File size: 3,346 Bytes
7d6d701 9621cc7 7d6d701 6f02f68 1ad0dcf 6f02f68 1ad0dcf 7d6d701 9ed9edc 7d6d701 6f02f68 52f3a4a 6f02f68 752918c b610816 9ed9edc 6f02f68 bef0bbf 6f02f68 7d6d701 3e7c183 6ed6ed9 58981a1 7d6d701 52f3a4a 9ed9edc 908ded3 7d6d701 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 |
import gradio as gr
import shutil, openai, os
from langchain.chains import RetrievalQA
from langchain.chat_models import ChatOpenAI
from langchain.document_loaders.blob_loaders.youtube_audio import YoutubeAudioLoader
from langchain.document_loaders.generic import GenericLoader
from langchain.document_loaders.parsers import OpenAIWhisperParser
from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.prompts import PromptTemplate
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from dotenv import load_dotenv, find_dotenv
_ = load_dotenv(find_dotenv())
#openai.api_key = os.environ["OPENAI_API_KEY"]
template = """Use the following pieces of context to answer the question at the end. If you don't know the answer, just say that you don't know,
don't try to make up an answer. Keep the answer as concise as possible. Always say "\n\nThanks for using the app, Bernd Straehle 🔥" at the end of the answer.
{context} Question: {question} Helpful Answer: """
QA_CHAIN_PROMPT = PromptTemplate(input_variables = ["context", "question"], template = template)
def invoke(openai_api_key, youtube_url, prompt):
openai.api_key = openai_api_key
youtube_dir = "docs/youtube/"
loader = GenericLoader(YoutubeAudioLoader([youtube_url], youtube_dir), OpenAIWhisperParser())
docs = loader.load()
text_splitter = RecursiveCharacterTextSplitter(chunk_size = 1500, chunk_overlap = 150)
splits = text_splitter.split_documents(docs)
chroma_dir = "docs/chroma/"
vectordb = Chroma.from_documents(documents = splits, embedding = OpenAIEmbeddings(), persist_directory = chroma_dir)
llm = ChatOpenAI(model_name = "gpt-4", temperature = 0)
qa_chain = RetrievalQA.from_chain_type(llm, retriever = vectordb.as_retriever(), return_source_documents = True, chain_type_kwargs = {"prompt": QA_CHAIN_PROMPT})
result = qa_chain({"query": prompt})
shutil.rmtree(youtube_dir)
shutil.rmtree(chroma_dir)
return result["result"]
description = """The app demonstrates how to use a <strong>Large Language Model</strong> (LLM) with <strong>Retrieval Augmented Generation</strong> (RAG) on external data.
Enter an OpenAI API key, YouTube URL (external data), and prompt to search the video, analyse its sentiment, summarize it, and/or translate it, etc.\n\n
Implementation: <a href='https://www.gradio.app/'>Gradio</a> UI using <a href='https://platform.openai.com/'>OpenAI</a> API
via AI-first toolkit <a href='https://www.langchain.com/'>LangChain</a> with foundation models
<a href='https://openai.com/research/whisper'>Whisper</a> (speech to text) and <a href='https://openai.com/research/gpt-4'>GPT-4</a> (LLM use cases)."""
gr.close_all()
demo = gr.Interface(fn=invoke,
inputs = [gr.Textbox(label = "OpenAI API Key", value = "sk-", lines = 1), gr.Textbox(label = "YouTube URL", value = "https://www.youtube.com/watch?v=GJm7H9IP5SU", lines = 1), gr.Textbox(label = "Prompt", value = "Translate song into English", lines = 1)],
outputs = [gr.Textbox(label = "Completion", lines = 1)],
title = "Generative AI - LLM & RAG",
description = description)
demo.launch() |