diff --git a/unsloth/models/llama.py b/unsloth/models/llama.py
index 4edff4df6e..c53f5600ce 100644
--- a/unsloth/models/llama.py
+++ b/unsloth/models/llama.py
@@ -673,7 +673,7 @@ def LlamaForCausalLM_fast_forward(
     *args, **kwargs,
 ) -> Union[Tuple, CausalLMOutputWithPast]:
 
-    if causal_mask is None and past_key_values is None:
+    if causal_mask is None:# and past_key_values is None:
         causal_mask = xformers.attn_bias.LowerTriangularMask()
 
     output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions