add some documentation for transforms

mu_map/dataset/transform.py

@@ -5,13 +5,26 @@ from torch import Tensor
 
 
 class Transform:
+    """
+    Interface of a transformer. A transformer can be initialized and then applied to
+    an input tensor and expected output tensor as returned by a dataset. It can be
+    used for normalization and data augmentation.
+    """
+
     def __call__(
         self, inputs: Tensor, outputs_expected: Tensor
     ) -> Tuple[Tensor, Tensor]:
+        """
+        Apply the transformer to a pair of inputs and expected outputs in a dataset.
+        """
         return inputs, outputs_expected
 
 
 class SequenceTransform(Transform):
+    """
+    A transformer that applies a sequence of transformers sequentially.
+    """
+
     def __init__(self, transforms: List[Transform]):
         self.transforms = transforms