ExternalVendorCode/lollms-webui
1
0
Fork 0
mirror of https://github.com/ParisNeo/lollms-webui.git synced 2025-01-29 15:44:12 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

use only the last 5 messages as the prompt to fix the issue of "too much token". also make it a little bit faster when showing the generated new messages.

Browse Source
This commit is contained in:
Jerry 2023-04-11 18:29:10 +08:00
parent 48934ffb01
commit 6b80ae107c
1 changed files with 24 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
app.py
View File

@ -35,7 +35,8 @@ class Gpt4AllWebUI:
# workaround for non interactive mode
self.full_message = ""
self.full_message_list = []
self.prompt_message = ""
# This is the queue used to stream text to the ui as the bot spits out its response
self.text_queue = Queue(0)
@ -124,7 +125,7 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
"conditionner", conditionning_message, DiscussionsDB.MSG_TYPE_CONDITIONNING,0
)
self.full_message_list.append(conditionning_message)
# self.prepare_query(conditionning_message)
@ -144,11 +145,11 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
# print(f"Bot said:{self.bot_says}")
def prepare_query(self, message):
def prepare_query(self):
self.bot_says = ""
self.full_text = ""
self.is_bot_text_started = False
self.current_message = message
#self.current_message = message
def new_text_callback(self, text: str):
print(text, end="")
@ -158,7 +159,8 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
self.bot_says += text
self.full_message += text
self.text_queue.put(text)
if self.current_message in self.full_text:
#if self.current_message in self.full_text:
if len(self.prompt_message) <= len(self.full_text):
self.is_bot_text_started = True
def add_endpoint(
@ -202,7 +204,7 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
gc.collect()
self.chatbot_bindings.generate(
self.full_message,#self.current_message,
self.prompt_message,#self.full_message,#self.current_message,
new_text_callback=self.new_text_callback,
n_predict=len(self.current_message)+self.args.n_predict,
temp=self.args.temp,
@ -239,7 +241,13 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
self.current_message = "\nUser: " + message + "\nGPT4All: "
self.full_message += self.current_message
self.prepare_query(self.full_message)
self.full_message_list.append(self.current_message)
if len(self.full_message_list) > 5:
self.prompt_message = '\n'.join(self.full_message_list[-5:])
else:
self.prompt_message = self.full_message
self.prepare_query()
self.generating = True
app.config['executor'].submit(self.generate_message)
while self.generating or not self.text_queue.empty():
@ -252,6 +260,7 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
self.current_discussion.update_message(response_id, self.bot_says)
self.full_message_list.append(self.bot_says)
#yield self.bot_says# .encode('utf-8').decode('utf-8')
# TODO : change this to use the yield version in order to send text word by word
@ -286,8 +295,13 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
return "renamed successfully"
def restore_discussion(self, full_message):
self.prompt_message = full_message
if len(self.full_message_list)>5:
self.prompt_message = "\n".join(self.full_message_list[-5:])
self.chatbot_bindings.generate(
full_message,
self.prompt_message,#full_message,
new_text_callback=self.new_text_callback,
n_predict=0,#len(full_message),
temp=self.args.temp,
@ -305,8 +319,10 @@ GPT4All:Welcome! I'm here to assist you with anything you need. What can I do fo
messages = self.current_discussion.get_messages()
self.full_message = ""
self.full_message_list = []
for message in messages:
self.full_message += message['sender'] + ": " + message['content'] + "\n"
self.full_message_list.append(message['sender'] + ": " + message['content'])
app.config['executor'].submit(self.restore_discussion, self.full_message)
return jsonify(messages)