Clearer documentation on how to set different loss weights

docs/src/architectures/alchemical-model.rst

@@ -59,6 +59,15 @@ hyperparameters to tune are (in decreasing order of importance):
   This hyperparameter controls the size and depth of the descriptors and the neural
   network. In general, increasing this might lead to better accuracy,
   especially on larger datasets, at the cost of increased training and evaluation time.
+- ``loss_weights``: This controls the weighting of different contributions to the loss
+  (e.g., energy, forces, virial, etc.). The default values work well for most datasets,
+  but they might need to be adjusted. For example, to set a weight of 1.0 for the energy
+  and 0.1 for the forces, you can set the following in the ``options.yaml`` file:
+  ```yaml
+  loss_weights:
+    energy: 1.0
+    forces: 0.1
+  ```
 
 
 Architecture Hyperparameters
@@ -110,7 +119,7 @@ bpnn
 
 training
 ########
-The parameters for the training loop are
+The hyperparameters for training are
 
 :param batch_size: batch size
 :param num_epochs: number of training epochs

docs/src/architectures/soap-bpnn.rst

@@ -99,7 +99,7 @@ bpnn
 
 training
 ########
-The parameters for the training loop are
+The parameters for training are
 
 :param batch_size: batch size
 :param num_epochs: number of training epochs
@@ -111,7 +111,7 @@ The parameters for the training loop are
     any case, the final summary will be per-structure.
 :param loss_weights: Specifies the weights to be used in the loss for each target. The
     weights should be a dictionary of floats, one for each target. All missing targets
-    are assigned a weight of 1.0.
+    are assigned a weight of 1.0. For example, for a 
 
 
 
@@ -148,7 +148,13 @@ hyperparameters to tune are (in decreasing order of importance):
   might need to be adjusted for specific datasets.
 - ``loss_weights``: This controls the weighting of different contributions to the loss
   (e.g., energy, forces, virial, etc.). The default values work well for most datasets,
-  but they might need to be adjusted.
+  but they might need to be adjusted. For example, to set a weight of 1.0 for the energy
+  and 0.1 for the forces, you can set the following in the ``options.yaml`` file:
+  ```yaml
+  loss_weights:
+    energy: 1.0
+    forces: 0.1
+  ```
 - ``layernorm``: Whether to use layer normalization before the neural network. Setting
   this hyperparameter to ``false`` will lead to slower convergence of training, but
   might lead to better generalization outside of the training set distribution.