Update linear.py

Fix bias handling with tgemm
prarit · Jun 3, 2024 · 34e010c · 34e010c
1 parent f3e1926
commit 34e010c
Showing 1 changed file with 2 additions and 0 deletions.
diff --git a/vllm/model_executor/layers/linear.py b/vllm/model_executor/layers/linear.py
@@ -104,6 +104,8 @@ def apply_weights(self,
             if bias is not None:
                 return F.linear(x, weight) + bias
             return F.linear(x, weight)
+        elif bias is not None:
+            return F.linear(x, weight, bias)
         return tgemm.mm(x, weight)