minor comment fix

Signed-off-by: Sage Moore <[email protected]>
vllm-project · Dec 9, 2024 · 6915fa2 · 6915fa2
1 parent e2fda7f
commit 6915fa2
Showing 1 changed file with 0 additions and 2 deletions.
diff --git a/vllm/compilation/activation_quant_fusion.py b/vllm/compilation/activation_quant_fusion.py
@@ -53,8 +53,6 @@ class ActivationQuantFusionPass(VllmInductorPass):
     """
     This pass fuses a pre-defined set of custom ops into fused ops.
     It uses the torch pattern matcher to find the patterns and replace them.
-    It also manually processes multi-output matches, as those are broken in
-    the torch pattern matcher.
 
     Because patterns can only be registered once, the pass is a singleton.
     This will be addressed in a future version of PyTorch: