fix arguments

app.py

@@ -60,7 +60,7 @@ def load_model():
     generator_mini = pipeline(task="text-generation", model=ankerbot_model, tokenizer=ankerbot_tokenizer, torch_dtype=torch.float16, attn_implementation="eager", use_flash_attention_2=False) # True for flash-attn2 else False
     return (generator, generator_mini)
     
-model_cache = None
+_model_cache = None
 
 @spaces.GPU
 def get_model():
@@ -159,7 +159,6 @@ def search_qdrant_with_context(query_text, collection_name, top_k=3):
 def respond(
     query,
     history: list[tuple[str, str]],
-    system_message,
     max_tokens,
     temperature,
     top_p,
@@ -184,7 +183,7 @@ def respond(
         colleciton_name = "products"
 
     context = search_qdrant_with_context(query + " " + refined_context[0]["generated_text"].split("assistant\n").pop(), collection_name)
-    answer = generate_response(query, context, max_tokens, temperature, top_p, generator[0])
+    answer = generate_response(query, context, last_messages, max_tokens, temperature, top_p, generator[0])
     full_conv = f"Nutzer:{query};Assistent:{answer}"
     if len(last_messages) > 5:
         last_messages.pop(0)
@@ -207,6 +206,7 @@ demo = gr.ChatInterface(
             label="Top-p (nucleus sampling)",
         ),
     ],
+    chatbot=gr.Chatbot(type="messages"),
 )