Papers by Jiale Cheng
One Cognitive Loop Is Enough: SODA unlocks Pure-Text Spatial Reasoning in Large Language Models (2026.acl-long)
Copied to clipboard
Shunwen Bai, Jiahuan Zhang, Haoran Huang, Yurun Wang, Jiale Liu, Yanxi Wu, Ningzhe Yu, Yudong Gao, Mingjun Cheng
| Challenge: | Existing large language models (LLMs) lack visual input, leading to errors in basic numerical comparisons. |
| Approach: | They propose a spatial OODA framework that integrates the OODAC cognitive loop into multiple control tasks and integrates it into LLMs. |
| Outcome: | The proposed model significantly improves the spatial reasoning capabilities of large language models across multiple scenarios including SPOD-Bench, SPACE and applications. |
Data Efficient RLVR via Off-Policy Influence Guidance (2026.acl-long)
Copied to clipboard
Erle Zhu, Dazhi Jiang, Yuan Wang, Xujun Li, Jiale Cheng, Yuxian Gu, Yilin Niu, Aohan Zeng, Jie Tang, Minlie Huang, Hongning Wang
| Challenge: | Existing data selection methods for RLVR are heuristic-based, lacking theoretical guarantees and generalizability. |
| Approach: | They propose an off-policy influence estimation method that approximates data influence using offline trajectories. |
| Outcome: | The proposed method reduces the computational cost of policy rollouts and improves storage and computation efficiency. |
InstructSafety: A Unified Framework for Building Multidimensional and Explainable Safety Detector through Instruction Tuning (2023.findings-emnlp)
Copied to clipboard
| Challenge: | Existing safety detection systems have limitations in terms of their versatility and interpretability. |
| Approach: | They introduce a safety detection framework that unifies 7 common sub-tasks into a uniform formulation and process 39 human-annotated datasets for instruction tuning. |
| Outcome: | The proposed framework unifies 7 common sub-tasks into a uniform formulation and then runs on 39 human-annotated datasets to fine-tune it. |
AutoDetect: Towards a Unified Framework for Automated Weakness Detection in Large Language Models (2024.findings-emnlp)
Copied to clipboard
Jiale Cheng, Yida Lu, Xiaotao Gu, Pei Ke, Xiao Liu, Yuxiao Dong, Hongning Wang, Jie Tang, Minlie Huang
| Challenge: | Large Language Models (LLMs) exhibit significant but subtle weaknesses, such as mistakes in instruction-following or coding tasks. |
| Approach: | They propose a framework to automatically expose weaknesses in Large Language Models (LLMs) they use three LLM-powered agents to perform comprehensive weakness identification . |
| Outcome: | The proposed framework shows that it is more effective than untargeted data augmentation methods like Self-Instruct to identify weaknesses in LLMs. |
LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety (2026.acl-long)
Copied to clipboard
Junxiao Yang, Haoran Liu, Jinzhe Tu, Jiale Cheng, Zhexin Zhang, Shiyao Cui, Jiaqi Weng, Jialing Tao, Hui Xue, Hongning Wang, Han Qiu, Minlie Huang
| Challenge: | Large language models (LLMs) have demonstrated better safety performance in high-resource languages than in low-resourced languages. |
| Approach: | They propose language-agnostic semantic alignment (LASA) which anchors safety alignment directly in semantic bottlenecks. |
| Outcome: | The proposed approach significantly improves safety across all languages: average attack success rate drops from 24.7% to 2.8% on LLaMA-3.1-8B-Instruct and remains within 3–4% across Qwen2.5 and Qwend3 Instruct models (7B–32B). |
AlignBench: Benchmarking Chinese Alignment of Large Language Models (2024.acl-long)
Copied to clipboard
Xiao Liu, Xuanyu Lei, Shengyuan Wang, Yue Huang, Andrew Feng, Bosi Wen, Jiale Cheng, Pei Ke, Yifan Xu, Weng Lam Tam, Xiaohan Zhang, Lichao Sun, Xiaotao Gu, Hongning Wang, Jing Zhang, Minlie Huang, Yuxiao Dong, Jie Tang
| Challenge: | Effective evaluation of alignment for emerging Chinese LLMs is still significantly lacking, calling for real-scenario grounded, open-ended, challenging and automatic evaluations tailored for alignment. |
| Approach: | They propose a multi-dimensional benchmark for evaluating LLMs’ alignment in Chinese with 8 main categories, 683 real-scenario rooted queries and corresponding human verified references. |
| Outcome: | The benchmark uses a human-in-the-loop data curation pipeline, 683 real-scenario rooted queries and human verified references. |
A Survey of Post-Training Scaling in Large Language Models (2025.acl-long)
Copied to clipboard
Hanyu Lai, Xiao Liu, Junjie Gao, Jiale Cheng, Zehan Qi, Yifan Xu, Shuntian Yao, Dan Zhang, Jinhua Du, Zhenyu Hou, Xin Lv, Minlie Huang, Yuxiao Dong, Jie Tang
| Challenge: | Large language models (LLMs) have demonstrated proficiency in understanding and generating human natural languages. |
| Approach: | They propose a framework for scaling large language models using supervised fine-tuning, RLxF and test-time compute methodologies. |
| Outcome: | The proposed model can be used to understand and generate human natural languages. |
Constructing Highly Inductive Contexts for Dialogue Safety through Controllable Reverse Generation (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods to detect toxic generation of pretrained language models rely on templates, data extraction, crowdsourcing workers or automatic generation. |
