Update recipe documentation (#465)

Fix docs Signed-off-by: Kirthi Shankar Sivamani <[email protected]>
NVIDIA · Oct 10, 2023 · 29b4670 · 29b4670
1 parent 92d1ba0
commit 29b4670
Showing 1 changed file with 0 additions and 3 deletions.
diff --git a/transformer_engine/common/recipe.py b/transformer_engine/common/recipe.py
@@ -116,9 +116,6 @@ def scaling_factor_compute(amax: Tensor,
 
           FP8_MAX = maximum_representable_value(fp8_format)
           new_scaling_factor = (FP8_MAX / amax) / (2 ^ margin)
-
-    * The scaling factor should always be a power of 2 to not introduce numerical
-      error during the conversion from FP8 to higher precision format.
     """
 
     margin: int = 0