Papers by Yijun Liu

Graph-GRPO: Stabilizing Multi-Agent Topology Learning via Group Relative Policy Optimization (2026.findings-acl)

Copied to clipboard

Yueyang Cang, Xiaoteng Zhang, Erlu Zhao, Zehua Ji, Yuhang Liu, Yuchen He, Zhiyuan Ning, Chen Yijun, Wenge Que, Li Shi

Challenge:	Recent approaches to optimize communication topology rely on single-sample policy gradients with absolute rewards.
Approach:	They propose a topology optimization framework that integrates Group Relative Policy Optimization.
Outcome:	The proposed topology optimization framework outperforms state-of-the-art methods on reasoning and code generation benchmarks.

Lookahead Q-Cache: Achieving More Consistent KV Cache Eviction via Pseudo Query (2025.emnlp-main)

Copied to clipboard

Yixuan Wang, Shiyu Ji, Yijun Liu, Yuzhuang Xu, Yang Xu, Qingfu Zhu, Wanxiang Che

Challenge:	Existing KV cache eviction methods prune tokens using prefilling-stage attention scores, causing inconsistency with actual inference queries.
Approach:	They propose a lookahead q-cache framework that generates low-cost pseudo lookaheaded queries to better approximate the true decoding-stage queries.
Outcome:	The proposed framework outperforms existing methods on LongBench and Needle-in-a-Haystack benchmarks and can be flexibly combined to yield further improvements.

Make Some Noise: Unlocking Language Model Parallel Inference Capability through Noisy Training (2024.emnlp-main)

Copied to clipboard

Yixuan Wang, Xianzhen Luo, Fuxuan Wei, Yijun Liu, Qingfu Zhu, Xuanyu Zhang, Qing Yang, Dongliang Xu, Wanxiang Che

Language Models

Challenge:	Existing speculative decoding methods require additional model structure and training processes to assist the model for draft token generation.
Approach:	They propose a make some noise training framework that introduces some noise at the input for the model to learn the denoising task.
Outcome:	The proposed model improves inference speed by 2.3-2.7x times without compromising model performance.

Towards Safer Large Language Models through Machine Unlearning (2024.findings-acl)

Copied to clipboard

Zheyuan Liu, Guangyao Dou, Zhaoxuan Tan, Yijun Tian, Meng Jiang

Responsible & Trustworthy NLP

Challenge:	Existing work attempted to implement a gradient ascent based approach to prevent LLMs from producing harmful output when faced with problematic prompts.
Approach:	They propose a gradient ascent based approach to prevent LLMs from producing harmful output when faced with problematic prompts.
Outcome:	The proposed approach eliminates harmful knowledge while preserving utility on normal prompts.

Domain-aware and Co-adaptive Feature Transformation for Domain Adaption Few-shot Relation Extraction (2024.lrec-main)

Copied to clipboard

Yijun Liu, Feifei Dai, Xiaoyan Gu, Minghui Zhai, Bo Li, Meiou Zhang

Responsible & Trustworthy NLP

Challenge:	Existing approaches to relation extraction focus on the source domain, which makes it difficult to accurately transfer useful knowledge to the target domain.
Approach:	They propose a domain-aware and co-adaptive feature transformation approach to address these issues by leveraging the target domain distribution features to guide the domain-based feature transformations.
Outcome:	The proposed method outperforms existing models and achieves state-of-the-art performance on a benchmark dataset.

On the Perception Bottleneck of VLMs for Chart Understanding (2025.findings-emnlp)

Copied to clipboard

Junteng Liu, Weihao Zeng, Xiwen Zhang, Yijun Wang, Zifei Shan, Junxian He

Visual Data in NLP

Challenge:	a perception bottleneck in large vision-language models is critical for chart understanding . instruction tuning improves the extraction capability of LVLMs, but the vision encoder remains a bottleneck .
Approach:	They propose to decompose the perception bottleneck into two components . the vision encoder bottleneck is where visual representation fails to encapsulate the correct information .
Outcome:	The proposed approach significantly mitigates the vision encoder bottleneck and improves the ability of LVLMs to comprehend charts.

Seer Self-Consistency: Advance Budget Estimation for Adaptive Test-Time Scaling (2026.findings-acl)

Copied to clipboard

Shiyu Ji, Yixuan Wang, Yijun Liu, Qingfu Zhu, Wanxiang Che

Challenge:	Large Language Models (LLMs) have demonstrated exceptional performance in reasoning tasks, particularly in mathematics.
Approach:	They propose a dynamic self-consistency framework that integrates System 1 and System 2 reasoning to improve token efficiency and latency.
Outcome:	The proposed method outperforms existing methods, achieving up to 47% reduction in token consumption and 43% reduction in inference latency without significant performance loss.

