Papers by Chengzhi Li

11 papers
Temporal Token Matters: Investigating and Interpreting the Consistency of Temporal Ordering in Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Large Language Models (LLMs) exhibit notable deficiencies in temporal reasoning . phrasing changes can lead LLMs to produce inconsistent outputs .
Approach: They investigate the mechanistic interpretability of temporal ordering within event temporal reasoning . they identify a sparse subset of attention heads that are causally responsible for reasoning outcomes .
Outcome: The proposed model outperforms other models in a variety of tasks and is validated by intervention-based experiments.
MMRC: A Large-Scale Benchmark for Understanding Multimodal Large Language Model in Real-World Conversation (2025.acl-long)

Copied to clipboard

Challenge: Existing multimodal large language models lack the ability to memorize, recall, and reason in sustained interactions.
Approach: They propose a multimodal real-world conversation benchmark for evaluating open-ended abilities of multimodal large language models.
Outcome: The proposed benchmarks show that the models perform better in open-ended conversations.
NovBench: Evaluating Large Language Models on Academic Paper Novelty Assessment (2026.findings-acl)

Copied to clipboard

Challenge: Existing methods for evaluating novelty have been proposed, but there is no systematic evaluation of their ability to generate novelty evaluations.
Approach: They propose a benchmark to evaluate large language models’ ability to generate novelty evaluations in support of human peer review.
Outcome: The proposed framework evaluates the quality of LLM-generated novelty evaluations under different prompting strategies.
Option Symbol Matters: Investigating and Mitigating Multiple-Choice Option Symbol Bias of Large Language Models (2025.naacl-long)

Copied to clipboard

Challenge: Multiple-Choice Question Answering (MCQA) is a widely used task in the evaluation of large language models (LLMs).
Approach: They propose a tuning-free, causal effect driven debiasing method which intervenes the activations of identified components according to their causal effects.
Outcome: The proposed method alleviates the aforementioned bias and improves the performance of LLMs.
Learning to Rewrite: Generalized LLM-Generated Text Detection (2025.acl-long)

Copied to clipboard

Challenge: Existing detectors for Large Language Models (LLMs) struggle to generalize in open-world settings.
Approach: They propose a framework to detect LLM-generated text with exceptional generalization to unseen domains by reinforcing LLMs’ inherent rewriting tendencies.
Outcome: The proposed framework outperforms state-of-the-art detection methods by 23.04% in AUROC, 35.10% for out-of distribution tests, and 48.66% under adversarial attacks.
Constructing Your Model’s Value Distinction: Towards LLM Alignment with Anchor Words Tuning (2025.findings-emnlp)

Copied to clipboard

Challenge: a study of large language models (LLMs) shows that they can generate outputs that are honest, positive, harmless, etc.
Approach: They propose a method that amplifies logits difference between positive and negative tokens . they propose to use the logits gap to generate positive and positive tokens after alignment .
Outcome: The proposed method achieves effective alignment, but requires fewer computational resources compared to training-time alignment methods.
How Do LLMs and VLMs Understand Viewpoint Rotation Without Vision? An Interpretability Study (2026.acl-long)

Copied to clipboard

Challenge: Existing studies on spatial intelligence from the perspective of visual-spatial intelligence have not explored whether visual intelligence alone is sufficient to endow models with spatial intelligence.
Approach: They propose to use a linguistic perspective to investigate spatial intelligence from a theoretical perspective.
Outcome: The proposed model performs poorly on the proposed dataset while human can easily achieve 100% accuracy.
RAFT: Realistic Attacks to Fool Text Detectors (2024.emnlp-main)

Copied to clipboard

Challenge: Large language models (LLMs) have exhibited remarkable fluency across tasks, but their unethical applications are unclear.
Approach: They propose a grammar error-free black-box attack that exploits LLM embeddings at the word-level while preserving original text quality.
Outcome: The proposed attack compromises all detectors across domains and is transferable across source models.
Memory or Reasoning? Explore How LLMs Compute Mixed Arithmetic Expressions (2025.findings-acl)

Copied to clipboard

Challenge: Large language models (LLMs) can solve complex multi-step math reasoning problems, but their internal implementation is limited.
Approach: They propose to use a "C**ausal **E**ffect **D**riven **F**ine-tuning method" to improve LLMs' reasoning ability.
Outcome: The proposed method improves the model's reasoning ability by enhancing key components that are used to execute mixed arithmetic calculations.
Can LLMs See Without Pixels? Benchmarking Spatial Intelligence from Textual Descriptions (2026.findings-acl)

Copied to clipboard

Challenge: Existing advances in Spatial Intelligence rely on vision-Language Models . however, a critical question remains: does spatial understanding originate from visual encoders?
Approach: They propose to evaluate the SI performance of Large Language Models without pixel-level input.
Outcome: The proposed benchmark challenges large language models to perform symbolic reasoning rather than visual pattern matching.
Encoding Conversation Context for Neural Keyphrase Extraction from Microblog Posts (N18-1)

Copied to clipboard

Challenge: Existing keyphrase extraction methods suffer from data sparsity problem when conducted on short and informal texts.
Approach: They propose a neural keyphrase extraction framework for microblog posts that takes conversation context into account and uses four types of neural encoders to represent conversation context.
Outcome: The proposed framework outperforms state-of-the-art keyphrase extraction methods on Twitter and Weibo datasets.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations