Sachin21112004/distilbart-news-summarizer

📰 News Summarizer (DistilBART-Based)

A lightweight, fast, and highly efficient news summarization model built on top of the distilled BART architecture.
This model converts long news articles into clear, concise, and human-like summaries, making it perfect for:

• News aggregation platforms
• Research workflows
• Content automation
• Browser extensions
• Educational tools
• AI agents & chatbots

---

🚀 Features

✔ High-quality abstractive summaries

Not extractive — the model generates a new summary in natural language.

✔ Fast & lightweight

Based on the 12-6 distilled BART variant, giving strong performance at a fraction of the size.

✔ Trained on real news sources

Understands journalistic writing, factual structure, headlines, and key-point extraction.

✔ Ideal for production & APIs

Minimal latency, optimized for cloud/server use.

---

📦 How to Use

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

model_name = "Sachin21112004/news-summarizer"

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

article = """
Your long news article text here...
"""

inputs = tokenizer(article, return_tensors="pt", max_length=1024, truncation=True)
summary_ids = model.generate(
    inputs["input_ids"],
    max_length=150,
    min_length=40,
    no_repeat_ngram_size=3,
    length_penalty=2.0,
    num_beams=4,
    early_stopping=True
)

print(tokenizer.decode(summary_ids[0], skip_special_tokens=True))