Update vllm/model_executor/layers/fused_moe/shared_fused_moe.py

Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Signed-off-by: Sage Moore <sagemoore@utexas.edu>

Author: SageMoore

vllm/model_executor/layers/fused_moe/shared_fused_moe.py

@@ -28,8 +28,8 @@ def __init__(
         super().__init__(**kwargs)
         self._shared_experts = shared_experts
 
-        # Disable shared expert overlap if we are using eplb, because there
-        # are correctness issues, or not using flashinfer + DP since there
+        # Disable shared expert overlap if we are using eplb, because of
+        # correctness issues, or if using flashinfer with DP, since there
         # is nothing to be gained in this case. Disabling the overlap
         # optimization also prevents the shared experts from being hidden
         # from torch.compile.