Papers by Zhecheng Li
Mapping the Minds of LLMs: A Graph-Based Analysis of Reasoning LLMs (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large Reasoning Models (LRMs) often display unstable behaviors, e.g., hallucinating unsupported premises, overthinking simple tasks, and displaying higher sensitivity to prompt variations. |
| Approach: | They propose a graph-based analytical framework that clusters long, verbose CoT outputs into semantically coherent reasoning steps, then constructs directed reasoning graphs to capture contextual and logical dependencies among these steps. |
| Outcome: | The proposed framework enables quantitative evaluation of internal reasoning structure and quality beyond conventional metrics and provides practical insights for prompt engineering and cognitive analysis of LLMs. |
Are Large Language Models Reliable Reviewers? A Benchmark for Error Detection in Financial Documents (2026.findings-acl)
Copied to clipboard
Ying He, Zhouhong Gu, Zhecheng Hu, Yubo Zhou, Hao Shen, Jiaqing Liang, Zhaoqian Dai, Ma Shuguang, Fei Yu, Yanghua Xiao, Zhixu Li
| Challenge: | Existing LLMs struggle to identify errors in financial documents, a study shows . 18% of financial practitioners make errors daily, one-third make errors several times weekly, and 59% make errors multiple times monthly. |
| Approach: | They introduce FinED-Bench, a publicly available Benchmark for financial error detection . it covers nine real-world financial scenarios and includes over 900 documents in 2025 . supervised fine-tuning can significantly improve the performance of weaker LLMs, they show . |
| Outcome: | The proposed benchmark covers nine real-world financial scenarios and includes over 900 documents reported in 2025 that are unseen by existing language models. |
Vulnerability of LLMs to Vertically Aligned Text Manipulations (2025.acl-long)
Copied to clipboard
| Challenge: | Recent research shows that vertical text input significantly degrades the accuracy of large language models (LLMs) in text classification tasks. |
| Approach: | They investigate the impact of vertical text input on the performance of LLMs . they find that chain of thought reasoning does not help LLM recognize vertical input . |
| Outcome: | The proposed model can significantly mislead models, posing a risk of bypassing detection in real-world scenarios involving harmful or sensitive information. |
Too Big to Fail: Larger Language Models are Disproportionately Resilient to Induction of Dementia-Related Linguistic Anomalies (2024.findings-acl)
Copied to clipboard
| Challenge: | Existing studies show that the attention mechanism in transformer-based NLMs may present an analogue to the notions of cognitive and brain reserve. |
| Approach: | They propose a bidirectional ablation method that masks attention heads to display degradation of similar magnitude to masking in smaller models. |
| Outcome: | The proposed method exhibits properties attributed to the concepts of cognitive and brain reserve in human brain studies. |
Mitigating Confounding in Speech-Based Dementia Detection through Weight Masking (2025.acl-long)
Copied to clipboard
| Challenge: | Pre-trained neural language models fine-tuned on AD transcripts perform well, but little research has explored the effects of the gender of the speakers represented by these transcripts. |
| Approach: | They propose to use the Extended Confounding Filter and the Dual Filter to isolate and ablate weights associated with gender in dementia datasets. |
| Outcome: | The proposed methods overfit to training data distributions and disrupt gender-related weights, with the trade-off of slightly reduced dementia detection performance. |
Who is in the Spotlight: The Hidden Bias Undermining Multimodal Retrieval-Augmented Generation (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing RAG models are sensitive to the order in which evidence is presented, resulting in unstable performance and biased reasoning. |
| Approach: | They propose to quantify position bias in multimodal RAG systems by using position sensitivity index . they also develop a visualization framework to trace attention allocation patterns across decoder layers . |
| Outcome: | The proposed framework shows that multimodal interactions intensify position bias compared to unimodal settings and that this bias increases logarithmically with retrieval range. |
DRS: Deep Question Reformulation With Structured Output (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing models like GPT-3 and Instruct-GPT lack the ability to reformulate unanswerable questions. |
| Approach: | They propose a zero-shot method that combines the strengths of LLMs with a DFS-based algorithm to iteratively explore potential entity combinations and constrain outputs using predefined entities. |
| Outcome: | The proposed method outperforms all baselines, including the GPT-3.5 model, on the unanswerable question reformulation task. |