diff --git a/timm/models/factory.py b/timm/models/factory.py
index 40453380..d040a9ff 100644
--- a/timm/models/factory.py
+++ b/timm/models/factory.py
@@ -29,7 +29,6 @@ def create_model(
         scriptable=None,
         exportable=None,
         no_jit=None,
-        use_ml_decoder_head=False,
         **kwargs):
     """Create a model
 
@@ -81,10 +80,6 @@ def create_model(
     with set_layer_config(scriptable=scriptable, exportable=exportable, no_jit=no_jit):
         model = create_fn(pretrained=pretrained, **kwargs)
 
-    if use_ml_decoder_head:
-        from timm.models.layers.ml_decoder import add_ml_decoder_head
-        model = add_ml_decoder_head(model)
-
     if checkpoint_path:
         load_checkpoint(model, checkpoint_path)
 
diff --git a/train.py b/train.py
index 42985e12..10d839be 100755
--- a/train.py
+++ b/train.py
@@ -115,7 +115,6 @@ parser.add_argument('-b', '--batch-size', type=int, default=128, metavar='N',
                     help='input batch size for training (default: 128)')
 parser.add_argument('-vb', '--validation-batch-size', type=int, default=None, metavar='N',
                     help='validation batch size override (default: None)')
-parser.add_argument('--use-ml-decoder-head', type=int, default=0)
 
 # Optimizer parameters
 parser.add_argument('--opt', default='sgd', type=str, metavar='OPTIMIZER',
@@ -380,8 +379,7 @@ def main():
         bn_momentum=args.bn_momentum,
         bn_eps=args.bn_eps,
         scriptable=args.torchscript,
-        checkpoint_path=args.initial_checkpoint,
-        use_ml_decoder_head=args.use_ml_decoder_head)
+        checkpoint_path=args.initial_checkpoint)
     if args.num_classes is None:
         assert hasattr(model, 'num_classes'), 'Model must have `num_classes` attr if not set on cmd line/config.'
         args.num_classes = model.num_classes  # FIXME handle model default vs config num_classes more elegantly