Update llama.py

unsloth/models/llama.py

@@ -683,7 +683,7 @@ pass
 @torch.inference_mode
 def LlamaModel_fast_forward_inference(
     self,
-    input_ids
+    input_ids,
     past_key_values,
 ):
     # Fix out of bounds tokenization
@@ -753,7 +753,7 @@ def LlamaForCausalLM_fast_forward(
 
     if past_key_value is not None and \
         hasattr(self.model.layers[0].self_attn, "paged_attention"):
-        
+
         outputs = LlamaModel_fast_forward_inference(
             self.model,
             input_ids,