Papers by He Bai
Meta-rater: A Multi-dimensional Data Selection Method for Pre-training Language Models (2025.acl-long)
Copied to clipboard
Xinlin Zhuang, Jiahui Peng, Ren Ma, Yinfan Wang, Tianyi Bai, Xingjian Wei, Qiu Jiantao, Chi Zhang, Ying Qian, Conghui He
| Challenge: | composition of pre-training datasets for large language models remains undisclosed . current methods for evaluating data quality are limited by single-dimensional evaluation or redundancy-focused strategies. |
| Approach: | They propose a multi-dimensional data selection method that integrates dimensions with existing quality metrics through learned optimal weightings. |
| Outcome: | The proposed method doubles convergence speed for 1.3B model models and improves downstream task performance by 3.23%. |
Semantics of the Unwritten: The Effect of End of Paragraph and Sequence Tokens on Text Generation with GPT2 (2021.acl-srw)
Copied to clipboard
| Challenge: | Experimental results show that pre-trained language model GPT2 can generate better continuations by learning to generate the in the fine-tuning stage. |
| Approach: | They conduct experiments on an English essay dataset using Chinese-GPT2 . they find that the model can generate better continuations by learning to generate the in the fine-tuning stage. |
| Outcome: | The pre-trained language model GPT2 can generate better continuations by learning to generate the in the fine-tuning stage. |
Memory Consolidation for Contextual Spoken Language Understanding with Dialogue Logistic Inference (P19-1)
Copied to clipboard
| Challenge: | Existing models for SLU use explicit memory representations, but the context memory is under-exploited. |
| Approach: | They propose a dialogue logistic inference task to consolidate the context memory with SLU in a multi-task framework. |
| Outcome: | The proposed model improves slot filling and domain classification performance in a multi-task framework. |
GRAIT: Gradient-Driven Refusal-Aware Instruction Tuning for Effective Hallucination Mitigation (2025.findings-naacl)
Copied to clipboard
Runchuan Zhu, Xinke Jiang, Jiang Wu, Zhipeng Ma, Jiahe Song, Fengshuo Bai, Dahua Lin, Lijun Wu, Conghui He
| Challenge: | Experimental evaluations on open-ended and multiple-choice questions demonstrate GRAIT significantly outperforms existing RAIT methods in the overall performance. |
| Approach: | They propose a framework to reduce the risk of over-refusal and reduce hallucinations by rejecting unknown questions to minimize hallucinism and ensuring correct answers are not rejected. |
| Outcome: | The proposed framework outperforms existing methods on open-ended and multiple-choice questions. |
MinerU2.5: A Decoupled Vision-Language Model for Efficient High-Resolution Document Parsing (2026.acl-industry)
Copied to clipboard
Junbo Niu, Zheng Liu, Zhuangcheng Gu, Bin Wang, Linke Ouyang, Zhiyuan Zhao, Tao Chu, Tianyao He, Fan Wu, Qintong Zhang, Zhenjiang Jin, Guang Liang, Rui Zhang, Wenzheng Zhang, Yuan Qu, Zhifei Ren, Yuefeng Sun, Zirui Tang, Boyu Niu, Yuanhong Zheng, Dongsheng Ma, Ziyang Miao, Hejun Dong, Siyi Qian, Junyuan Zhang, Fangdong Wang, Jingzhou Chen, Xiaomeng Zhao, Liqun Wei, Wei Li, Shasha Wang, RuiLiang Xu, Yuanyuan Cao, Lu Chen, Qianqian Wu, Huaiyu Gu, Lindong Lu, Dechen Lin, null Shenguanlin, Xuanhe Zhou, Linfeng Zhang, Yuhang Zang, Xiaoyi Dong, Jiaqi Wang, Bo Zhang, Lei Bai, Pei Chu, Weijia Li, Jiang Wu, Lijun Wu, Zhenxiang Li, Guangyu Wang, Zhongying Tu, Chao Xu, Kai Chen, Bowen Zhou, Dahua Lin, Wentao Zhang, Conghui He
| Challenge: | Document images are characterized by higher resolutions, denser content, and more complex structural layouts. |
| Approach: | They propose a 1.2B-parameter document parsing vision-language model that decouples layout analysis from local content recognition. |
| Outcome: | The proposed model surpasses general-purpose and domain-specific models on multiple benchmarks while maintaining significantly lower computational overhead. |
Robustness via Referencing: Defending against Prompt Injection Attacks by Referencing the Executed Instruction (2026.findings-acl)
Copied to clipboard
Yulin Chen, Haoran Li, Yuan Sui, Yue Liu, Yufei He, Xiaoling Bai, Chi Fei, Li Yabo, Haozhe Ma, Yangqiu Song, Bryan Hooi
