|
|
--- |
|
|
license: llama3.2 |
|
|
base_model: meta-llama/Meta-Llama-3-3B |
|
|
tags: |
|
|
- summarization |
|
|
- news |
|
|
- llama3 |
|
|
- fine-tuned |
|
|
- llama3.2 |
|
|
--- |
|
|
|
|
|
# 📰 Automatic News Summarizer (Fine-tuned on LLaMA 3.2 3B) |
|
|
|
|
|
This model is a fine-tuned version of **Meta-LLaMA-3-3B**, optimized for **automatic news summarization** tasks. It is designed to generate concise and coherent summaries of news articles using state-of-the-art language modeling techniques. |
|
|
|
|
|
## 📌 Model Details |
|
|
|
|
|
- **Base model**: [Meta-LLaMA-3-3B](https://huggingface.co/meta-llama/Meta-Llama-3-3B) |
|
|
- **Fine-tuned for**: Summarization |
|
|
- **Architecture**: Decoder-only Transformer (LLaMA 3.2 3B) |
|
|
- **License**: [Meta LLaMA 3 Community License](https://ai.meta.com/resources/models-and-libraries/llama-downloads/) |
|
|
|
|
|
## 📚 Training |
|
|
|
|
|
- Fine-tuned on a dataset of curated news articles and summaries |
|
|
- Optimized for relevance, coherence, and brevity |
|
|
- Training framework: Hugging Face Transformers |
|
|
- Fine-tuning platform: Google Colab |
|
|
|
|
|
## 🚀 Usage Example |
|
|
|
|
|
```python |
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
|
|
|
model_name = "punit16/automatic_news_summarizer" |
|
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name) |
|
|
model = AutoModelForCausalLM.from_pretrained(model_name) |
|
|
|
|
|
input_text = "The government has announced a new policy today aimed at reducing air pollution in major cities..." |
|
|
|
|
|
inputs = tokenizer(input_text, return_tensors="pt") |
|
|
summary_ids = model.generate(**inputs, max_new_tokens=512) |
|
|
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True) |
|
|
|
|
|
print("Summary:", summary) |
|
|
|
