import gradio as gr import os import time from langchain.document_loaders import OnlinePDFLoader #for laoding the pdf from langchain.embeddings import OpenAIEmbeddings # for creating embeddings from langchain.vectorstores import Chroma # for the vectorization part from langchain.chains import ConversationalRetrievalChain # for conversing with chatGPT from langchain.chat_models import ChatOpenAI # the LLM model we'll use (ChatGPT) def loading_pdf(): return "Loading..." def pdf_changes(pdf_doc, open_ai_key): if openai_key is not None: os.environ['OPENAI_API_KEY'] = open_ai_key #Load the pdf file loader = OnlinePDFLoader(pdf_doc.name) pages = loader.load_and_split() #Create an instance of OpenAIEmbeddings, which is responsible for generating embeddings for text embeddings = OpenAIEmbeddings() #To create a vector store, we use the Chroma class, which takes the documents (pages in our case), the embeddings instance, and a directory to store the vector data vectordb = Chroma.from_documents(pages, embedding=embeddings) #Finally, we create the bot using the ConversationalRetrievalChain class #A ConversationalRetrievalChain is similar to a RetrievalQAChain, except that the ConversationalRetrievalChain allows for #passing in of a chat history which can be used to allow for follow up questions. global pdf_qa pdf_qa = ConversationalRetrievalChain.from_llm(ChatOpenAI(temperature=0, model_name="gpt-4"), vectordb.as_retriever(), return_source_documents=False) return "Ready" else: return "Please provide an OpenAI API key" def add_text(history, text): history = history + [(text, None)] return history, "" def bot(history): response = infer(history[-1][0], history) history[-1][1] = "" for character in response: history[-1][1] += character time.sleep(0.05) yield history def infer(question, history): results = [] for human, ai in history[:-1]: pair = (human, ai) results.append(pair) chat_history = results print(chat_history) query = question result = pdf_qa({"question": query, "chat_history": chat_history}) print(result) return result["answer"] css=""" #col-container {max-width: 700px; margin-left: auto; margin-right: auto;} """ title = """

Chatbot for PDFs - GPT-4

Upload a .PDF, click the "Load PDF to LangChain" button,
Wait for the Status to show Ready, start typing your questions.
The app is set to store chat-history and is built on GPT-4

""" with gr.Blocks(css=css) as demo: with gr.Column(elem_id="col-container"): gr.HTML(title) with gr.Column(): openai_key = gr.Textbox(label="Your OpenAI API key", type="password") pdf_doc = gr.File(label="Load a pdf", file_types=['.pdf'], type="file") with gr.Row(): langchain_status = gr.Textbox(label="Status", placeholder="", interactive=False) load_pdf = gr.Button("Load PDF to LangChain") 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(loading_pdf, None, langchain_status, queue=False) load_pdf.click(pdf_changes, inputs=[pdf_doc, openai_key], outputs=[langchain_status], queue=False) question.submit(add_text, [chatbot, question], [chatbot, question]).then( bot, chatbot, chatbot ) submit_btn.click(add_text, [chatbot, question], [chatbot, question]).then( bot, chatbot, chatbot) demo.launch()