| Challenge: | Prompt injection attacks manipulate large language models (LLMs) by misleading them to deviate from the original input instructions and execute maliciously injected instructions. |
| Approach: | They propose a prompt injection defense method that suppresses the model's instruction-following tendencies rather than suppressing them. |
| Outcome: | The proposed method outperforms prompt-engineering-based approaches and fine-tuning methods and reduces the ASR to nearly 0% in some scenarios. |
RESF: Regularized-Entropy-Sensitive Fingerprinting for Black-Box Tamper Detection of Large Language Models (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for tamper detection rely on model stability, not inherently stochastic models. |
| Approach: | They propose a hypothesis-testing method for black-box tamper detection for LLMs . they propose regularized entropy-sensitive fingerprinting to enable efficient fingerprinting . |
| Outcome: | The proposed method achieves 98.80% detection accuracy under challenging conditions . it is based on a first-order surrogate for KL divergence to identify prompts most responsive to parameter perturbations. |
Let’s Rectify Step by Step: Improving Aspect-based Sentiment Analysis with Diffusion Models (2024.lrec-main)
Copied to clipboard
| Challenge: | Empirical evaluations conducted on eight benchmark datasets underscore the compelling advantages offered by DiffusionABSA when compared against robust baseline models. |
| Approach: | They propose a diffusion model which extracts aspects step by step and learns a denoising process that progressively restores them in a reverse manner. |
| Outcome: | Empirical evaluations on eight benchmark datasets underscore the compelling advantages offered by DiffusionABSA when compared against robust baseline models. |
Cross-Lingual Training of Neural Models for Document Ranking (2020.findings-emnlp)
Copied to clipboard
| Challenge: | a recent study shows that multi-lingual BERT models can be used for document ranking in non-English languages . a blog post by Google suggests that the company is exploring this approach to improve web search across a number of languages. |
| Approach: | They propose to leverage relevance judgments in English to train neural document ranking models for mono-lingual retrieval in multiple target languages. |
| Outcome: | The proposed approach improves search quality in non-English languages while requiring low resources. |
Cross-lingual Text-to-SQL Semantic Parsing with Representation Mixup (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Experimental results show that Rex can benefit from cross-lingual training and improve the effectiveness of semantic parsers. |
| Approach: | They propose a Representation Mixup Framework for effectively exploiting translations in the cross-lingual Text-to-SQL task. |
| Outcome: | The proposed framework can benefit from cross-lingual training and improve the effectiveness of semantic parsers, achieving state-of-the-art performance. |
Efficient Pretraining Data Selection for Language Models via Multi-Actor Collaboration (2025.acl-long)
Copied to clipboard
Tianyi Bai, Ling Yang, Zhen Hao Wong, Fupeng Sun, Xinlin Zhuang, Jiahui Peng, Chi Zhang, Lijun Wu, Qiu Jiantao, Wentao Zhang, Binhang Yuan, Conghui He
| Challenge: | Efficient data selection is crucial to accelerate the pretraining of language models . limited research has addressed the inherent conflicts between data selection methods . |
| Approach: | They propose a multi-actor collaborative data selection mechanism that prioritizes data based on its specific criterion and updates prioritization rules using the current state of the model. |
| Outcome: | The proposed model accelerates convergence in LM pretraining and achieves an average relative performance gain of 10.5% across multiple language model benchmarks. |
AlphaOne: Reasoning Models Thinking Slow and Fast at Test Time (2025.emnlp-main)
Copied to clipboard
Junyu Zhang, Runpei Dong, Han Wang, Xuying Ning, Haoran Geng, Peihao Li, Xialin He, Yutong Bai, Jitendra Malik, Saurabh Gupta, Huan Zhang
| Challenge: | Existing monotonic scaling methods for large reasoning models are not reliable. |
| Approach: | They propose a universal framework for modulating reasoning progress in large reasoning models at test time. |
