Spaces:
Sleeping
Sleeping
Download app.py from abdurrehmansaeed/Research_Decoder: direct link, hf CLI and curl.
- Browser
- Download file 4.23 kB
-
https://huggingface.co/spaces/abdurrehmansaeed/Research_Decoder/resolve/main/app.py
- Command line
-
hf download hf://spaces/abdurrehmansaeed/Research_Decoder/app.py
-
curl -L -o app.py https://huggingface.co/spaces/abdurrehmansaeed/Research_Decoder/resolve/main/app.py
4.23 kB
| import os | |
| import tempfile | |
| import mimetypes | |
| from flask import Flask, request, jsonify | |
| from dotenv import load_dotenv | |
| from langchain_community.document_loaders import PyMuPDFLoader | |
| from langchain_text_splitters import RecursiveCharacterTextSplitter | |
| from langchain_huggingface import HuggingFaceEmbeddings | |
| from langchain_groq import ChatGroq | |
| from langchain_community.vectorstores import FAISS | |
| from langchain_classic.chains import create_retrieval_chain | |
| from langchain_classic.chains.combine_documents import create_stuff_documents_chain | |
| from langchain_core.prompts import ChatPromptTemplate | |
| # --- FORCE MIME TYPES FOR WINDOWS --- | |
| mimetypes.add_type('text/css', '.css') | |
| mimetypes.add_type('application/javascript', '.jsx') | |
| app = Flask(__name__, static_folder='frontend', static_url_path='/') | |
| global_vector_store = None | |
| # --- LLM --- | |
| load_dotenv() | |
| groq_api_key = os.getenv("GROQ_API_KEY") | |
| embeddings = HuggingFaceEmbeddings(model_name="BAAI/bge-small-en-v1.5") | |
| llm = ChatGroq( | |
| temperature=0.1, | |
| model_name="llama-3.3-70b-versatile", | |
| groq_api_key=groq_api_key | |
| ) | |
| def serve_index(): | |
| return app.send_static_file('index.html') | |
| # --- UPLOAD & PROCESS PDF --- | |
| def upload_file(): | |
| global global_vector_store | |
| if 'file' not in request.files: | |
| return jsonify({"error": "No file part"}), 400 | |
| file = request.files['file'] | |
| if file.filename == '': | |
| return jsonify({"error": "No selected file"}), 400 | |
| try: | |
| with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp_file: | |
| file.save(tmp_file.name) | |
| tmp_file_path = tmp_file.name | |
| loader = PyMuPDFLoader(tmp_file_path) | |
| documents = loader.load() | |
| text_splitter = RecursiveCharacterTextSplitter( | |
| separators=["\n\n", "\n", " ", ""], | |
| chunk_size=1000, | |
| chunk_overlap=200 | |
| ) | |
| chunks = text_splitter.split_documents(documents) | |
| global_vector_store = FAISS.from_documents(chunks, embeddings) | |
| intro_text = " ".join([doc.page_content for doc in chunks[:4]]) | |
| extraction_prompt = f""" | |
| Based on the following introductory text from a research paper, provide: | |
| 1. A 2-sentence summary of the paper. | |
| 2. The core problem the authors are trying to solve. | |
| 3. The primary methodology or model architecture proposed. | |
| Text: {intro_text} | |
| """ | |
| response = llm.invoke([("human", extraction_prompt)]) | |
| os.remove(tmp_file_path) | |
| return jsonify({"insights": response.content}), 200 | |
| except Exception as e: | |
| return jsonify({"error": str(e)}), 500 | |
| # --- CHAT WITH PDF --- | |
| def chat(): | |
| global global_vector_store | |
| if not global_vector_store: | |
| return jsonify({"error": "Please upload a document first."}), 400 | |
| data = request.json | |
| user_message = data.get('message') | |
| if not user_message: | |
| return jsonify({"error": "No message provided."}), 400 | |
| try: | |
| retriever = global_vector_store.as_retriever(search_kwargs={"k": 4}) | |
| system_prompt = ( | |
| "You are a highly technical AI assistant helping a machine learning engineer understand a research paper. " | |
| "Use the following pieces of retrieved context to answer the question. " | |
| "If the answer is not in the context, state clearly that the paper does not mention it. " | |
| "Keep the answer concise and highly technical.\n\n" | |
| "Context: {context}" | |
| ) | |
| prompt = ChatPromptTemplate.from_messages([ | |
| ("system", system_prompt), | |
| ("human", "{input}"), | |
| ]) | |
| question_answer_chain = create_stuff_documents_chain(llm, prompt) | |
| rag_chain = create_retrieval_chain(retriever, question_answer_chain) | |
| response = rag_chain.invoke({"input": user_message}) | |
| return jsonify({"answer": response["answer"]}), 200 | |
| except Exception as e: | |
| return jsonify({"error": str(e)}), 500 | |
| if __name__ == '__main__': | |
| app.run(host='0.0.0.0', port=7860, debug=True) |