Papers by Jinghui Zhang
Jailbreak Open-Sourced Large Language Models via Enforced Decoding (2024.acl-long)
Copied to clipboard
Hangfan Zhang, Zhimeng Guo, Huaisheng Zhu, Bochuan Cao, Lu Lin, Jinyuan Jia, Jinghui Chen, Dinghao Wu
| Challenge: | Existing studies show that Large Language Models can be misused to generate undesired content. |
| Approach: | They propose to use large language models to manipulate the generation process to generate undesired content without heavy computations or prompt designs. |
| Outcome: | The proposed method shows that open-sourced large language models could be misused to generate undesired content without heavy computations or prompt designs. |
ServImage: An Image Generation and Editing Benchmark from Real-world Commercial Imaging Services (2026.acl-long)
Copied to clipboard
Fengxian Ji, Jingpu Yang, Zirui Song, Lang Gao, Junhong Liang, Zhenhao Chen, Jinghui Zhang, Xiuying Chen
| Challenge: | Recent image generation and editing models demonstrate robust adherence to instructions and high visual quality on academic benchmarks. |
| Approach: | They propose a benchmark that correlates image outputs with economic value in commercial design projects. |
| Outcome: | ServImage benchmarks show image generation models perform well on academic benchmarks but are uncertain on commercial projects. |
A Rationale-Centric Framework for Human-in-the-loop Machine Learning (2022.acl-long)
Copied to clipboard
| Challenge: | Experimental results show that RDL leads to significant prediction benefits on both in-distribution and out-of-district tests, especially for few-shot learning scenarios. |
| Approach: | They propose a rational-centric framework with human-in-the-loop to exploit spurious associations and bias models towards generally applicable underlying distributions. |
| Outcome: | The proposed framework leads to significant prediction benefits on in-distribution and out-of-district tests, compared to state-of the-art benchmarks. |
ToolCPT: Improving Tool Utilization in LLM Agents via Continuous Pre-training (2026.findings-acl)
Copied to clipboard
| Challenge: | Current approaches to enhancing tool use for LLM-based agents focus on post-training fine-tuning or test-time context extension. |
| Approach: | They propose to enhance tool knowledge for LLM-based agents during continuous pre-training . they curate 5.1 million code artifacts from large-scale, high-quality code repositories . |
| Outcome: | The proposed model outperforms existing methods on out-of-distribution tools on multiple benchmarks. |
WordGame: Efficient & Effective LLM Jailbreak via Simultaneous Obfuscation in Query and Response (2025.findings-naacl)
Copied to clipboard
| Challenge: | Recent advances in large language models have raised concerns about their susceptibility to jailbreaking attacks, which leads to harmful content inadvertently. |
| Approach: | They propose to exploit the safety alignment patterns of LLMs by simultaneous obfuscation in queries and responses to break down adversarial intent of query. |
| Outcome: | The proposed attack breaks down adversarial intent of query and encourages benign content regarding the games to precede anticipated harmful content in the response. |
PromptFix: Few-shot Backdoor Removal via Adversarial Prompt Tuning (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing studies have shown that pre-trained language models can be backdoored such that model behavior is manipulated when trigger tokens are presented. |
| Approach: | They propose a backdoor mitigation strategy for NLP models via adversarial prompt-tuning in few-shot settings that uses two extra sets of soft tokens which approximate the trigger and counteract it respectively. |
| Outcome: | The proposed method keeps model parameters intact and only utilizes two extra sets of soft tokens which approximate the trigger and counteract it respectively. |
Semantically-Aligned Universal Tree-Structured Solver for Math Word Problems (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing models focus on one-unknown linear MWPs. |
| Approach: | They propose a universal expression tree-structured solver that integrates multiple expression trees underlying a MWP into a single expression tree. |
| Outcome: | The proposed method outperforms state-of-the-art models on a MWPs dataset and generates a universal expression tree explicitly by deciding which symbol to generate . |
When Personalization Tricks Detectors: The Feature-Inversion Trap in Machine-Generated Text Detection (2026.acl-long)
Copied to clipboard
Lang Gao, Xuhui Li, Chenxi Wang, Mingzhe Li, Wei Liu, Zirui Song, Jinghui Zhang, Rui Yan, Preslav Nakov, Xiuying Chen
| Challenge: | Personalized MGT detection remains largely underexplored due to personalization challenges . large language models (LLMs) can imitate personal writing styles, but they can generate fake news and misinformation. |
| Approach: | They propose a benchmark to evaluate detector robustness under personalization . they attribute this limitation to a feature-inversion trap that flips the effect in personalized contexts . |
| Outcome: | The proposed framework predicts detector robustness under personalization with an 85% correlation to actual results. |
Phi: Preference Hijacking in Multi-modal Large Language Models at Inference Time (2025.emnlp-main)
Copied to clipboard
| Challenge: | Recent advances in Multimodal Large Language Models have raised serious safety concerns. |
| Approach: | They propose a method for manipulating the output preference of MLLMs using a preference hijacked image. |
| Outcome: | The proposed method works at inference time and requires no model modifications. |