Update examples/flax/language-modeling/README.md

Co-authored-by: Steven Liu <[email protected]>
huggingface · Dec 4, 2024 · d7fc2d1 · d7fc2d1
1 parent b95143f
commit d7fc2d1
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/examples/flax/language-modeling/README.md b/examples/flax/language-modeling/README.md
@@ -565,4 +565,4 @@ By changing the dtype for `FlaxBertModel `to `jax.numpy.bfloat16`, you get the p
 import jax
 model = FlaxBertModel.from_pretrained("bert-base-uncased", config=config, dtype=jax.numpy.bfloat16)
 ```
-On a AWS c7i.4xlarge with Intel Sapphire Rapids, we get  > 2X speedup by changing precision from float32 to bfloat16.
+Switching from float32 to bfloat16 can increase the speed of an AWS c7i.4xlarge with Intel Sapphire Rapids by more than 2x.