Update app.py

app.py

@@ -51,7 +51,7 @@ def respond(
     top_k,
     repeat_penalty,
 ):
-    chat_template = MessagesFormatterType.MISTRAL
+    # chat_template = MessagesFormatterType.MISTRAL
 
     global llm
     global llm_model
@@ -66,46 +66,40 @@ def respond(
         )
         llm_model = model
 
-    provider = LlamaCppPythonProvider(llm)
+    # provider = LlamaCppPythonProvider(llm)
 
-    agent = LlamaCppAgent(
-        provider,
-        system_prompt=f"{system_message}",
-        predefined_messages_formatter_type=chat_template,
-        debug_output=True
-    )
-    
-    settings = provider.get_provider_default_settings()
-    settings.temperature = temperature
-    settings.top_k = top_k
-    settings.top_p = top_p
-    settings.min_p = min_p
-    settings.max_tokens = max_tokens
-    settings.repeat_penalty = repeat_penalty
-    settings.stream = True
-
-    messages = BasicChatHistory()
-
-    for msn in history:
-        user = {
-            'role': Roles.user,
-            'content': msn[0]
-        }
-        assistant = {
-            'role': Roles.assistant,
-            'content': msn[1]
-        }
-        messages.add_message(user)
-        messages.add_message(assistant)
+    # agent = LlamaCppAgent(
+    #     provider,
+    #     system_prompt=f"{system_message}",
+    #     predefined_messages_formatter_type=chat_template,
+    #     debug_output=True
+    # )
     
-    print(history)
-    stream = agent.get_chat_response(
-        message,
-        llm_sampling_settings=settings,
-        chat_history=messages,
-        returns_streaming_generator=True,
-        print_output=False
-    )
+    # settings = provider.get_provider_default_settings()
+    # settings.temperature = temperature
+    # settings.top_k = top_k
+    # settings.top_p = top_p
+    # settings.min_p = min_p
+    # settings.max_tokens = max_tokens
+    # settings.repeat_penalty = repeat_penalty
+    # settings.stream = True
+
+    # messages = BasicChatHistory()
+    messages=[{"role":"system","content":system_message}]
+    chat=[{"role":"user","content":message})]
+    chat_b=[]
+
+    i=1
+    if history!=[]:
+        for msn in history:
+            if i%2==0:
+                messages.append({"role":"user","content":msn})
+            else:
+                messages.append({"role":"assistant","content":msn})
+            i+=1
+    messages+=chat
+    print(messages)
+    stream = llm.create_chat_completion(messages=messages,temperature = temperature,top_k = top_k,top_p = top_p,min_p = min_p,max_tokens = max_tokens,repeat_penalty = repeat_penalty,stream = True)
     
     outputs = ""
     for output in stream: