diff --git a/server/text_generation_server/models/custom_modeling/flash_qwen2_modeling.py b/server/text_generation_server/models/custom_modeling/flash_qwen2_modeling.py
index 849dd80e..42a2cbde 100644
--- a/server/text_generation_server/models/custom_modeling/flash_qwen2_modeling.py
+++ b/server/text_generation_server/models/custom_modeling/flash_qwen2_modeling.py
@@ -26,7 +26,7 @@ def load_attention(config, prefix, weights):
             prefixes=[f"{prefix}.q_proj", f"{prefix}.k_proj", f"{prefix}.v_proj"],
             dim=0,
             weights=weights,
-            bias=False,
+            bias=True,
         )