Update app.py

app.py

@@ -1,65 +1,48 @@
 import gradio as gr
 import random
+
 import time
 import os
+
 import gradio as gr
 import numpy as np
+
 from langchain.chains import LLMChain
-from langchain import PromptTemplate
+
+from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
 import re
 import requests
 from bs4 import BeautifulSoup
+from langchain.chat_models import ChatOpenAI
+from langchain.agents import AgentType, Tool, initialize_agent
+from langchain.tools.render import format_tool_to_openai_function
+from langchain.agents.format_scratchpad import format_to_openai_function_messages
+from langchain.agents.output_parsers import OpenAIFunctionsAgentOutputParser
+from langchain.agents import AgentExecutor
+
+
 from langchain.schema import (
     SystemMessage,
     HumanMessage,
     AIMessage
 )
-from utils import ClaudeLLM
 
-personalities = ["You are Marcus Aurelius. You have the exact same traits. And you always quote him.", "I want you to act as a travel guide. I will write you my location and you will suggest a place to visit near my location. In some cases, I will also give you the type of places I will visit. You will also suggest me places of similar type that are close to my first location.", "I want you to act as an astrologer. You will learn about the zodiac signs and their meanings, understand planetary positions and how they affect human lives, be able to interpret horoscopes accurately, and share your insights with those seeking guidance or advice.", "I want you to act as a social media influencer. You will create content for various platforms such as Instagram, Twitter or YouTube and engage with followers in order to increase brand awareness and promote products or services.",]
+llm = ChatOpenAI(
+            temperature=0,
+            model='gpt-3.5-turbo-16k'
+        )
+
 
 
+personalities = ["You are a weather fact checker. You will check if the user prompts about the temperature in a certain city. You need to use the functions provided to you when needed.",]
+
 def user(user_message, history):
     return "", history + [[user_message, None]]
-def suggest(chatlog):
-  latest_msg = chatlog[-1][1]
-  llm = ClaudeLLM()
-  instruction = f"Generate 3 short responses for the following text message: {latest_msg}"
-
-  prompt = PromptTemplate(
-                input_variables=["instruction",],
-                template="""Generate 3 short replies that answer the text message. Keep your answer short and concise to the task.
-Here is an example:
-Message: Hi Marc, How are you doing today?
-Response:
-1- I'm good, how are you?
-2- I'm fine, thanks!
-3- Great, how about you?
-
-### Task
-{instruction}
-
-### Response:""",
-            )
-
-  chain = LLMChain(llm=llm, prompt = prompt)
-
-  response = chain.run(instruction = instruction,).strip()
-  suggestions = [ remove_numbers(resp.strip().replace('.', '').replace('-', '')).replace('\n', '') for resp in response.split('\n')]
-  print(suggestions)
-  if len(suggestions) < 3:
-    fillers = [''] * (3 - len(suggestions))
-    suggestions += fillers
-  if len(suggestions) > 3:
-    suggestions = suggestions[1:4]
-  
-  return suggestions
 
 def remove_numbers(question):
   return question.translate(str.maketrans('', '', '0123456789'))
 
-
-llm = ClaudeLLM()
+# llm = ClaudeLLM()
 
 def add_text(history, text):
 
@@ -68,103 +51,106 @@ def add_text(history, text):
 
     return history, ""
 
-
 def qa_retrieve(chatlog, index):
 
