Merge pull request #756 from Aske-Rosted/train_example_comments

Train example comments

examples/04_training/01_train_dynedge.py

@@ -81,10 +81,17 @@ def main(
         # Log configuration to W&B
         wandb_logger.experiment.config.update(config)
 
-    # Define graph representation
+    # Define graph/data representation, here the KNNGraph is used.
+    # The KNNGraph is a graph representation, which uses the
+    # KNNEdges edge definition with 8 neighbours as default.
+    # The graph representation is defined by the detector,
+    # in this case the Prometheus detector.
+    # The standard node definition is used, which is NodesAsPulses.
     graph_definition = KNNGraph(detector=Prometheus())
 
-    # Use GraphNetDataModule to load in data
+    # Use GraphNetDataModule to load in data and create dataloaders
+    # The input here depends on the dataset being used,
+    # in this case the Prometheus dataset.
     dm = GraphNeTDataModule(
         dataset_reference=config["dataset_reference"],
         dataset_args={
@@ -110,17 +117,28 @@ def main(
 
     # Building model
 
+    # Define architecture of the backbone, in this example
+    # the DynEdge architecture is used.
+    # https://iopscience.iop.org/article/10.1088/1748-0221/17/11/P11003
     backbone = DynEdge(
         nb_inputs=graph_definition.nb_outputs,
         global_pooling_schemes=["min", "max", "mean", "sum"],
     )
+    # Define the task.
+    # Here an energy reconstruction, with a LogCoshLoss function.
+    # The target and prediction are transformed using the log10 function.
+    # When infering the prediction is transformed back to the
+    # original scale using 10^x.
     task = EnergyReconstruction(
         hidden_size=backbone.nb_outputs,
         target_labels=config["target"],
         loss_function=LogCoshLoss(),
         transform_prediction_and_target=lambda x: torch.log10(x),
         transform_inference=lambda x: torch.pow(10, x),
     )
+    # Define the full model, which includes the backbone, task(s),
+    # along with typical machine learning options such as
+    # learning rate optimizers and schedulers.
     model = StandardModel(
         graph_definition=graph_definition,
         backbone=backbone,