| Approach: | They propose a method to construct adversarial contexts conditioned on a given response . they augment existing dataset BAD+ and construct a new dataset B AD+ . |
| Outcome: | The proposed method can detect toxic or biased content in large pretrained language models. |
Making Pre-trained Language Models Better Continual Few-Shot Relation Extractors (2024.lrec-main)
Copied to clipboard
| Challenge: | Existing methods to learn new relations with limited labeled data are prone to catastrophic forgetting and overfitting. |
| Approach: | They propose a framework that uses prompts to acquire more generalized knowledge . they propose CFRE to continuously learn new relations while retaining knowledge of old ones . |
| Outcome: | The proposed method outperforms state-of-the-art methods by a large margin and significantly mitigates catastrophic forgetting and overfitting in low-resource scenarios. |
Training Language Model to Critique for Better Refinement (2025.findings-acl)
Copied to clipboard
Tianshu Yu, Chao Xiang, Mingchuan Yang, Pei Ke, Bosi Wen, Cunxiang Wang, Jiale Cheng, Li Zhang, Xinyu Mu, Chuxiong Sun, Minlie Huang
| Challenge: | Large language models (LLMs) have remarkable evaluation and critique capabilities, providing insightful feedback and identifying flaws in various tasks. |
| Approach: | They propose a framework to train critic models using refinement signals to generate feedback loops where critiques guide the model in refining its responses. |
| Outcome: | The proposed framework outperforms traditional methods and open-source models in terms of critique quality and refinement outcomes. |
Glyph: Scaling Context Windows via Visual-Text Compression (2026.acl-long)
Copied to clipboard
Jiale Cheng, Yusen Liu, Xinyu Zhang, Yulin Fei, Wenyi Hong, Ruiliang Lyu, Weihan Wang, Zhe Su, Xiaotao Gu, Xiao Liu, Yushi Bai, Jie Tang, Hongning Wang, Minlie Huang
| Challenge: | Large language models (LLMs) traditionally represent text as sequences of discrete tokens . a long-context scaling problem requires processing more tokens more efficiently . |
| Approach: | They propose a framework that renders long texts into compact visual pages and processes them with a vision-language model. |
| Outcome: | The proposed framework renders long texts into compact visual pages and processes them with a vision-language model. |
LogicGame: Benchmarking Rule-Based Reasoning Abilities of Large Language Models (2025.findings-acl)
Copied to clipboard
Jiayi Gui, Yiming Liu, Jiale Cheng, Xiaotao Gu, Xiao Liu, Hongning Wang, Yuxiao Dong, Jie Tang, Minlie Huang
| Challenge: | Large Language Models (LLMs) have demonstrated notable capabilities across various tasks, showcasing complex problem-solving abilities. |
| Approach: | They propose a benchmark to evaluate the rule-based logical reasoning capabilities of Large Language Models (LLMs) they create simulated scenarios in which models execute or plan operations to achieve specific outcomes. |
| Outcome: | The proposed benchmark evaluates the performance of large language models on a variety of scenarios with varying difficulty levels. |
On the Safety of Conversational Models: Taxonomy, Dataset, and Benchmark (2022.findings-acl)
Copied to clipboard
Hao Sun, Guangxuan Xu, Jiawen Deng, Jiale Cheng, Chujie Zheng, Hao Zhou, Nanyun Peng, Xiaoyan Zhu, Minlie Huang
| Challenge: | Dialogue safety problems severely limit the real-world deployment of generative conversational models. |
| Approach: | They propose a taxonomy for dialogue safety specifically designed to capture unsafe behaviors in human-bot dialogue settings. |
| Outcome: | The proposed taxonomy captures unsafe behaviors in human-bot dialogue settings with rich context-sensitive unsafe examples. |
VideoQA-TA: Temporal-Aware Multi-Modal Video Question Answering (2025.coling-main)
Copied to clipboard
| Challenge: | Existing methods for video question answering align visual or textual features directly with large language models, limiting the deep semantic association between modalities and hindering a comprehensive understanding of interactions within spatial and temporal contexts. |
| Approach: | They propose a temporal-aware framework for multi-modal video question answering that aligns videos and questions at fine-grained levels. |
| Outcome: | The proposed framework improves reasoning ability and accuracy of videoQA by aligning videos and questions at fine-grained levels. |
PAL: Persona-Augmented Emotional Support Conversation Generation (2023.findings-acl)
Copied to clipboard
| Challenge: | Recent work has demonstrated the effectiveness of dialogue models in providing emotional support due to the lack of human resources for mental health support. |
| Approach: | They propose a framework for dynamically inferring and modeling seekers’ persona from the conversation history and a model that leverages persona information to provide personalized emotional support. |
| Outcome: | The proposed model outperforms baseline models on the studied benchmark. |
Black-Box Prompt Optimization: Aligning Large Language Models without Model Training (2024.acl-long)
Copied to clipboard
| Challenge: | Large language models are often not well aligned with human intents, which requires additional training. |
| Approach: | They propose to use Black-Box Prompt Optimization (BPO) to perform alignments on large language models that are not well aligned with human intents. |
