Spaces:
Runtime error
Runtime error
| import streamlit as st | |
| from langchain_community.llms import HuggingFaceEndpoint | |
| from langchain_core.runnables import RunnablePassthrough | |
| from langchain_core.output_parsers import StrOutputParser | |
| from langchain.prompts import ChatPromptTemplate | |
| from PyPDF2 import PdfReader | |
| from langchain_text_splitters import RecursiveCharacterTextSplitter | |
| import os | |
| from langchain_community.vectorstores import Chroma | |
| from langchain.chains.question_answering import load_qa_chain | |
| from langchain.prompts import PromptTemplate | |
| from langchain_community.document_loaders import PyPDFLoader | |
| from langchain_chroma import Chroma | |
| from langchain_community.embeddings import HuggingFaceEmbeddings | |
| # Load model directly | |
| #from transformers import AutoModelForCausalLM | |
| access_token = os.getenv("HUGGINGFACE_API_KEY") | |
| #st.set_page_config(page_title="Document Genie", layout="wide") | |
| #st.markdown(""" | |
| ### PDFChat: Get instant insights from your PDF | |
| #This chatbot is built using the Retrieval-Augmented Generation (RAG) framework, leveraging Google's Generative AI model Gemini-PRO. It processes uploaded PDF documents by breaking them down into manageable chunks, creates a searchable vector store, and generates accurate answers to user queries. This advanced approach ensures high-quality, contextually relevant responses for an efficient and effective user experience. | |
| #### How It Works | |
| #Follow these simple steps to interact with the chatbot: | |
| #1. **Upload Your Document**: The system accepts a PDF file at one time, analyzing the content to provide comprehensive insights. | |
| #2. **Ask a Question**: After processing the document, ask any question related to the content of your uploaded document for a precise answer. | |
| #""") | |
| #def get_pdf(pdf_docs): | |
| # loader = PyPDFLoader(pdf_docs) | |
| # docs = loader.load() | |
| # return docs | |
| def get_pdf(pdf_docs): | |
| docs=[] | |
| for pdf in pdf_docs: | |
| temp_file = "./temp.pdf" | |
| # Delete the existing temp.pdf file if it exists | |
| if os.path.exists(temp_file): | |
| os.remove(temp_file) | |
| with open(temp_file, "wb") as file: | |
| file.write(pdf.getvalue()) | |
| file_name = pdf.name | |
| loader = PyPDFLoader(temp_file) | |
| docs.extend(loader.load()) | |
| return docs | |
| def text_splitter(text): | |
| text_splitter = RecursiveCharacterTextSplitter( | |
| # Set a really small chunk size, just to show. | |
| chunk_size=10000, | |
| chunk_overlap=500, | |
| separators=["\n\n","\n"," ",".",","]) | |
| chunks=text_splitter.split_documents(text) | |
| return chunks | |
| #GOOGLE_API_KEY = os.getenv("GOOGLE_API_KEY") | |
| #COHERE_API_KEY = os.getenv("COHERE_API_KEY") | |
| HUGGINGFACE_API_KEY = os.getenv("HUGGINGFACE_API_KEY") | |
| def get_conversational_chain(retriever): | |
| prompt_template = """ | |
| Given the following extracted parts of a long document and a question, create a final answer. | |
| Answer the question as detailed as possible from the provided context, make sure to provide all the details, if the answer is not in | |
| provided context just say, "answer is not available in the context", and then ignore the context and add the answer from your knowledge like a simple llm prompt. | |
| Try to give atleast the basic information.Donot return blank answer.\n\n | |
| Make sure to understand the question and answer as per the question. | |
| The answer should be a detailed one and try to incorporate examples for better understanding. | |
| If the question involves terms like detailed or explained , give answer which involves complete detail about the question.\n\n | |
| Context:\n {context}?\n | |
| Question: \n{question}\n | |
| Answer: | |
| """ | |
| #model = ChatGoogleGenerativeAI(model="gemini-pro", temperature=0.3, google_api_key=GOOGLE_API_KEY) | |
| repo_id ='google/gemma-1.1-2b-it' | |
| #repo_id='meta-llama/Meta-Llama-3-70B' | |
| #repo_id = 'mistralai/Mixtral-8x7B-Instruct-v0.1' | |
| llm = HuggingFaceEndpoint(repo_id=repo_id, temperature=0.3,token = access_token) | |
| #tokenizer = AutoTokenizer.from_pretrained("google/gemma-1.1-2b-it") | |
| #llm = AutoModelForCausalLM.from_pretrained("google/gemma-1.1-2b-it") | |
| #llm = AutoModelForCausalLM.from_pretrained("microsoft/Phi-3-mini-128k-instruct", trust_remote_code=True, token=access_token) | |
| #llm = pipeline("text-generation", model="google/gemma-1.1-2b-it") | |
| pt = ChatPromptTemplate.from_template(prompt_template) | |
| # Retrieve and generate using the relevant snippets of the blog. | |
| #retriever = db.as_retriever() | |
| rag_chain = ( | |
| {"context": retriever, "question": RunnablePassthrough()} | |
| | pt | |
| | llm | |
| | StrOutputParser() | |
| ) | |
| return rag_chain | |
| def embedding(chunk,query): | |
| #embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001") | |
| #embeddings = CohereEmbeddings(model="embed-english-v3.0") | |
| embeddings=HuggingFaceEmbeddings() | |
| db = Chroma.from_documents(chunk,embeddings) | |
| doc = db.similarity_search(query) | |
| print(doc) | |
| #Retrieve and generate using the relevant snippets of the blog. | |
| retriever = db.as_retriever() | |
| chain = get_conversational_chain(retriever) | |
| response = chain.invoke(query) | |
| print(response) | |
| return response | |
| #st.write("Reply: ", response["output_text"]) | |
| if 'messages' not in st.session_state: | |
| st.session_state.messages = [{'role': 'assistant', "content": 'Hello! Upload a PDF and ask me anything about its content.'}] | |
| st.header("Chat with your pdf💁") | |
| with st.sidebar: | |
| st.title("PDF FILE UPLOAD:") | |
| pdf_docs = st.file_uploader("Upload your PDF File and Click on the Submit & Process Button", accept_multiple_files=True, key="pdf_uploader") | |
| query = st.chat_input("Ask a Question from the PDF File") | |
| if query: | |
| raw_text = get_pdf(pdf_docs) | |
| text_chunks = text_splitter(raw_text) | |
| st.session_state.messages.append({'role': 'user', "content": query}) | |
| response = embedding(text_chunks,query) | |
| st.session_state.messages.append({'role': 'assistant', "content": response}) | |
| for message in st.session_state.messages: | |
| with st.chat_message(message['role']): | |
| st.write(message['content']) |