src/chatbot.py

from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_core.prompts import ChatPromptTemplate
from langchain_community.llms.huggingface_hub import HuggingFaceHub
from langchain_community.embeddings import HuggingFaceEmbeddings

from src.vectordatabase import RAG, get_vectorstore
import pandas as pd
import os
#from dotenv import load_dotenv, find_dotenv

#Load environmental variables from .env-file
#load_dotenv(find_dotenv())



embeddings = HuggingFaceEmbeddings(model_name="paraphrase-multilingual-MiniLM-L12-v2")
llm = HuggingFaceHub(
    # Try different model here
    repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1",
    # repo_id="CohereForAI/c4ai-command-r-v01", # too large 69gb
    # repo_id="CohereForAI/c4ai-command-r-v01-4bit", # too large 22 gb
    # repo_id="meta-llama/Meta-Llama-3-8B", # too large 16 gb
    task="text-generation",
    model_kwargs={
        "max_new_tokens": 512,
        "top_k": 30,
        "temperature": 0.1,
        "repetition_penalty": 1.03,
        }
        #,huggingfacehub_api_token

)
# To Do: Experiment with different templates replying in german or english depending on the input language
prompt1 = ChatPromptTemplate.from_template("""<s>[INST] 
                    Instruction: Beantworte die folgende Frage auf deutsch und nur auf der Grundlage des angegebenen Kontexts:

                    Context: {context}

                    Question: {input}  
                    [/INST]"""
                    # Returns the answer in English!?
) 

prompt2 = ChatPromptTemplate.from_template("""Beantworte die folgende Frage auf deutsch und nur auf der Grundlage des angegebenen Kontexts:

        <context>
        {context}
        </context>

        Frage: {input}
        """
        # Returns the answer in German
)

 
folder_path = "./src/FAISS"
#index_name = "speeches_1949_09_12"
index_name = "legislature20"
db = get_vectorstore(embeddings=embeddings, folder_path=folder_path, index_name=index_name)

def chatbot(message, history, db=db, llm=llm, prompt=prompt2):
    raw_response = RAG(llm=llm, prompt=prompt, db=db, question=message)
    response = raw_response['answer'].split("Antwort: ")[1]
    return response  

# Retrieve speech contents based on keywords
def keyword_search(query, db=db, embeddings=embeddings):
    query_embedding = embeddings.embed_query(query)
    results =  db.max_marginal_relevance_search_with_score_by_vector(query_embedding)
    # Format vector store query results into dataframe
    #print(results[0][0].metadata.keys())
    
    df_res = pd.DataFrame(columns=['Speech Content', 'Relevance']) # Add Date/Party/Politician
    for doc in results:
        speech_content = doc[0].page_content
        #speech_date = doc[0].metadata["date"]
        score = doc[1] # Relevance based on relevance search
        df_res = pd.concat([df_res, pd.DataFrame({'Speech Content': [speech_content],
                                                          #'Date': [speech_date],
                                                          'Relevance': [score]})], ignore_index=True)
    
    df_res.sort_values('Relevance', inplace=True)
    return df_res