L

Longt5 Tglobal Large 16384 Pubmed 3k Steps

Developed by Stancld
LongT5 is a long-sequence text-to-text Transformer model based on T5, employing transient-global attention mechanism, suitable for long-text processing tasks.
Downloads 1,264
Release Time : 6/10/2022

Model Overview

This model is a large-scale configuration of LongT5, specifically designed for long text sequences, fine-tuned on the PubMed abstract dataset, and suitable for tasks like text summarization.

Model Features

Long Sequence Processing Capability
Supports input sequences of up to 16,384 tokens, suitable for processing long documents.
Transient-Global Attention Mechanism
Combines local and global attention to capture long-range dependencies while maintaining efficiency.
Medical Domain Adaptation
Fine-tuned on the PubMed medical abstract dataset, suitable for medical text processing tasks.

Model Capabilities

Long Text Processing
Text Summarization Generation
Medical Text Understanding

Use Cases

Medical Research
Medical Literature Summarization
Automatically generates concise summaries of medical research papers
Rouge-1 score of 47.44, close to the original paper's performance
Academic Assistance
Long Document Content Extraction
Extracts key information from lengthy academic papers
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
Š 2025AIbase