Spaces:
Runtime error
Runtime error
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,50 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from llama_index import SimpleDirectoryReader, GPTSimpleVectorIndex, ServiceContext
|
| 3 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
| 4 |
+
from PyPDF2 import PdfReader
|
| 5 |
+
|
| 6 |
+
# Load tokenizer and model from HuggingFace (StableLM)
|
| 7 |
+
tokenizer = AutoTokenizer.from_pretrained("StabilityAI/stablelm-tuned-alpha-3b")
|
| 8 |
+
model = AutoModelForCausalLM.from_pretrained("StabilityAI/stablelm-tuned-alpha-3b")
|
| 9 |
+
|
| 10 |
+
# Create service context for the LLM
|
| 11 |
+
service_context = ServiceContext.from_defaults(
|
| 12 |
+
llm_predictor=(model, tokenizer), # Attach the model and tokenizer
|
| 13 |
+
chunk_size=1024
|
| 14 |
+
)
|
| 15 |
+
|
| 16 |
+
# Function to load PDF
|
| 17 |
+
def load_pdf(file):
|
| 18 |
+
reader = PdfReader(file.name)
|
| 19 |
+
text = ""
|
| 20 |
+
for page in reader.pages:
|
| 21 |
+
text += page.extract_text()
|
| 22 |
+
return text
|
| 23 |
+
|
| 24 |
+
# Function to create an index and query it
|
| 25 |
+
def chat_with_pdf(pdf, query):
|
| 26 |
+
# Read the PDF content
|
| 27 |
+
pdf_text = load_pdf(pdf)
|
| 28 |
+
|
| 29 |
+
# Use llama-index to create a document index
|
| 30 |
+
documents = [pdf_text]
|
| 31 |
+
index = GPTSimpleVectorIndex.from_documents(documents, service_context=service_context)
|
| 32 |
+
|
| 33 |
+
# Query the index
|
| 34 |
+
response = index.query(query)
|
| 35 |
+
return response.response
|
| 36 |
+
|
| 37 |
+
# Gradio interface
|
| 38 |
+
def chatbot(pdf, query):
|
| 39 |
+
if not pdf or not query:
|
| 40 |
+
return "Please upload a PDF and enter a query."
|
| 41 |
+
|
| 42 |
+
response = chat_with_pdf(pdf, query)
|
| 43 |
+
return response
|
| 44 |
+
|
| 45 |
+
# Define Gradio inputs and interface
|
| 46 |
+
pdf_input = gr.inputs.File(label="Upload your PDF")
|
| 47 |
+
query_input = gr.inputs.Textbox(label="Ask a question about the PDF")
|
| 48 |
+
output = gr.outputs.Textbox(label="Chatbot Response")
|
| 49 |
+
|
| 50 |
+
gr.Interface(fn=chatbot, inputs=[pdf_input, query_input], outputs=output, title="PDF Chatbot").launch()
|