix} to {model_dir} for caching')
 
     def model_exists(self, params, is_unattacked_model):
         """ Check if a model with the given parameters already exists. """
@@ -51,9 +51,6 @@ def model_exists(self, params, is_unattacked_model):
         hash_id = self.hash_parameters(params)
         params_dir = self.cache_directory / f"{hash_id}"
         if self.folder_exists(params_dir):
-            # params_path = params_dir / 'params.json'
-            # with params_path.open('r') as file:
-            #     saved_params = json.load(file)
             if is_unattacked_model:
                 model_name = f"{params['model']['name']}_{params['dataset']['name']}.pt"
                 model_path = params_dir / model_name

gnn_toolbox/experiment_handler/create_modules.py

@@ -21,7 +21,7 @@ def create_transforms(configs):
         transform_cls = get_from_registry("transform", name, registry)
         if transform_cls is None:
             raise ValueError(f"Transform '{name}' not found in the registry.")
-        logging.info(f"Creating transform '{name}' with parameters: {params}")
+        logging.debug(f"For dataset, creating transform '{name}' with parameters: {params}")
         transforms.append(transform_cls(**params))
 
     return Compose(transforms)

gnn_toolbox/experiment_handler/result_saver.py

@@ -22,9 +22,9 @@ def log_experiment_config(self):
             file_path = self.log_to / "experiment_config.json"
             with open(file_path, 'w') as f:
                 json.dump(self.experiment_config, f, indent=4)
-            logging.info("Experiment configuration logged successfully.")
+            logging.info(f"Experiment configuration logged to {file_path} successfully.")
         except Exception as e:
-            logging.error(f"Failed to log experiment configuration: {e}")
+            logging.error(f"Failed to log experiment configuration to {file_path}: {e}")
 
     def log_results(self):
         try:
@@ -35,9 +35,9 @@ def log_results(self):
                         json.dump(value, f, indent=2)
                     if self.csv_save and (key=='clean_result' or self.perturbed_result2csv):
                         self.save_to_csv(key, value)
-            logging.info("Results logged successfully.")
+            logging.info(f"Results logged successfully to {self.log_to}.")
         except Exception as e:
-            logging.error(f"Failed to log results: {e}")
+            logging.error(f"Failed to log results to {self.log_to}: {e}")
 
     def save_to_csv(self, key, value):
         try:
@@ -47,7 +47,7 @@ def save_to_csv(self, key, value):
                 writer.writerow(value[0].keys())
                 for result in value:
                     writer.writerow(result.values())
-            logging.info(f"{key} saved to CSV successfully.")
+            logging.info(f"{key} saved to CSV at {file_path} successfully.")
         except Exception as e:
-            logging.error(f"Failed to save {key} to CSV: {e}")
+            logging.error(f"Failed to save {key} to CSV at {file_path}: {e}")
 

requirements.txt

@@ -1,14 +1,14 @@
-numba==0.59.1
 numpy==1.26.4
 ogb==1.3.6
 pydantic==2.7.1
 PyYAML==6.0.1
 scipy==1.13.0
 tensorboardX==2.6.2.2
+--find-links https://download.pytorch.org/whl/torch_stable.html
 torch==2.2.1+cu118
-torch_geometric==2.5.3
 torch_scatter==2.1.2+pt22cu118
 torch_sparse==0.6.18+pt22cu118
+torch_geometric==2.5.3
 torchtyping==0.1.4
 tqdm==4.66.2
 typeguard==4.2.1