Spaces:
Sleeping
Sleeping
File size: 3,303 Bytes
184bec9 fc697fd 184bec9 fc697fd 184bec9 fc697fd |
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 66 67 68 69 70 71 72 73 74 75 76 |
import os
import gradio as gr
import asyncio
from langchain_core.prompts import PromptTemplate
from langchain_community.output_parsers.rail_parser import GuardrailsOutputParser
from langchain_community.document_loaders import PyPDFLoader
from langchain_google_genai import ChatGoogleGenerativeAI
import google.generativeai as genai
from langchain.chains.question_answering import load_qa_chain
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
# Gemini PDF QA System
async def initialize(file_path, question):
genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
model = genai.GenerativeModel('gemini-pro')
model = ChatGoogleGenerativeAI(model="gemini-pro", temperature=0.3)
prompt_template = """Answer the question as precise as possible using the provided context. If the answer is
not contained in the context, say "answer not available in context" \n\n
Context: \n {context}?\n
Question: \n {question} \n
Answer:
"""
prompt = PromptTemplate(template=prompt_template, input_variables=["context", "question"])
if os.path.exists(file_path):
pdf_loader = PyPDFLoader(file_path)
pages = pdf_loader.load_and_split()
context = "\n".join(str(page.page_content) for page in pages[:30])
stuff_chain = load_qa_chain(model, chain_type="stuff", prompt=prompt)
stuff_answer = await stuff_chain({"input_documents": pages, "question": question, "context": context}, return_only_outputs=True)
return stuff_answer['output_text']
else:
return "Error: Unable to process the document. Please ensure the PDF file is valid."
async def pdf_qa(file, question):
answer = await initialize(file.name, question)
return answer
# Mistral Text Completion
def load_mistral_model():
model_path = "nvidia/Mistral-NeMo-Minitron-8B-Base"
tokenizer = AutoTokenizer.from_pretrained(model_path)
device = 'cuda' if torch.cuda.is_available() else 'cpu'
dtype = torch.bfloat16
model = AutoModelForCausalLM.from_pretrained(model_path, torch_dtype=dtype, device_map=device)
return tokenizer, model
def generate_text(prompt, max_length=50):
tokenizer, model = load_mistral_model()
inputs = tokenizer.encode(prompt, return_tensors='pt').to(model.device)
outputs = model.generate(inputs, max_length=max_length)
return tokenizer.decode(outputs[0])
# Gradio Interface
input_file = gr.File(label="Upload PDF File")
input_question = gr.Textbox(label="Ask about the document")
output_text_gemini = gr.Textbox(label="Answer - GeminiPro")
input_prompt = gr.Textbox(label="Enter prompt for text completion")
output_text_mistral = gr.Textbox(label="Completed Text - Mistral")
def pdf_qa_wrapper(file, question):
return asyncio.run(pdf_qa(file, question))
# Create Gradio Interface
iface = gr.Interface(
fn=[pdf_qa_wrapper, generate_text],
inputs=[
[input_file, input_question],
input_prompt
],
outputs=[output_text_gemini, output_text_mistral],
title="Combined PDF QA and Text Completion System",
description="Upload a PDF file to ask questions about its content, or enter a prompt for text completion."
)
iface.launch() |