Fix a comment

Signed-off-by: Kamil Tokarski <[email protected]>

dali/python/nvidia/dali/auto_aug/augmentations.py

@@ -121,9 +121,6 @@ def contrast_mean_centered(sample, parameter):
     """
     This variant follows PIL implementation of Contrast enhancement, which does not use
     the middle of the `dtype` range as the contrast center, but a channel-weighted mean.
-    NOTE: Currently, it is not possible to run this variant with `sample` that resides
-    in GPU memory due to reduction values residing in GPU and the limitation that named arguments
-    (i.e. contrast_center) must reside in CPU memory.
     """
     mean = fn.reductions.mean(sample, axes=[0, 1])
     rgb_weights = types.Constant(np.array([0.299, 0.587, 0.114], dtype=np.float32))