| Outcome: | The proposed framework unifies and generalizes existing monotonic scaling methods and enables flexible and dense slow-to-fast reasoning modulation. |
ART: rule bAsed futuRe-inference deducTion (2023.emnlp-main)
Copied to clipboard
Mengze Li, Tianqi Zhao, Bai Jionghao, Baoyi He, Jiaxu Miao, Wei Ji, Zheqi Lv, Zhou Zhao, Shengyu Zhang, Wenqiao Zhang, Fei Wu
| Challenge: | Existing studies focus on language-based premises and deduce valid conclusions from visual observations. |
| Approach: | They propose a rule-based deductive reasoning task that uses video to deduce the correct future event . they use commonsense knowledge to annotate video and a strong baseline to conduct reasoning . |
| Outcome: | Empirical studies validate the rationality of ARTNet in deductive reasoning upon visual observations . ART is a method that rigorously follows a set of explicit constraints to deduce valid conclusions from empirical facts . |
Chumor 2.0: Towards Better Benchmarking Chinese Humor Understanding from (Ruo Zhi Ba) (2025.findings-acl)
Copied to clipboard
Ruiqi He, Yushu He, Longju Bai, Jiarui Liu, Zhenjie Sun, Zenghao Tang, He Wang, Hanchen Xia, Rada Mihalcea, Naihao Deng
| Challenge: | Existing studies on humor in non-English languages lack culturally nuanced humor in other languages. |
| Approach: | They construct a Chinese humor explanation dataset using a reddit-like platform . they test ten LLMs and find they are significantly better than existing LLM models . |
| Outcome: | The proposed dataset is the first and largest Chinese humor explanation dataset. |
Better Language Model with Hypernym Class Prediction (2022.acl-long)
Copied to clipboard
| Challenge: | Class-based language models (LMs) have been devised to address context sparsity in n-gram LMs for decades. |
| Approach: | They propose to use class-based prediction to improve generalization for rare words by annealing from predicting the class to token prediction during training. |
| Outcome: | The proposed model improves perplexity without sacrificing performance on rare words. |
MoleculeQA: A Dataset to Evaluate Factual Accuracy in Molecular Comprehension (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing models generate erroneous information and evaluations fail to assess factual correctness of models. |
| Approach: | They propose to use MoleculeQA to evaluate molecular factual correctness in large language models by organizing molecules into a taxonomy and building QA pairs through human and LLM efforts. |
| Outcome: | The proposed model improves the factual correctness of generated information and enables the development of new models. |
OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problems (2024.acl-long)
Copied to clipboard
Chaoqun He, Renjie Luo, Yuzhuo Bai, Shengding Hu, Zhen Thai, Junhao Shen, Jinyi Hu, Xu Han, Yujie Huang, Yuxiang Zhang, Jie Liu, Lei Qi, Zhiyuan Liu, Maosong Sun
| Challenge: | Large Language Models (LLMs) and Large Multimodal Models have exceeded general human capabilities in various tasks. |
| Approach: | They present an Olympiad-level bilingual multimodal scientific benchmark featuring 8,476 problems from Olympiad level mathematics and physics competitions. |
| Outcome: | The best performing model, GPT-4V, attains an average score of 17.97% on OlympiadBench, with a mere 10.74% in physics, highlighting the benchmark rigor and the intricacy of physical reasoning. |
Learning to Evolve: A Self-Improving Framework for Multi-Agent Systems via Textual Parameter Graph Optimization (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing methods for designing and optimizing multi-agent systems are static and do not learn from experience. |
| Approach: | They propose a framework that enables a multi-agent system to learn to evolve . they use "textual gradients" to pinpoint failures and suggest granular modifications . |
| Outcome: | a new framework enables a multi-agent system to learn to evolve . it learns from historical optimization experiences to improve its performance . |
MTR-DuplexBench: Towards a Comprehensive Evaluation of Multi-Round Conversations for Full-Duplex Speech Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing benchmarks focus on evaluating single-round interactions, neglecting other critical aspects. |
