diff --git a/unsloth/models/llama.py b/unsloth/models/llama.py
index 852388e4f8..bcc26b36a5 100644
--- a/unsloth/models/llama.py
+++ b/unsloth/models/llama.py
@@ -487,7 +487,6 @@ def LlamaModel_fast_forward(
         inputs_embeds = self.embed_tokens(input_ids)
 
     # Ignore attention_mask
-    print(attention_mask)
     if attention_mask is None:
         padding_mask = None
     elif False:#self.training: