---
library_name: transformers
license: other
base_model: Qwen/Qwen2.5-0.5B-Instruct
tags:
- llama-factory
- full
- generated_from_trainer
model-index:
- name: entity_extraction_new
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# QuCo-extractor-0.5B

[![arXiv](https://img.shields.io/badge/arXiv-2512.19134-b31b1b.svg)](https://arxiv.org/abs/2512.19134)
[![GitHub](https://img.shields.io/badge/GitHub-QuCo--RAG-blue?logo=github)](https://github.com/ZhishanQ/QuCo-RAG)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://github.com/ZhishanQ/QuCo-RAG/blob/main/LICENSE)

## Model Description

**QuCo-extractor-0.5B** is a specialized entity extraction model fine-tuned from [Qwen2.5-0.5B-Instruct](https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct) for the **QuCo-RAG** system. This model extracts knowledge triples (entity-relation-entity) from sentences to support corpus-grounded uncertainty quantification in Retrieval-Augmented Generation.

This model is part of the QuCo-RAG project presented in:

> **QuCo-RAG: Quantifying Uncertainty from the Pre-training Corpus for Dynamic Retrieval-Augmented Generation**
> 
> Dehai Min, Kailin Zhang, Tongtong Wu, Lu Cheng
> 
> [[Paper]](https://arxiv.org/abs/2512.19134) [[Code]](https://github.com/ZhishanQ/QuCo-RAG)


## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 5e-06
- train_batch_size: 28
- eval_batch_size: 32
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 2
- total_train_batch_size: 56
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 2.0


### Framework versions

- Transformers 4.57.1
- Pytorch 2.5.1+cu121
- Datasets 4.0.0
- Tokenizers 0.22.1