Update app.py

app.py

@@ -12,6 +12,12 @@ from app_utils import (
     get_chat_history, initialize_knowledge_base, 
     text_to_speech_gen, logging, buzz_user)
 
+global max_response_tokens
+global token_limit
+max_response_tokens = 500
+token_limit= 15000
+
+
 global FUNC_CALL #全局变量 用于判断角色动作
 FUNC_CALL = 0
 
@@ -27,6 +33,17 @@ AUDIO_HTML = ''
 nltk.download('averaged_perceptron_tagger')  #下载语料库
 conv_model, voice_model = initialize_knowledge_base()  #初始化知识库
 
+def num_tokens_from_messages(messages, model="gpt-3.5-turbo-16k"):
+    encoding = tiktoken.encoding_for_model(model)
+    num_tokens = 0
+    for message in messages:
+        num_tokens += 4  # every message follows <im_start>{role/name}\n{content}<im_end>\n
+        for key, value in message.items():
+            num_tokens += len(encoding.encode(value))
+            if key == "name":  # if there's a name, the role is omitted
+                num_tokens += -1  # role is always required and always 1 token
+    num_tokens += 2  # every reply is primed with <im_start>assistant
+    return num_tokens
 
 def idle_timer():
     global BUZZ_TIMEOUT
@@ -54,6 +71,7 @@ def get_response(history, audio_input):
 
     query_type = 'text'
     question =history[-1][0]
+    conv_history_tokens = 0
 
     global BUZZ_TIMEOUT
     BUZZ_TIMEOUT = 80
@@ -80,12 +98,18 @@ def get_response(history, audio_input):
     LOGGER.info("\ndocument_response: %s", answer)
     print('\ndocument_response:', answer)
 
+    conv_history_tokens = num_tokens_from_messages(conversation)
+    while (conv_history_tokens + max_response_tokens >= token_limit):
+        del MESSAGES[1] 
+        conv_history_tokens = num_tokens_from_messages(conversation)
+
     for trigger in GENERAL_RSPONSE_TRIGGERS:
         if trigger in answer:    
             MESSAGES.append({"role": "user", "content": question})
             chat = openai.ChatCompletion.create(
-                    model="gpt-3.5-turbo", 
+                    model="gpt-3.5-turbo-16k", 
                     messages=MESSAGES,
+                    max_tokens=500,
                     temperature=0.7,
                     n=128,
                     stop="\n"