Mosaic-IT: Cost-Free Compositional Data Synthesis for Instruction Tuning (2025.findings-acl)

Copied to clipboard

Ming Li, Pei Chen, Chenguang Wang, Hongyu Zhao, Yijun Liang, YuPeng Hou, Fuxiao Liu, Tianyi Zhou

Responsible & Trustworthy NLP

Challenge:	Current instruction tuning relies on teacher models or human intervention to generate and refine the instructions and responses for training, which are costly, non-sustainable, and may lack diversity.
Approach:	They propose a human/model-free compositional data synthesis method that can create rich and diverse augmentations from existing instruction tuning data to enhance large language models.
Outcome:	The proposed method improves performance over benchmarks and reduces training costs by 80% compared with original instruction tuning.

Improving Grammatical Error Correction via Contextual Data Augmentation (2024.findings-acl)

Copied to clipboard

Yixuan Wang, Baoxin Wang, Yijun Liu, Qingfu Zhu, Dayong Wu, Wanxiang Che

Low-Resource NLP

Challenge:	Increasing use of synthetic data due to inconsistent error distribution and noisy labels is limiting the use of these data.
Approach:	They propose a method for augmentation of synthetic data with a more consistent error distribution.
Outcome:	The proposed method outperforms strong baselines and achieves state-of-the-art with only a few synthetic data.

Chain-of-Talkers (CoTalk): Fast Human Annotation of Dense Image Captions (2025.emnlp-main)

Copied to clipboard

Yijun Shen, Delong Chen, Fan Liu, Xingyu Wang, Chuanyi Zhang, Liang Yao, Yuhui Zheng

Multimodality

Challenge:	Existing approaches for optimizing human annotation efforts are limited . et al., 2015) suggest that densely annotated image captions improve vision-language alignment .
Approach:	They propose an AI-in-the-loop methodology to maximize the number of annotated samples and improve their comprehensiveness under fixed budget constraints.
Outcome:	The proposed method improves annotation speed and retrieval performance over the parallel method.

LM-Combiner: A Contextual Rewriting Model for Chinese Grammatical Error Correction (2024.lrec-main)

Copied to clipboard

Yixuan Wang, Baoxin Wang, Yijun Liu, Dayong Wu, Wanxiang Che

Text Generation

Challenge:	Recent work using model ensemble methods based on voting can effectively mitigate over-correction and improve the precision of the GEC system.
Approach:	They propose a rewriting model that can directly modify the over-correction of GEC system outputs without a model ensemble.
Outcome:	The proposed model can mitigate over-correction and improve accuracy of Chinese grammatical error correction tasks without a model ensemble.

Few Clean Instances Help Denoising Distant Supervision (2022.coling-1)

Copied to clipboard

Yufang Liu, Ziyin Huang, Yijun Wang, Changzhi Sun, Man Lan, Yuanbin Wu, Xiaofeng Mou, Ding Wang

Challenge:	Existing distantly supervised entity relation extractors rely on noisy data for training and evaluation.
Approach:	They propose a criterion for clean instance selection based on influence functions to collect sample-level evidence for recognizing good instances.
Outcome:	The proposed method shows strong performance on real and synthetic noisy datasets.

Democratizing Large Language Models via Personalized Parameter-Efficient Fine-tuning (2024.emnlp-main)

Copied to clipboard

Zhaoxuan Tan, Qingkai Zeng, Yijun Tian, Zheyuan Liu, Bing Yin, Meng Jiang

Responsible & Trustworthy NLP

Challenge:	Experimental results demonstrate that OPPU significantly outperforms existing prompt-based methods across seven diverse tasks in the LaMP benchmark.
Approach:	They propose to integrate parametric user knowledge into the personal PEFT parameters and non-parametric knowledge from retrieval and profiles, adapting LLMs to user behavior shifts.
Outcome:	The proposed method outperforms existing prompt-based methods across seven diverse tasks in the LaMP benchmark.

Confusion is the Final Barrier: Rethinking Jailbreak Evaluation and Investigating the Real Misuse Threat of LLMs (2025.findings-emnlp)

Copied to clipboard

Yu Yan, Sheng Sun, Zhe Wang, Yijun Lin, Zenghao Duan, Zhifei Zheng, Min Liu, Zhiyi Yin, Jianping Zhang

Responsible & Trustworthy NLP

Challenge:	Large Language Models have been developed to deal with real-world crimes, but it remains unclear whether they internalize authentic knowledge or are forced to simulate toxic language patterns.
Approach:	They construct knowledge-intensive Q&A to investigate misuse threats of Large Language Models in terms of dangerous knowledge possession, harmful task planning utility, and harmfulness judgment robustness.
Outcome:	The findings raise concerns that jailbreak success is often attributable to a hallucination loop between jailbroken LLM and judger LLM .