diff --git a/docs/api.rst b/docs/api.rst
index 1dc066926..23f81cfcd 100644
--- a/docs/api.rst
+++ b/docs/api.rst
@@ -31,7 +31,7 @@ Datasets
 .. autoclass:: torchani.data.CachedDataset
     :members:
 .. autofunction:: torchani.data.load_ani_dataset
-.. autoclass:: torchani.data.BatchedANIDataset
+.. autoclass:: torchani.data.PaddedBatchChunkDataset
 
 
 
diff --git a/torchani/data/__init__.py b/torchani/data/__init__.py
index e4dabe124..bc90fedca 100644
--- a/torchani/data/__init__.py
+++ b/torchani/data/__init__.py
@@ -7,7 +7,6 @@
 from ._pyanitools import anidataloader
 import torch
 from .. import utils
-import warnings
 from .new import CachedDataset, ShuffledDataset, find_threshold
 
 default_device = 'cuda' if torch.cuda.is_available() else 'cpu'
@@ -159,6 +158,13 @@ def split_whole_into_batches_and_chunks(atomic_properties, properties, batch_siz
 
 
 class PaddedBatchChunkDataset(Dataset):
+    r""" Dataset that contains batches in 'chunks', with padded structures
+
+    This dataset acts as a container of batches to be used when training. Each
+    of the batches is broken up into 'chunks', each of which is a tensor has
+    molecules with a smiliar number of atoms, but which have been padded with
+    dummy atoms in order for them to have the same tensor dimensions.
+    """
 
     def __init__(self, atomic_properties, properties, batch_size,
                  dtype=torch.get_default_dtype(), device=default_device):
@@ -193,26 +199,6 @@ def __len__(self):
         return len(self.batches)
 
 
-class BatchedANIDataset(PaddedBatchChunkDataset):
-    """Same as :func:`torchani.data.load_ani_dataset`. This API has been deprecated."""
-
-    def __init__(self, path, species_tensor_converter, batch_size,
-                 shuffle=True, properties=('energies',), atomic_properties=(), transform=(),
-                 dtype=torch.get_default_dtype(), device=default_device):
-        self.properties = properties
-        self.atomic_properties = atomic_properties
-        warnings.warn("BatchedANIDataset is deprecated; use load_ani_dataset()", DeprecationWarning)
-
-        atomic_properties, properties = load_and_pad_whole_dataset(
-            path, species_tensor_converter, shuffle, properties, atomic_properties)
-
-        # do transformations on data
-        for t in transform:
-            atomic_properties, properties = t(atomic_properties, properties)
-
-        super().__init__(atomic_properties, properties, batch_size, dtype, device)
-
-
 def load_ani_dataset(path, species_tensor_converter, batch_size, shuffle=True,
                      rm_outlier=False, properties=('energies',), atomic_properties=(),
                      transform=(), dtype=torch.get_default_dtype(), device=default_device,
@@ -361,4 +347,4 @@ def load_ani_dataset(path, species_tensor_converter, batch_size, shuffle=True,
     return tuple(ret)
 
 
-__all__ = ['load_ani_dataset', 'BatchedANIDataset', 'CachedDataset', 'ShuffledDataset', 'find_threshold']
+__all__ = ['load_ani_dataset', 'PaddedBatchChunkDataset', 'CachedDataset', 'ShuffledDataset', 'find_threshold']
diff --git a/torchani/utils.py b/torchani/utils.py
index 126d9407d..ac22621ff 100644
--- a/torchani/utils.py
+++ b/torchani/utils.py
@@ -172,9 +172,7 @@ def sae(self, species):
         return self_energies.sum(dim=1) + intercept
 
     def subtract_from_dataset(self, atomic_properties, properties):
-        """Transformer for :class:`torchani.data.BatchedANIDataset` that
-        subtract self energies.
-        """
+        """Transformer that subtracts self energies from a dataset"""
         if self.self_energies is None:
             self_energies = self.sae_from_dataset(atomic_properties, properties)
             self.self_energies = torch.tensor(self_energies, dtype=torch.double)