T5 Model for Text Summarization

Model Details

• Base Model: T5 (Text-to-Text Transfer Transformer)
• Downstream Task: Text Summarization
• Dataset: News Summary
  • Size: 4515 examples
  • Content: Each example includes:
    • Author Name
    • Headlines
    • URL of Article
    • Short Text
    • Complete Article
  • Data Sources: Summarized news from Inshorts, with scraped news articles from The Hindu, Indian Times, and The Guardian.
  • Time Period: February to August 2017

Fine-Tuning Details

• The model was fine-tuned using a training script based on a Colab Notebook originally created by Abhishek Kumar Mishra. Special thanks to him for the excellent work!
• Modifications Made:
  • A slightly modified version of the original script was used.
  • The model was trained for more epochs (6 epochs) to improve performance on the News Summary dataset.
  • Minor tweaks were applied to the preprocessing and hyperparameters to better suit the data.