From 34e010c45b5ca1eb255e0d42ae8c7c05ad0d9fee Mon Sep 17 00:00:00 2001
From: Gregory Shtrasberg <156009573+gshtras@users.noreply.github.com>
Date: Mon, 3 Jun 2024 18:19:47 -0400
Subject: [PATCH] Update linear.py

Fix bias handling with tgemm
---
 vllm/model_executor/layers/linear.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/vllm/model_executor/layers/linear.py b/vllm/model_executor/layers/linear.py
index 1d2a8782a96f1..6a0f55101e660 100644
--- 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)