app.py

import torch
from fastai.text.all import *
from blurr.text.data.all import *
from blurr.text.modeling.all import *  # Import only needed functions
from transformers import T5Tokenizer, T5ForConditionalGeneration  # Use T5 specifically

# Load the pre-trained model and tokenizer (adjust for Bart if needed)
pretrained_model_name = "facebook/bart-large-cnn"  # Or "facebook/bart-base"
hf_tokenizer = T5Tokenizer.from_pretrained(pretrained_model_name)

def summarize(article):
  # Define your data transformation pipeline here, if applicable
  # ...

  # Load the exported model
  learn = load_learner('article_highlights.pkl')

  # Generate the summary
  summary = learn.predict(article)[0]['highlights']

  return summary

# Create the Gradio interface
iface = gr.Interface(
    fn=summarize,
    inputs="text",
    outputs="text",
    title="Article Summarizer (Part 3)",
    description="Enter an article and get a summary.",
    examples=[["This is an example article..."]]
)

# Launch the Gradio interface
iface.launch()