You can deploy this model and start earning money today!

mia-project-2025/
T5-base-feature-extraction-financial-news-articles

Name: T5-base-feature-extraction-financial-news-articles
Availability: InStock

apache-2.0

Views

No views yet

T5-Feature-Based-Financial: Feature-Based Transfer Learning for Financial News Summarization

This repository contains a T5 encoder used as a frozen feature extractor, paired with a custom lightweight Transformer decoder trained from scratch to generate concise titles for financial news articles.

Model Overview

Encoder: mrm8488/t5-base-finetuned-summarize-news (frozen)
Decoder: Lightweight Transformer decoder (trained from scratch)
Task: Text summarization (financial news → news title)
Dataset: Financial News Articles (80/20 train/test split)
Evaluation Metric: ROUGE
Framework: PyTorch + Hugging Face Transformers + Accelerate

Evaluation Results

Metric	Score
ROUGE-1	81.74
ROUGE-2	68.98
ROUGE-L	80.68
ROUGE-Lsum	80.67

These results were obtained after 12 epochs of feature-based transfer learning on 4 GPUs.

mia-project-2025/T5-base-feature-extraction-financial-news-articles

T5-Feature-Based-Financial: Feature-Based Transfer Learning for Financial News Summarization

Model Overview

Evaluation Results

mia-project-2025/
T5-base-feature-extraction-financial-news-articles