+    msgs = [[('assistant',chat[1]), ('user', chat[0])] for chat in chatlog[:-1]]
+    flat_msgs = [y for x in msgs for y in x]
+    flat_msgs = list([("system", personalities[0])] + flat_msgs + [("user", "{input}")] + [MessagesPlaceholder(variable_name="agent_scratchpad")])
+
+    print(flat_msgs)
+    print(type(flat_msgs))
+    msgs = flat_msgs
+    prompt = ChatPromptTemplate.from_messages(
+    msgs
+)
+
+    tools = [
+    Tool(
+        name="Search",
+        func= extract_temperature,
+        description="useful for when you want to retrieve temperature degrees in a certain city or country. Input should be in the form of a string containing the city or country provided.",
+    ),]
+    llm_with_tools = llm.bind(functions=[format_tool_to_openai_function(t) for t in tools])
+
+    agent = (
+    {
+        "input": lambda x: x["input"],
+        "agent_scratchpad": lambda x: format_to_openai_function_messages(
+            x["intermediate_steps"]
+        ),
+    }
+    | prompt
+    | llm_with_tools
+    | OpenAIFunctionsAgentOutputParser()
+    )
+
     print(f"Chatlog qa: {chatlog}")
     query = chatlog[-1][0]
-    #docs = ""
- 
-
-   # docs = db.similarity_search(query, k=4)
-    
-    #docs_page_content = " ".join([d.page_content for d in docs])
-    
-    
-    prompt = PromptTemplate(
-            input_variables=["query"],
-            template="""
-            Your name is Smartovate bot and you're being developed by Smartovate team. As a consultant, your role is to assist the user in various ways. You're a professional smart assistant and will help users formulating emails, giving ideas to productivity
-            and will always give your best help the users.
-
-            {query}
-    
-            """,
-        )
-    
-    
+    agent_executor = AgentExecutor(agent=agent, tools=tools, verbose=True)
+
+    gen = agent_executor.invoke(
+    {
+        "input": query
+    })
+
+    # prompt = PromptTemplate(
+    #         input_variables=["query"],
+    #         template="""
+
+    #         {personality}
+
+    #         {query}
+
+    #         """,
+    #     )
+
+
     # llm = BardLLM()
-    chain = LLMChain(llm=llm, prompt = prompt)
-    
-    response = chain.run(query=query,)
+    # chain = LLMChain(llm=llm, prompt = prompt, )
 
-    chatlog[-1][1] = response
+    # response = chain.run(query=query, personality = personalities[0])
+
+    chatlog[-1][1] = gen['output']
     return chatlog
-  
+
 def flush():
   global db
   db = ""
   return None
 
-
-#txt.submit(add_text, [chatbot, txt], [chatbot, txt]).then(
-  #  qa_retrieve, [chatbot], chatbot)#
-#btn.upload(add_file, [chatbot, btn], [chatbot,], batch = True).then(qa_retrieve, [chatbot], chatbot)
-
-#send_btn.click(add_text, [chatbot, txt, ], [chatbot, txt]).then(
- #   qa_retrieve, [chatbot, ], chatbot)
-
-#clear.click(flush, None, outputs = chatbot, queue=False)
-
-#demo.queue(concurrency_count = 4)
-#demo.launch()
-
 with gr.Blocks(css = """#white-button {
   background-color: #FFFFFF;
   color: #000000;
 }
-
 #orange-button-1 {
   background-color: #FFDAB9;
   color: #000000;
-
-
 }
-
 #orange-button-2 {
   background-color: #FFA07A;
   color: #FFFFFF;
-
 }
-
 #orange-button-3 {
   background-color: #FF4500;
   color: #FFFFFF;
 }""", theme=gr.themes.Soft()) as demo:
     chatbot = gr.Chatbot().style(height=750)
 
-    with gr.Row():
-        with gr.Column():
-            marcus = gr.Button("Suggestion 1", elem_id = "white-button")
-        with gr.Column(min_width=0):
-            travel_guide = gr.Button("Suggestion 2", elem_id = "orange-button-1")
-        with gr.Column(min_width=0):
-            astrologer = gr.Button("Suggestion 3", elem_id = "orange-button-2")
-
     with gr.Row():
       with gr.Column(scale = 0.75, min_width=0):
         msg = gr.Textbox(placeholder = "Enter text and press enter",show_label=False).style(container = False)
       with gr.Column(scale = 0.25, min_width=0):
             clear = gr.Button("Clear")
 
-    index = gr.Textbox(value = "0", visible = False)      
+    index = gr.Textbox(value = "0", visible = False)
     msg.submit(user, [msg, chatbot], [msg, chatbot], queue=False).then(
       qa_retrieve, [chatbot, index], chatbot
-      ).then(suggest, chatbot, [marcus, travel_guide, astrologer, ])
-    
-    marcus.click(lambda x: x, marcus, msg)
-    travel_guide.click(lambda x: x, travel_guide, msg)
-    astrologer.click(lambda x: x, astrologer, msg)
+
+      )
+
+    # marcus.click(lambda x: x, marcus, msg)
+    # travel_guide.click(lambda x: x, travel_guide, msg)
+    # astrologer.click(lambda x: x, astrologer, msg)
     clear.click(lambda: None, None, chatbot, queue=False)
 
 demo.launch()