app.py

import gradio as gr
from langchain.llms import Replicate
from langchain.vectorstores import Pinecone
from langchain.text_splitter import CharacterTextSplitter
from langchain.document_loaders import PyPDFLoader
from langchain.llms import HuggingFaceHub
from langchain.vectorstores import Chroma
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.chains import ConversationalRetrievalChain
from datasets import load_dataset
import os


key = os.environ.get('API')
os.environ["REPLICATE_API_TOKEN"] = key

import sentence_transformers

def loading_pdf():
    return "Loading..."

def pdf_changes(pdf_doc):
    
    loader = PyPDFLoader(pdf_doc.name)
    documents = loader.load()
    text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
    texts = text_splitter.split_documents(documents)
    
    embeddings = HuggingFaceEmbeddings()

    db = Chroma.from_documents(texts, embeddings)
    retriever = db.as_retriever(search_kwargs={'k': 2})

    llm = Replicate(
        model="a16z-infra/llama13b-v2-chat:df7690f1994d94e96ad9d568eac121aecf50684a0b0963b25a41cc40061269e5",
        input={"temperature": 0.2, "max_length": 3000, "length_penalty":0.1, "num_beams":3}
    )
    global qa 
    qa = ConversationalRetrievalChain.from_llm(
        llm,
        retriever,
        return_source_documents=True
    )
    return "Ready"


def query(history, text):
    langchain_history = [(msg[1], history[i+1][1] if i+1 < len(history) else "") for i, msg in enumerate(history) if i % 2 == 0]
    result = qa({"question": text, "chat_history": langchain_history})
    new_history = history + [(text,result['answer'])]
    return new_history,""

css="""
#col-container {max-width: 700px; margin-left: auto; margin-right: auto;}
"""

title = """
<div style="text-align: center;max-width: 700px;">
    <h1>Chat with PDF</h1>   
"""


with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.HTML(title)
        with gr.Column():
            pdf_doc = gr.File(label="Load a PDF", file_types=['.pdf'], type="file")
            load_pdf = gr.Button("Load PDF")
            langchain_status = gr.Textbox(label="Status", placeholder="", interactive=False)
        chatbot = gr.Chatbot([], elem_id="chatbot").style(height=350)
        question = gr.Textbox(label="Question", placeholder="Type your question and hit Enter ")
        submit_btn = gr.Button("Send message") 
    load_pdf.click(pdf_changes, inputs=[pdf_doc], outputs=[langchain_status], queue=False)
    question.submit(query, [chatbot, question], [chatbot, question])
    submit_btn.click(query, [chatbot, question], [chatbot, question])

demo.launch()