diff --git a/timm/data/dataset.py b/timm/data/dataset.py
index 5f5fc9e64b..1c481bffa5 100644
--- a/timm/data/dataset.py
+++ b/timm/data/dataset.py
@@ -30,13 +30,15 @@ def __init__(
             input_img_mode='RGB',
             transform=None,
             target_transform=None,
+            **kwargs,
     ):
         if reader is None or isinstance(reader, str):
             reader = create_reader(
                 reader or '',
                 root=root,
                 split=split,
-                class_map=class_map
+                class_map=class_map,
+                **kwargs,
             )
         self.reader = reader
         self.load_bytes = load_bytes
diff --git a/timm/data/readers/reader_hfds.py b/timm/data/readers/reader_hfds.py
index 3dd6dd5d26..7784660676 100644
--- a/timm/data/readers/reader_hfds.py
+++ b/timm/data/readers/reader_hfds.py
@@ -35,7 +35,7 @@ def __init__(
             root: Optional[str] = None,
             split: str = 'train',
             class_map: dict = None,
-            image_key: str = 'image',
+            input_key: str = 'image',
             target_key: str = 'label',
             download: bool = False,
     ):
@@ -50,9 +50,9 @@ def __init__(
             cache_dir=self.root,  # timm doesn't expect hidden cache dir for datasets, specify a path
         )
         # leave decode for caller, plus we want easy access to original path names...
-        self.dataset = self.dataset.cast_column(image_key, datasets.Image(decode=False))
+        self.dataset = self.dataset.cast_column(input_key, datasets.Image(decode=False))
 
-        self.image_key = image_key
+        self.image_key = input_key
         self.label_key = target_key
         self.remap_class = False
         if class_map: