Skip to content
TheChatArchiveThe Chat Archive
Grok 4.1 Fast· xaiHugging Face Frontier Reasoning Traces

Implement a Python script to build a text summarizer using transformers.

  1. User

    Implement a Python script to build a text summarizer using transformers.

  2. Assistant
    #!/usr/bin/env python3
    """
    Text Summarizer using Hugging Face Transformers.
    
    Requirements:
        pip install transformers torch accelerate
    
    Usage:
        python text_summarizer.py
        - Paste your text when prompted.
        - Or pipe text: echo "your long text here" | python text_summarizer.py
    """
    
    import sys
    from transformers import pipeline
    
    def main():
        # Load the pre-trained BART model for summarization (good for CNN/DailyMail style summaries)
        # You can swap with 't5-small' or 'google/pegasus-xsum' for different styles
        print("Loading summarization model... (This may take a moment on first run.)")
        summarizer = pipeline(
            "summarization",
            model="facebook/bart-large-cnn",
            device=-1  # Use CPU; set to 0 for GPU if available
        )
        
        # Get input text
        if not sys.stdin.isatty():
            # If piped input, read from stdin
            text = sys.stdin.read().strip()
        else:
            # Interactive mode
            print("\nEnter your text to summarize (Ctrl+D or empty line to finish):")
            lines = []
            try:
                while True:
                    line = input()
                    if not line.strip():
                        break
                    lines.append(line)
            except EOFError:
                pass
            text = ' '.join(lines).strip()
        
        if not text:
            print("No text provided. Exiting.")
            return
        
        print("\nGenerating summary...")
        try:
            # Generate summary with length constraints
            # Adjust max_length/min_length as needed (BART max tokens ~1024 input, ~142 output)
            summary = summarizer(
                text,
                max_length=130,
                min_length=30,
                do_sample=False,  # Deterministic output
                truncation=True
            )[0]['summary_text']
            
            print("\n" + "="*50)
            print("SUMMARY:")
            print(summary)
            print("="*50)
            print(f"\nOriginal length: {len(text.split())} words")
            print(f"Summary length: {len(summary.split())} words")
            
        except Exception as e:
            print(f"Error generating summary: {e}")
            print("Tip: Try shorter text or a lighter model like 'sshleifer/distilbart-cnn-12-6'.")
    
    if __name__ == "__main__":
        main()
    

    How to Use:

    1. Save the script as text_summarizer.py.
    2. Install dependencies:
      pip install transformers torch accelerate
      
    3. Run interactively:
      python text_summarizer.py
      
      • Paste your text and press Enter twice (or Ctrl+D on Unix).
    4. Pipe long text:
      python text_summarizer.py < article.txt
      
      or
      echo "Your very long article here..." | python text_summarizer.py
      
    5. Customization:
      • Change model for different results (e.g., 'google/pegasus-xsum' for abstractive XSum-style).
      • For very long texts (>1024 tokens), consider splitting into chunks (advanced extension).

    This uses BART-large-CNN, a state-of-the-art model fine-tuned on news articles. It handles truncation automatically and produces high-quality summaries! 🚀

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.