pages/2_Twitter_GPT_Search.py

from langchain.embeddings import HuggingFaceEmbeddings,HuggingFaceInstructEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import FAISS
from langchain.chat_models.openai import ChatOpenAI
from langchain import VectorDBQA
import pandas as pd
from variables import *

from langchain.chat_models import ChatOpenAI
from langchain.prompts.chat import (
    ChatPromptTemplate,
    SystemMessagePromptTemplate,
    AIMessagePromptTemplate,
    HumanMessagePromptTemplate,
)
from langchain.schema import (
    AIMessage,
    HumanMessage,
    SystemMessage
)

from datetime import datetime as dt


if 'tlist' not in st.session_state:
    st.session_state['tlist'] = ''

system_template="""Use the following pieces of context to answer the users question. 
If you don't know the answer, just say that you don't know, don't try to make up an answer.
ALWAYS return a "SOURCES" part in your answer.
The "SOURCES" part should be a reference to the source of the document from which you got your answer.

Example of your response should be:

```
The answer is foo
SOURCES: xyz
```

Begin!
----------------
{context}"""
messages = [
    SystemMessagePromptTemplate.from_template(system_template),
    HumanMessagePromptTemplate.from_template("{question}")
]
prompt = ChatPromptTemplate.from_messages(messages)

current_time = dt.strftime(dt.today(),'%d_%m_%Y_%H_%M')

st.markdown("## Financial Tweets GPT Search")

twitter_link = """
[![](https://img.shields.io/twitter/follow/nickmuchi?label=@nickmuchi&style=social)](https://twitter.com/nickmuchi)
"""

st.markdown(twitter_link)

bi_enc_dict = {'mpnet-base-v2':"sentence-transformers/all-mpnet-base-v2",
              'instructor-base': 'hkunlp/instructor-base'}

search_input = st.text_input(
        label='Enter Your Search Query',value= "What are the most topical risks?", key='search')
        
sbert_model_name = st.sidebar.selectbox("Embedding Model", options=list(bi_enc_dict.keys()), key='sbox')

tweets = ','.join(st.session_state['tlist'])

try:

    if search_input:
        
        model = bi_enc_dict[sbert_model_name]
                        
        with st.spinner(
            text=f"Loading {model} embedding model and Generating Response..."
        ):

            vectorstore = create_vectorstore(tweets,model)
            
            tweets = embed_tweets(search_input,prompt,vectorstore)


        references = [doc.page_content for doc in tweets['source_documents']]

        answer = tweets['result']
            
        ##### Sematic Search #####
        
        with st.expander(label='Query Result', expanded=True):
            st.write(answer)
            
        with st.expander(label='References from Corpus used to Generate Result'):
            for ref in references:
                st.write(ref)
            
    else:
    
        st.write('Please ensure you have entered the YouTube URL or uploaded the Earnings Call file')  
        
except RuntimeError:
  
    st.write('Please ensure you have entered the YouTube URL or uploaded the Earnings Call file')