Update app.py

app.py

@@ -99,12 +99,13 @@ def vicuna_model():
     return model, tokenizer 
 
 def hermes_model():
-    model_name_or_path = "TheBloke/CapybaraHermes-2.5-Mistral-7B-AWQ"
-    tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
-    model = AutoModelForCausalLM.from_pretrained(model_name_or_path,low_cpu_mem_usage=True,device_map="auto")
+    
+    tokenizer = AutoTokenizer.from_pretrained("TheBloke/CapybaraHermes-2.5-Mistral-7B-AWQ")
+    model = AutoModelForCausalLM.from_pretrained("TheBloke/CapybaraHermes-2.5-Mistral-7B-AWQ",low_cpu_mem_usage=True,device_map="auto")
 
     return model, tokenizer
 
+
 model, tokenizer = hermes_model()
 
 def hermes_generate_response(msg_prompt: str) -> dict: