diff --git a/keras_hub/src/models/gemma/gemma_attention.py b/keras_hub/src/models/gemma/gemma_attention.py
index 22b228f958..680685dd64 100644
--- a/keras_hub/src/models/gemma/gemma_attention.py
+++ b/keras_hub/src/models/gemma/gemma_attention.py
@@ -141,11 +141,8 @@ def _compute_attention(
                 query=q,
                 key=k,
                 value=v,
-                bias=None,
                 mask=attention_mask,
                 scale=query_normalization,
-                is_causal=True,
-                flash_attention=True,
             )
             return attention_output