src/models/summarization.py

"""
Summarization Model Handler
Manages the fine-tuned BART model for text summarization.
"""

from transformers import BartTokenizer, BartForConditionalGeneration
import torch
import streamlit as st

class Summarizer:
    def __init__(self):
        """Initialize the summarization model."""
        self.model = None
        self.tokenizer = None

    def load_model(self):
        """Load the fine-tuned BART summarization model."""
        try:
            with open('bart_ami_finetuned.pkl','rb') as f:
                self.model = pickle.load(f)
            # Load the tokenizer
            self.tokenizer = BartTokenizer.from_pretrained("facebook/bart-base")
            
            # Move model to appropriate device (GPU if available)
            self.model.to(torch.device('cuda' if torch.cuda.is_available() else 'cpu'))
            return self.model
        except Exception as e:
            st.error(f"Error loading fine-tuned summarization model: {str(e)}")
            return None

    def process(self, text: str, max_length: int = 130, min_length: int = 30):
        """Process text for summarization.
        
        Args:
            text (str): Text to summarize
            max_length (int): Maximum length of summary
            min_length (int): Minimum length of summary
            
        Returns:
            str: Summarized text
        """
        try:
            # Tokenize input text
            inputs = self.tokenizer(text, return_tensors="pt", truncation=True, max_length=1024, padding="max_length")
            
            # Move inputs to the same device as the model
            inputs = {key: value.to(self.model.device) for key, value in inputs.items()}
            
            # Generate summary
            summary_ids = self.model.generate(
                inputs["input_ids"],
                max_length=max_length,
                min_length=min_length,
                num_beams=4,  # Beam search for better quality
                early_stopping=True
            )
            
            # Decode summary tokens to text
            summary = self.tokenizer.decode(summary_ids[0], skip_special_tokens=True)
            return summary
        except Exception as e:
            st.error(f"Error in summarization: {str(e)}")
            return None