Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 3 |
+
from PyPDF2 import PdfReader
|
| 4 |
+
import torch
|
| 5 |
+
|
| 6 |
+
# Load IBM Granite model (use a smaller one if needed, e.g., granite-3.0-3b-instruct)
|
| 7 |
+
model_name = "ibm-granite/granite-3.0-8b-instruct"
|
| 8 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 9 |
+
model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto")
|
| 10 |
+
|
| 11 |
+
# Global variable for PDF context
|
| 12 |
+
pdf_context = ""
|
| 13 |
+
|
| 14 |
+
def upload_pdf(file):
|
| 15 |
+
global pdf_context
|
| 16 |
+
if file is None:
|
| 17 |
+
return "No file uploaded."
|
| 18 |
+
reader = PdfReader(file)
|
| 19 |
+
pdf_context = ""
|
| 20 |
+
for page in reader.pages:
|
| 21 |
+
pdf_context += page.extract_text() + "\n"
|
| 22 |
+
return "PDF uploaded and text extracted successfully!"
|
| 23 |
+
|
| 24 |
+
def chat(message, history):
|
| 25 |
+
# Build messages with history and PDF context
|
| 26 |
+
messages = [{"role": "system", "content": f"You are a helpful assistant. Answer based on this context: {pdf_context}"}]
|
| 27 |
+
for user_msg, assistant_msg in history:
|
| 28 |
+
messages.append({"role": "user", "content": user_msg})
|
| 29 |
+
messages.append({"role": "assistant", "content": assistant_msg})
|
| 30 |
+
messages.append({"role": "user", "content": message})
|
| 31 |
+
|
| 32 |
+
# Apply chat template and generate
|
| 33 |
+
input_text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
| 34 |
+
inputs = tokenizer(input_text, return_tensors="pt").to(model.device)
|
| 35 |
+
outputs = model.generate(**inputs, max_new_tokens=300, temperature=0.7)
|
| 36 |
+
response = tokenizer.decode(outputs[0][len(inputs["input_ids"][0]):], skip_special_tokens=True)
|
| 37 |
+
return response
|
| 38 |
+
|
| 39 |
+
# Gradio interface
|
| 40 |
+
with gr.Blocks() as demo:
|
| 41 |
+
gr.Markdown("# Basic PDF Q&A Chat with IBM Granite")
|
| 42 |
+
|
| 43 |
+
with gr.Row():
|
| 44 |
+
pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
|
| 45 |
+
upload_btn = gr.Button("Upload PDF")
|
| 46 |
+
|
| 47 |
+
status = gr.Textbox(label="Status")
|
| 48 |
+
|
| 49 |
+
chat_interface = gr.ChatInterface(
|
| 50 |
+
fn=chat,
|
| 51 |
+
title="Ask questions about the PDF"
|
| 52 |
+
)
|
| 53 |
+
|
| 54 |
+
upload_btn.click(upload_pdf, inputs=pdf_input, outputs=status)
|
| 55 |
+
|
| 56 |
+
demo.launch()
|