| Approach: | They propose a benchmark to evaluate full-duplex speech language models in multi-round settings . they segment continuous full-dual dialogues into discrete turns for evaluation . |
| Outcome: | The proposed benchmark compared full-duplex speech language models with full-dual speech models . the results show that the models perform better in multi-round settings than standard models compared to benchmarks . |
Pre-trained Language Model Based Active Learning for Sentence Matching (2020.coling-main)
Copied to clipboard
| Challenge: | Existing active learning approaches for natural language processing ignore the characteristics of natural language. |
| Approach: | They propose a pre-trained language model based active learning approach for sentence matching that provides linguistic criteria to measure instances and help select more effective instances for annotation. |
| Outcome: | The proposed approach can achieve greater accuracy with fewer labeled training instances. |
Incremental Intent Detection for Medical Domain with Contrast Replay Networks (2022.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to detect medical intents require fixed pre-defined intent categories . however, novel medical intent categories incessantly emerge with new data and intents in the real world . |
| Approach: | They propose to incrementally learn emerged medical intents from continually arriving data of new intents while avoiding catastrophically forgetting old ones. |
| Outcome: | The proposed method outperforms the state-of-the-art model on two benchmarks by 5.7% and 9.1% accuracy. |
Construction of Paired Knowledge Graph - Text Datasets Informed by Cyclic Evaluation (2024.lrec-main)
Copied to clipboard
Ali Mousavi, Xin Zhan, He Bai, Peng Shi, Theodoros Rekatsinas, Benjamin Han, Yunyao Li, Jeffrey Pound, Joshua M. Susskind, Natalie Schluter, Ihab F. Ilyas, Navdeep Jaitly
| Challenge: | Prior studies have shown that sequence-to-sequence models learn to hallucinate when the conditioning data has poor correlation with the sequence being produced. |
| Approach: | They construct a dataset that pairs Knowledge Graphs (KG) and text together and compare their results to a cyclic evaluation model. |
| Outcome: | The proposed model performs better on cyclic generation of KGs than on KG-T, but less well on synchronization of KTs. |
Rethinking Table Pruning in TableQA: From Sequential Revisions to Gold Trajectory-Supervised Parallel Search (2026.acl-long)
Copied to clipboard
Yu Guo, Shenghao Ye, Shuangwu Chen, Zijian Wen, Tao Zhang, Bai Qirui, Dong Jin, Yunpeng Hou, Huasen He, null Jianyang, Xiaobin Tan
| Challenge: | Existing pruning methods rely on sequential revisions and unreliable critique signals . Existing methods fail to detect the loss of answer-critical data . |
| Approach: | They propose a table pruning framework which transforms table pruning to gold trajectory-supervised parallel search. |
| Outcome: | The proposed framework outperforms the strongest baseline pruning framework by 3.2% on various tabular reasoning tasks. |
XRICL: Cross-lingual Retrieval-Augmented In-Context Learning for Cross-lingual Text-to-SQL Semantic Parsing (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Existing work focuses on English datasets, and it is unclear whether large language models can serve as competitive semantic parsers for other languages. |
| Approach: | They propose a framework that learns to retrieve relevant English exemplars for a given query to construct prompts. |
| Outcome: | The proposed framework learns to retrieve relevant English exemplars for a given query to construct prompts. |
EcomScriptBench: A Multi-task Benchmark for E-commerce Script Planning via Step-wise Intention-Driven Product Association (2025.acl-long)
Copied to clipboard
Weiqi Wang, Limeng Cui, Xin Liu, Sreyashi Nag, Wenju Xu, Chen Luo, Sheikh Muhammad Sarwar, Yang Li, Hansu Gu, Hui Liu, Changlong Yu, Jiaxin Bai, Yifan Gao, Haiyang Zhang, Qi He, Shuiwang Ji, Yangqiu Song
| Challenge: | Goal-oriented script planning is used by humans to plan for typical activities . however, this capability remains underexplored due to several challenges . |
| Approach: | They propose a framework that enables product-enriched scripts by associating products with each step based on the semantic similarity between the actions and their purchase intentions. |
