Yeop9690/t5-small-cnn-dailymail-summarization

Model Card for t5_small Summarization Model

Model Details

• Model Name: T5-small Summarization Model
• Architecture: T5-small
• Purpose: Summarization of news articles from the CNN/DailyMail dataset.

Training Data

• Dataset: CNN/DailyMail dataset (version 3.0.0)

Training Procedure

• Learning Rate: 2e-5
• Batch Size: 4 (per device)
• Epochs: 3
• Evaluation: ROUGE and BLEU scores were used to evaluate the summarization quality.

How to Use

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("Yeop9690/t5-small-cnn-dailymail-summarization")
model = AutoModelForSeq2SeqLM.from_pretrained("Yeop9690/t5-small-cnn-dailymail-summarization")

Evaluation

Test Results:

• eval_rouge1: 0.49
• eval_rouge2: 0.30
• eval_rougeL: 0.45
• eval_bleu1: 38.46
• eval_bleu2: 25.00
• eval_bleu4: 15.00

Limitations

• This model may not perform well on highly technical or domain-specific content.
• The summaries may sometimes miss important context or nuances in the original text.

Ethical Considerations