diff --git a/apex/normalization/fused_layer_norm.py b/apex/normalization/fused_layer_norm.py
index 8558f7a5e..d873969f4 100644
--- a/apex/normalization/fused_layer_norm.py
+++ b/apex/normalization/fused_layer_norm.py
@@ -310,6 +310,7 @@ class FusedRMSNorm(torch.nn.Module):
     :attr:`normalized_shape`.
     :math:`\gamma` is a learnable affine transform parameter of
     :attr:`normalized_shape` if :attr:`elementwise_affine` is ``True``.
+    `epsilon` is added to the mean-square, then the root of the sum is taken.
 
     .. note::
         Unlike Batch Normalization and Instance Normalization, which applies