Spaces:
Sleeping
Sleeping
File size: 2,496 Bytes
6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 bd2041d 6378bf1 |
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 54 55 56 57 58 59 60 61 62 63 64 65 |
import gradio as gr
from langchain.vectorstores import Chroma
from langchain_community.document_loaders import PyPDFLoader
from langchain_community.embeddings import HuggingFaceEmbeddings
from transformers import LayoutLMv3Processor, AutoModelForSeq2SeqLM
from langchain.chains import RetrievalQA
from langchain.prompts import PromptTemplate
from pdf2image import convert_from_path
import os
class LayoutLMv3OCR:
def __init__(self):
self.processor = LayoutLMv3Processor.from_pretrained("microsoft/layoutlmv3-base")
self.model = AutoModelForSeq2SeqLM.from_pretrained("microsoft/layoutlmv3-base")
def extract_text(self, pdf_path):
images = convert_from_path(pdf_path)
text_pages = []
for image in images:
inputs = self.processor(images=image, return_tensors="pt")
outputs = self.model.generate(**inputs)
text = self.processor.batch_decode(outputs, skip_special_tokens=True)[0]
text_pages.append(text)
return text_pages
ocr_tool = LayoutLMv3OCR()
def process_pdf_and_query(pdf_path, question):
loader = PyPDFLoader(pdf_path)
documents = loader.load()
embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
vectordb = Chroma.from_documents(documents, embeddings)
retriever = vectordb.as_retriever()
prompt_template = "Beantworte die folgende Frage basierend auf dem Dokument: {context}\nFrage: {question}\nAntwort:"
prompt = PromptTemplate(input_variables=["context", "question"], template=prompt_template)
qa_chain = RetrievalQA.from_chain_type(llm=None, retriever=retriever, chain_type_kwargs={"prompt": prompt})
response = qa_chain.run(input_documents=documents, question=question)
return response
def chatbot_response(pdf, question):
pdf_path = "uploaded_pdf.pdf"
pdf.save(pdf_path)
extracted_text = ocr_tool.extract_text(pdf_path)
answer = process_pdf_and_query(pdf_path, question)
os.remove(pdf_path)
return answer
pdf_input = gr.inputs.File(label="PDF-Datei hochladen")
question_input = gr.inputs.Textbox(label="Frage eingeben")
response_output = gr.outputs.Textbox(label="Antwort")
interface = gr.Interface(
fn=chatbot_response,
inputs=[pdf_input, question_input],
outputs=response_output,
title="RAG Chatbot mit PDF-Unterstützung",
description="Lade eine PDF-Datei hoch und stelle Fragen zu ihrem Inhalt."
)
if __name__ == "__main__":
interface.launch()
|