adjust for hf gradio

README.md

@@ -3,7 +3,7 @@ title: Insurance Advisor Agents PoC
 emoji: 🤖
 colorFrom: red
 colorTo: indigo
-sdk: docker
+sdk: gradio
 python: 3.11
 app_file: app_gui.py
 pinned: false

app_gui.py

@@ -1,12 +1,6 @@
 # Import Gradio for UI, along with other necessary libraries
 import gradio as gr
-from fastapi import FastAPI
-from rag_app.agents.react_agent import agent_executor, llm
-from rag_app.chains import user_response_sentiment_prompt
-# need to import the qa!
-
-app = FastAPI()
-user_sentiment_chain = user_response_sentiment_prompt | llm
+from rag_app.agents.react_agent import agent_executor
 
 
 if __name__ == "__main__":
@@ -21,13 +15,6 @@ if __name__ == "__main__":
     def bot(history):
         # Obtain the response from the 'infer' function using the latest input
         response = infer(history[-1][0], history)
-        #sources = [doc.metadata.get("source") for doc in response['source_documents']]
-        #src_list = '\n'.join(sources)
-        #print_this = response['result'] + "\n\n\n Sources: \n\n\n" + src_list
-
-
-        #history[-1][1] = print_this #response['answer']
-        # Update the history with the bot's response
         history[-1][1] = response['output']
         return history
 
@@ -35,10 +22,6 @@ if __name__ == "__main__":
     def infer(question, history):
         # Use the question and history to query the RAG model
         #result = qa({"query": question, "history": history, "question": question})
-        try:
-            data = user_sentiment_chain.invoke({"user_reponse":question})
-        except Exception as e:
-            raise e
         try:
             result = agent_executor.invoke(
                 {
@@ -46,17 +29,15 @@ if __name__ == "__main__":
                     "chat_history": history
                 }
             )
-            
-           
             return result
         except Exception:
             raise gr.Error("Model is Overloaded, Please retry later!")
         
     def vote(data: gr.LikeData):
         if data.liked:
-            print("You upvoted this response: " + data.value)
+            print("You upvoted this response: ")
         else:
-            print("You downvoted this response: " + data.value)
+            print("You downvoted this response: ")
 
     # CSS styling for the Gradio interface
     css = """
@@ -94,6 +75,4 @@ if __name__ == "__main__":
         clear.click(lambda: None, None, chatbot, queue=False)
 
     # Launch the Gradio demo interface
-    demo.queue().launch(share=False, debug=True)
-
-    app = gr.mount_gradio_app(app, demo, path="/")
+    demo.queue().launch(share=False, debug=True)

rag_app/agents/react_agent.py

@@ -7,17 +7,13 @@ from langchain.agents.output_parsers import ReActJsonSingleInputOutputParser
 from langchain.tools.render import render_text_description
 import os
 from dotenv import load_dotenv
-from rag_app.structured_tools.agent_tools import (
-    web_research, ask_user, get_email
+from rag_app.structured_tools.structured_tools import (
+    google_search, knowledgeBase_search
 )
 
 from langchain.prompts import PromptTemplate
 from rag_app.templates.react_json_with_memory_ger import template_system
 # from innovation_pathfinder_ai.utils import logger
-# from langchain.globals import set_llm_cache
-# from langchain.cache import SQLiteCache
-
-# set_llm_cache(SQLiteCache(database_path=".cache.db"))
 # logger = logger.get_console_logger("hf_mixtral_agent")
 
 config = load_dotenv(".env")
@@ -25,10 +21,6 @@ HUGGINGFACEHUB_API_TOKEN = os.getenv('HUGGINGFACEHUB_API_TOKEN')
 GOOGLE_CSE_ID = os.getenv('GOOGLE_CSE_ID')
 GOOGLE_API_KEY = os.getenv('GOOGLE_API_KEY')
 LLM_MODEL = os.getenv('LLM_MODEL')
-# LANGCHAIN_TRACING_V2 = "true"
-# LANGCHAIN_ENDPOINT = "https://api.smith.langchain.com"
-# LANGCHAIN_API_KEY = os.getenv('LANGCHAIN_API_KEY')
-# LANGCHAIN_PROJECT = os.getenv('LANGCHAIN_PROJECT')
 
 # Load the model from the Hugging Face Hub
 llm = HuggingFaceEndpoint(repo_id=LLM_MODEL, 
@@ -40,11 +32,10 @@ llm = HuggingFaceEndpoint(repo_id=LLM_MODEL,
 
 
 tools = [
-    #knowledgeBase_search,
-    #google_search,
-    web_research,
-    ask_user,
-    get_email
+    knowledgeBase_search,
+    google_search,
+    #web_research,
+    #ask_user
     ]
 
 prompt = PromptTemplate.from_template(