Papers by Jingyu Peng
Harmonizing Diverse Models: A Layer-wise Merging Strategy for Consistent Generation (2025.emnlp-industry)
Copied to clipboard
| Challenge: | RAG systems often generate inconsistent outputs for semantically equivalent inputs . this unpredictability undermines the reliability of RAG and poses challenges for adoption in high-stakes or knowledge-sensitive domains such as finance, healthcare, and scientific research. |
| Approach: | They propose a method that integrates knowledge from specialized models into a single model to improve output consistency. |
| Outcome: | The proposed model significantly improves output consistency, achieving approximately 47.5% improvement in response similarity over baseline. |
On the Continued Value of Universal Dependencies in the Era of Large Language Models (2026.acl-long)
Copied to clipboard
| Challenge: | a growing belief that explicit linguistic representations are no longer necessary is questioned in large language models . a recent study examines whether and in what ways this cross-lingual syntactic framework can still benefit LLMs . |
| Approach: | They use Universal Dependencies (UD) to examine whether and in what ways it can still benefit LLMs. |
| Outcome: | The proposed model outperforms its syntax-agnostic counterparts in a cross-lingual evaluation task. |
Stepwise Reasoning Disruption Attack of LLMs (2025.acl-long)
Copied to clipboard
Jingyu Peng, Maolin Wang, Xiangyu Zhao, Kai Zhang, Wanyu Wang, Pengyue Jia, Qidong Liu, Ruocheng Guo, Qi Liu
| Challenge: | Existing attacks on LLM reasoning are constrained by specific settings or lack of imperceptibility, limiting their feasibility and generalizability. |
| Approach: | They propose a stepwise rEasoning error disruption attack that subtly injects errors into prior reasoning steps to mislead the model into producing incorrect subsequent reasoning and final answers. |
| Outcome: | The proposed attack is compatible with zero-shot and few-shot settings, maintains the natural reasoning flow, and ensures covert execution without modification of the instruction. |
JI2S: Joint Influence‐Aware Instruction Data Selection for Efficient Fine‐Tuning (2025.emnlp-main)
Copied to clipboard
| Challenge: | Prior selection strategies score samples using generalpurpose LLMs, leveraging their strong language understanding but introducing inherent biases that misalign with the target model’s behavior and yield unstable downstream performance. |
| Approach: | They propose a framework that jointly models marginal and combinatorial influences within sample groups and evaluate them on Open LLM Benchmarks, MTBench, and GPT4–judged pairwise comparisons. |
| Outcome: | The proposed framework outperforms fulldataset training and strong baselines on Open LLM Benchmarks, MTBench, and GPT4–judged pairwise comparisons. |
Logic Jailbreak: Efficiently Unlocking LLM Safety Restrictions Through Formal Logical Expression (2026.findings-acl)
Copied to clipboard
Jingyu Peng, Maolin Wang, Nan Wang, Jiatong Li, Yuchen Li, Yuyang Ye, Wanyu Wang, Pengyue Jia, Kai Zhang, Xiangyu Zhao
| Challenge: | Despite advances in aligning LLMs with human values, current safety mechanisms remain vulnerable to jailbreak attacks. |
| Approach: | They propose a black-box jailbreak method that uses logical expression translation to bypass LLM safety mechanisms. |
| Outcome: | The proposed method exploits the distributional gap between alignment data and logic-expressed inputs while preserving the underlying semantic intent and readability while evading safety constraints. |