GLiNER2: An Efficient Multi-Task Information Extraction System with Schema-Driven Interface
Summary
GLiNER2 is a unified transformer-based framework for multiple NLP tasks, offering improved efficiency and accessibility over large language models by supporting named entity recognition, text classification, and hierarchical data extraction in a single model.
View Cached Full Text
Cached at: 09/19/26, 02:54 PM
Paper page - GLiNER2: An Efficient Multi-Task Information Extraction System with Schema-Driven Interface
Source: https://huggingface.co/papers/2507.18546 Published on Jul 24, 2025
Abstract
GLiNER2 is a unified transformer-based framework that supports multiple NLP tasks with improved efficiency and accessibility compared to large language models.
Information extraction (IE) is fundamental to numerous NLP applications, yet existing solutions often require specialized models for different tasks or rely on computationally expensive large language models. We present GLiNER2, a unified framework that enhances the original GLiNER architecture to support named entity recognition, text classification, and hierarchical structured data extraction within a single efficient model. Built pretrained transformer encoder architecture, GLiNER2 maintains CPU efficiency and compact size while introducingmulti-task compositionthrough an intuitiveschema-based interface. Our experiments demonstrate competitive performance across extraction and classification tasks with substantial improvements in deployment accessibility compared to LLM-based alternatives. We release GLiNER2 as an open-source pip-installable library with pre-trained models and documentation at https://github.com/fastino-ai/GLiNER2.
View arXiv pageView PDFGitHub1.99kAdd to collection
Community
Paper submitter
So cool, really great work guys!
Paper author
Hi, nice work!
I am wondering when would you publish your model, thank you!
Please let us know whether the model will be open source.
·
multi lingual please, you so good
great work! Any idea when github repo will be available and are there any pretrained models already published?
Can i get the model weights?
Upload images, audio, and videos by dragging in the text input, pasting, orclicking here.
Tap or paste here to upload images
Get this paper in your agent:
hf papers read 2507\.18546
Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash
Models citing this paper14
#### fastino/gliner2.5-multi-v1 Token Classification• 0.3B• Updated3 days ago • 200k • 212
#### fastino/gliner2-base-v1 0.2B• Updated3 days ago • 401k • 99
#### fastino/gliner2-large-v1 0.5B• Updated3 days ago • 226k • 97
#### fastino/gliner2-multi-v1 0.3B• Updated3 days ago • 201k • 87
Browse 14 models citing this paper## Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2507.18546 in a dataset README.md to link it from this page.
Spaces citing this paper27
Browse 27 spaces citing this paper## Collections including this paper1
Similar Articles
GLiNER-Relex: A Unified Framework for Joint Named Entity Recognition and Relation Extraction
GLiNER-Relex is a unified framework for joint named entity recognition and relation extraction that leverages a shared transformer encoder for zero-shot capabilities. The paper demonstrates competitive performance on standard benchmarks and releases the model as an open-source Python package.
@george_onx: Seeing a lot of comparisons of Jev w/ GLiNER models lately. If you're benchmarking, GLiNER2.5 is the latest version we'…
GLiNER2.5 is a multilingual boundary checkpoint for unified schema-based information extraction, supporting entities, classification, structured records, relations, and span attributes in one model.
Enhancing Scientific Named Entity Recognition via Large Language Models: A Type-driven Multi-task Learning Approach
This paper proposes TdSciNER, a type-driven multi-task learning approach that leverages LLMs to improve scientific named entity recognition by filtering entity types, adding an auxiliary typing task, and using a demonstration selection strategy. Experiments on three datasets show performance comparable to fully supervised models.
fastino/GLiNER2.5-Decide
GLiNER2.5-Decide is a 340M parameter English classification model for operational decisions, supporting multiple label sets in a single forward pass without prompt templates.
SchemaRAG: Dynamic Large Schema Reduction for LLM-driven Structured Information Extraction
SchemaRAG is a retrieval-augmented generation framework that dynamically reduces the output schema space for LLM-driven structured information extraction, achieving improved performance and efficiency on healthcare and e-commerce datasets.