Add docstring to custom Sequential

Co-authored-by: Copilot <[email protected]>

Author: LarsKue

bayesflow/networks/sequential/sequential.py

@@ -7,6 +7,29 @@
 
 @serializable("bayesflow.networks")
 class Sequential(keras.Layer):
+    """
+    A custom sequential model for managing a sequence of Keras layers.
+
+    This class extends `keras.Layer` and provides functionality for building,
+    calling, and serializing a sequence of layers. Unlike `keras.Sequential`,
+    this implementation allows for more flexibility in handling layer arguments
+    and supports custom serialization through the `@serializable` decorator.
+
+    Parameters
+    ----------
+    layers : keras.Layer or Sequence[keras.Layer]
+        A single Keras layer or a sequence of Keras layers to be managed by this model.
+    **kwargs : dict
+        Additional keyword arguments passed to the base `keras.Layer` class.
+
+    Notes
+    -----
+    - This class differs from `keras.Sequential` in that it does not assume a strict
+      linear stack of layers and provides custom methods for serialization and
+      configuration.
+    - It is designed to integrate with the BayesFlow framework and supports
+      additional utilities like `layer_kwargs`.
+    """
     def __init__(self, *layers: keras.Layer | Sequence[keras.Layer], **kwargs):
         super().__init__(**layer_kwargs(kwargs))
         if len(layers) == 1 and isinstance(layers[0], Sequence):