added another argument in output_ids

app.py

@@ -37,10 +37,11 @@ def llama_generation(input_text):
                                        return_tensors='pt').to('cuda')
     
     # llama generation looks for the numeric vectors not the tensors so there is no need for **input_ids rather just input_ids
-    output_ids = llama_model.generate(input_ids=input_ids) 
+    output_ids = llama_model.generate(input_ids=input_ids,
+                                      max_new_tokens=400) 
 
     # Decode
-    output_text = llama_tokenizer.decode(output_ids,
+    output_text = llama_tokenizer.decode(output_ids[0],
                                          skip_special_tokens=True)
     
     return output_text