| Outcome: | The proposed model outperforms existing models and is model-agnostic. |
LongSafety: Evaluating Long-Context Safety of Large Language Models (2025.acl-long)
Copied to clipboard
Yida Lu, Jiale Cheng, Zhexin Zhang, Shiyao Cui, Cunxiang Wang, Xiaotao Gu, Yuxiao Dong, Jie Tang, Hongning Wang, Minlie Huang
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities in understanding and generating long sequences. |
| Approach: | They propose a benchmark to evaluate LLM safety in open-ended long-context tasks . they find that relevant context and extended input sequences can exacerbate safety risks . |
| Outcome: | The proposed benchmark identifies significant safety vulnerabilities in 16 LLMs . strong safety performance in short-context scenarios does not correlate with safety in long-contact tasks . |
Modelling Long-distance Node Relations for KBQA with Global Dynamic Graph (2020.coling-main)
Copied to clipboard
| Challenge: | Existing studies rely on deep graph neural networks (GNNs) to capture rich structural information, but they lack the structural information needed for QA. |
| Approach: | They propose a framework which captures structural information from KBs and models long-distance node relations from two perspectives. |
| Outcome: | The proposed framework models long-distance node relations from two perspectives . it is based on two widely used multi-hop KBQA datasets . |
CritiqueLLM: Towards an Informative Critique Generation Model for Evaluation of Large Language Model Generation (2024.acl-long)
Copied to clipboard
Pei Ke, Bosi Wen, Andrew Feng, Xiao Liu, Xuanyu Lei, Jiale Cheng, Shengyuan Wang, Aohan Zeng, Yuxiao Dong, Hongning Wang, Jie Tang, Minlie Huang
| Challenge: | Existing models for NLP evaluations lack the ability to generate informative critiques in pointwise grading and pairwise comparison especially without references. |
| Approach: | They propose a method which can acquire pointwise grading critiques with pseudo references and revise these critiques via multi-path prompting to obtain informative evaluation data in different tasks and settings. |
| Outcome: | The proposed method outperforms all open-source models and even GPT-4 in system-level correlations of pointwise grading. |
Beyond A Fixed Seal: Adaptive Stealing Watermark in Large Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing detection methods for large language models rely on fixed strategies to steal watermarks. |
| Approach: | They propose a novel steal-based watermark algorithm that derives watermark information from watermarked texts to craft highly targeted adversarial attacks. |
| Outcome: | The proposed system significantly increases steal efficiency against target watermarks under identical conditions. |
When and What to Ask: AskBench and Rubric-Guided RLVR for LLM Clarification (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models respond even when prompts omit critical details or include misleading information, leading to hallucinations or reinforced misconceptions. |
| Approach: | They propose an interactive benchmark that converts standard QA pairs into multi-turn interactions with explicit checkpoints. |
| Outcome: | The proposed benchmark improves accuracy, rubric adherence, and interaction efficiency with strong generalization to unseen domains. |
Exploring Task Difficulty for Few-Shot Relation Extraction (2021.emnlp-main)
Copied to clipboard
| Challenge: | Existing models do not distinguish hard tasks from easy ones in the learning process. |
| Approach: | They propose a novel approach that exploits relation label information to learn better representations by focusing on hard tasks. |
| Outcome: | Experiments on two standard datasets show the proposed approach performs better than previous methods. |
CM-TTS: Enhancing Real Time Text-to-Speech Synthesis Efficiency through Weighted Samplers and Consistency Models (2024.findings-naacl)
Copied to clipboard
| Challenge: | Neural Text-to-Speech systems are a promising approach for high-fidelity speech synthesis . but the efficiency of multi-step sampling in Diffusion Models presents challenges . |
| Approach: | They propose a novel architecture grounded in consistency models to improve model convergence. |
| Outcome: | The proposed architecture achieves top-quality speech synthesis in fewer steps without adversarial training or pre-trained model dependencies. |
Open Domain Question Answering based on Text Enhanced Knowledge Graph with Hyperedge Infusion (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods to improve knowledge base are incomplete and difficult to understand. |
| Approach: | They propose a novel QA method by leveraging text information to enhance the incomplete KB. |
| Outcome: | Extensive experiments on the WebQuestionsSP benchmark prove the effectiveness of the proposed model. |
Generative Prompt Tuning for Relation Classification (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing prompt tuning methods for RC are limited by label spaces and rigid prompt restrictions. |
| Approach: | They propose a generative prompt tuning method to reformulate relation classification as an infilling problem by adding cloze-style phrases to masked language modeling problems. |
| Outcome: | The proposed method exploits rich semantics of entity and relation types and can predict label verbalizations with varying lengths at multiple predicted positions. |