Download app.py from haksoo/emaster: direct link, hf CLI and curl.
- Browser
- Download file 2.29 kB
-
https://huggingface.co/spaces/haksoo/emaster/resolve/main/app.py
- Command line
-
hf download hf://spaces/haksoo/emaster/app.py
-
curl -L -o app.py https://huggingface.co/spaces/haksoo/emaster/resolve/main/app.py
2.29 kB
| import gradio as gr | |
| import os | |
| os.environ["OPENAI_API_KEY"] = os.getenv('OPENAI_API_KEY') | |
| from langchain.document_loaders import PyPDFLoader | |
| from langchain.text_splitter import RecursiveCharacterTextSplitter | |
| from langchain.vectorstores import Chroma | |
| from langchain.embeddings import OpenAIEmbeddings | |
| from langchain.chat_models import ChatOpenAI | |
| from langchain.retrievers.multi_query import MultiQueryRetriever | |
| from langchain.chains import RetrievalQA | |
| #Loader | |
| loader = PyPDFLoader("2020_emaster_keynote.pdf") | |
| pages = loader.load_and_split() | |
| #Split | |
| text_splitter = RecursiveCharacterTextSplitter( | |
| # Set a really small chunk size, just to show. | |
| chunk_size = 300, | |
| chunk_overlap = 20, | |
| length_function = len, | |
| is_separator_regex = False, | |
| ) | |
| texts = text_splitter.split_documents(pages) | |
| #Embedding | |
| embeddings_model = OpenAIEmbeddings() | |
| # load it into Chroma | |
| db = Chroma.from_documents(texts, embeddings_model) | |
| # question = "๋ฐ์ดํฐ๋ฒ ์ด์ค์ ์ ์?" | |
| llm = ChatOpenAI(model_name="gpt-3.5-turbo", temperature=0.5) | |
| qa_chain = RetrievalQA.from_chain_type(llm,retriever=db.as_retriever()) | |
| # ์ธํฐํ์ด์ค๋ฅผ ์์ฑ. | |
| with gr.Blocks() as demo: | |
| gr.Image('images/emaster.png') | |
| gr.Text(''' | |
| ์ด๋ฐ ์ง๋ฌธ ์ด๋ ์ธ์? | |
| ๋ฐ์ดํฐ๋ฒ ์ด์ค์ ์ ์ | |
| ''') | |
| chatbot = gr.Chatbot(label="์ ๋ณด์ฒ๋ฆฌ์ฐ์ ๊ธฐ์ฌ์ฑ๋ด") # ์ฒญ๋ ์ ์ฑ ์ฑ๋ด ๋ ์ด๋ธ์ ์ข์ธก ์๋จ์ ๊ตฌ์ฑ | |
| msg = gr.Textbox(label="์ง๋ฌธํด์ฃผ์ธ์!") # ํ๋จ์ ์ฑํ ์ฐฝ์ ๋ ์ด๋ธ | |
| clear = gr.Button("๋ํ ์ด๊ธฐํ") # ๋ํ ์ด๊ธฐํ ๋ฒํผ | |
| # ์ฑ๋ด์ ๋ต๋ณ์ ์ฒ๋ฆฌํ๋ ํจ์ | |
| def respond(message, chat_history): | |
| result = qa_chain({"query": message}) | |
| # result = qa_chain(message) | |
| bot_message = result['result'] | |
| # ์ฑํ ๊ธฐ๋ก์ ์ฌ์ฉ์์ ๋ฉ์์ง์ ๋ด์ ์๋ต์ ์ถ๊ฐ. | |
| chat_history.append((message, bot_message)) | |
| return "", chat_history | |
| # ์ฌ์ฉ์์ ์ ๋ ฅ์ ์ ์ถ(submit)ํ๋ฉด respond ํจ์๊ฐ ํธ์ถ. | |
| msg.submit(respond, [msg, chatbot], [msg, chatbot]) | |
| # '์ด๊ธฐํ' ๋ฒํผ์ ํด๋ฆญํ๋ฉด ์ฑํ ๊ธฐ๋ก์ ์ด๊ธฐํ. | |
| clear.click(lambda: None, None, chatbot, queue=False) | |
| # ์ธํฐํ์ด์ค ์คํ. | |
| demo.launch(debug=True, share=True) |