GLiNER multi-task: Generalist Lightweight Model for Various Information Extraction Tasks

TL;DR

This paper introduces GLiNER multi-task, a lightweight encoder model that achieves state-of-the-art zero-shot performance on NER and strong results on other information extraction tasks, offering an efficient alternative to large language models.

Contribution

The paper presents a novel lightweight encoder model capable of handling various information extraction tasks with competitive performance, reducing reliance on large language models.

Findings

Achieved state-of-the-art zero-shot NER performance

Led in question-answering, summarization, and relation extraction tasks

Demonstrated effectiveness of self-learning approaches for NER

Abstract

Information extraction tasks require both accurate, efficient, and generalisable models. Classical supervised deep learning approaches can achieve the required performance, but they need large datasets and are limited in their ability to adapt to different tasks. On the other hand, large language models (LLMs) demonstrate good generalization, meaning that they can adapt to many different tasks based on user requests. However, LLMs are computationally expensive and tend to fail to generate structured outputs. In this article, we will introduce a new kind of GLiNER model that can be used for various information extraction tasks while being a small encoder model. Our model achieved SoTA performance on zero-shot NER benchmarks and leading performance on question-answering, summarization and relation extraction tasks. Additionally, in this article, we will cover experimental results on self-learning approaches for named entity recognition using GLiNER models.