Spaces:
Sleeping
Sleeping
import os | |
import langchain | |
from langchain.text_splitter import RecursiveCharacterTextSplitter | |
from langchain.document_loaders import TextLoader | |
from langchain.vectorstores import FAISS | |
from langchain.embeddings import HuggingFaceBgeEmbeddings | |
from langchain.schema import Document | |
import streamlit as st | |
from langchain_groq import ChatGroq | |
from langchain.text_splitter import CharacterTextSplitter | |
from langchain.prompts import PromptTemplate | |
from langchain_core.output_parsers import StrOutputParser | |
from langchain.schema.runnable import RunnablePassthrough | |
text_splitter = RecursiveCharacterTextSplitter(chunk_size=75, chunk_overlap=0, separators=["\n",'. ']) | |
loader = TextLoader('data.txt') | |
docs = loader.load() | |
split_docs = text_splitter.split_documents(docs) | |
# Iterate over the Document objects and extract the content and metadata using the appropriate attributes | |
split_docss = [Document(page_content=t.page_content, metadata=t.metadata) for t in split_docs] | |
embeddings = HuggingFaceBgeEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2") | |
vectorstore = FAISS.from_documents(split_docss,embeddings) | |
retriever=vectorstore.as_retriever() | |
retriever.search_kwargs['k'] = 5 | |
os.environ['GROQ_API_KEY']='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' | |
def preprocess(text): | |
return "\n".join(x.page_content for x in text) | |
def format_docs(docs): | |
return "\n".join(doc.page_content for doc in docs) | |
def printer(text): | |
print(text) | |
return text | |
llm = ChatGroq(temperature=0, model_name="mixtral-8x7b-32768", streaming=True, verbose=False) | |
# llama3-8b-8192 | |
prompt='''You are a coupon recommender chatbot. | |
Understand the question with the category of the item and the location: | |
{summary} | |
and answer with only relevant coupons: | |
{context} | |
Instructions: | |
List only the coupons that match the category of the item along with the location in the question.Use three sentences maximum and keep the answer concise. | |
''' | |
template=PromptTemplate(template=prompt,input_variables=['summary','context']) | |
rag_chain=({ "summary": RunnablePassthrough(), "context": retriever | format_docs , | |
} | template|printer| llm| StrOutputParser()) | |
st.set_page_config(page_title="Lloyds Mobile Rewards Section", layout="wide") | |
st.title("Mock Rewards Section") | |
st.header("Welcome to Your Rewards") | |
st.write("Find exclusive offers and discounts just for you!") | |
st.sidebar.title("Check your eligibility to turn on the discover mode!") | |
# category = st.sidebar.selectbox("Select Category", ["All", "Electronics", "Food", "Groceries"]) | |
if 'count' not in st.session_state: | |
st.session_state.count = 0 | |
enable_controls = st.session_state.count == 5 | |
def increment_count(): | |
st.session_state.count += 1 | |
st.sidebar.title(f'Number of transactions : {st.session_state.count}') | |
# st.write(f"Button has been pressed {st.session_state.count} times") | |
button=st.sidebar.button("Mock Transaction") | |
if button: | |
increment_count() | |
response=None | |
# Display the updated count | |
if enable_controls: | |
st.sidebar.write("You are now eligible for discover mode ") | |
user_input = st.sidebar.text_input("Enter your request or choice of category", key='user_input') | |
response=retriever.get_relevant_documents(query=st.session_state.user_input) | |
st.write(f'{response}') | |
if not enable_controls: | |
st.sidebar.write(f"Keep going {20-st.session_state.count} times.") | |