Niranjann0902/Summarization_model

Model Details

Model Description

T5 Small - News Summarization Model Description This model is a fine-tuned version of google/t5-small for news summarization. It has been trained on the CNN/DailyMail dataset to generate concise summaries of news articles.

Base Model: google/t5-small Fine-tuned On: cnn_dailymail (version 3.0.0) Task: Text Summarization Framework: 🤗 Transformers (Trainer API)

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

• Developed by: [Niranjan R]
• Funded by [optional]: [More Information Needed]
• Shared by [optional]: [More Information Needed]
• Model type: [Summarization_Model]
• Language(s) (NLP): [English]
• License: [More Information Needed]
• Finetuned from model [optional]: [More Information Needed]

How to Get Started with the Model

Load model directly

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("Niranjann0902/Summarization_model") model = AutoModelForSeq2SeqLM.from_pretrained("Niranjann0902/Summarization_model")

[More Information Needed]

Training Details

GPU: Google Colab (T4 GPU) Epochs: 1 (for fast training) Batch Size: 16 Optimizer: AdamW with weight_decay=0.01 Mixed Precision: Enabled (fp16=True)

Training Data

"cnn_dailymail" Dataset is used, Version : 3.0.0

Limitations:

May struggle with long-form summarization. Fine-tuning on more diverse datasets could improve performance. Exploring even smaller models (like DistilT5) for faster inference.

Training Procedure

The model was fine-tuned using the Hugging Face Trainer API with the following setup:

Dataset: cnn_dailymail (Version 3.0.0) Training Split: 10,000 samples Validation Split: 2,000 samples Batch Size: 16 (for both training and evaluation) Epochs: 1 (for quick training) Optimizer: AdamW (weight_decay=0.01) Mixed Precision: Enabled (fp16=True) Hardware Used: Google Colab (T4 GPU)

The model was trained on a subset of the dataset to speed up training while maintaining reasonable performance. The learning process used the "summarize: " prefix before inputs, following the standard T5 fine-tuning approach for summarization tasks.

Testing Data, Training Data, Factors & Metrics

Training Set: 10,000 samples from cnn_dailymail Validation Set: 2,000 samples from cnn_dailymail Preprocessing: Tokenization using T5Tokenizer with max-length padding and truncation Text Format: Input: "summarize: " Output:

More Information

Summarization Model

Model Card Authors

Niranjan R

Model Card Contact

Contact Mail: niranjanraja.99@gmail.com