| Outcome: | The proposed framework can generate product-enriched scripts from 2.4 million scripts . human annotations are conducted to provide gold labels for a sampled subset . |
Learning to Edit Knowledge via Instruction-based Chain-of-Thought Prompting (2026.acl-long)
Copied to clipboard
| Challenge: | Existing knowledge editing methods focus on structured fact triples, overlooking diverse unstructured forms of factual information. |
| Approach: | They propose a method that allows LLMs to edit knowledge via **Chain of Thoughts** reasoning. |
| Outcome: | The proposed method achieves strong generalization across six diverse knowledge editing scenarios with a single round of training on three open-source language models. |
A Survey of Reinforcement Learning for Large Language Models under Data Scarcity: Challenges and Solutions (2026.acl-long)
Copied to clipboard
Zhiyin Yu, Yuchen Mou, Juncheng Yan, Junyu Luo, Chunchun Chen, Xing Wei, Yunhui Liu, Hongru Sun, Yuxing Zhang, Jun Xu, Yatao Bian, Ming Zhang, Wei Ye, Tieke He, Jie Yang, Guanjie Zheng, Zhonghai Wu, Bo Zhang, Lei Bai, Xiao Luo
| Challenge: | Existing research on reinforcement learning for LLMs under data scarcity has not been unified. |
| Approach: | They propose a top-up hierarchical framework built around three complementary perspectives: data-centric, training-centric and framework-centric. |
| Outcome: | The proposed framework provides a clear conceptual foundation for understanding the design space of data-efficient RL for large language models and to guide researchers working in this emerging area. |
Syntax for Semantic Role Labeling, To Be, Or Not To Be (P18-1)
Copied to clipboard
| Challenge: | Existing neural SRL models lack syntactic backbone for performance, limiting its use in deep learning. |
| Approach: | They propose an enhanced argument labeling model with extended korder argument pruning algorithm for effectively exploiting syntactic information. |
| Outcome: | The proposed model achieves state-of-the-art on the CoNLL-2008 and 2009 benchmarks in English and Chinese. |
LongAlign: A Recipe for Long Context Alignment of Large Language Models (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Existing studies to build long context language models focus on context extension and continual training on long text. |
| Approach: | They propose a recipe for instruction fine-tuning on input sequences of similar length . they adopt packing and sorted batching strategies to speed up supervised fine-uning . |
| Outcome: | The proposed model outperforms existing recipes for LLMs in long context tasks by 30% while maintaining proficiency in handling short, generic tasks. |
IntentionQA: A Benchmark for Evaluating Purchase Intention Comprehension Abilities of Language Models in E-commerce (2024.findings-emnlp)
Copied to clipboard
Wenxuan Ding, Weiqi Wang, Sze Kwok, Minghao Liu, Tianqing Fang, Jiaxin Bai, Xin Liu, Changlong Yu, Zheng Li, Chen Luo, Qingyu Yin, Bing Yin, Junxian He, Yangqiu Song
| Challenge: | Existing approaches that distill intentions from LMs fail to generate meaningful and human-centric intentions applicable in real-world E-commerce contexts. |
| Approach: | They propose a double-task multiple-choice question answering benchmark to evaluate LMs' comprehension of purchase intentions in E-commerce. |
| Outcome: | The proposed benchmark consists of 4,360 carefully curated problems across three difficulty levels, constructed using an automated pipeline to ensure scalability on large E-commerce platforms. |
Source Critical Reinforcement Learning for Transferring Spoken Language Understanding to a New Language (C18-1)
Copied to clipboard
| Challenge: | a study aims to develop a language transferring system to avoid the trouble of acquiring and labeling a new big SLU corpus . general-purpose translators cannot handle the lot of semantic labels, not to mention cultural differences . a RL-based language transfer method can be used to adapt the adapted translator to a target language . |
| Approach: | They propose to use reinforcement learning to adapt a spoken language understanding model to a target language. |
| Outcome: | The proposed language transferring method improves domain classification accuracy by 22% compared with naive translation . the proposed language transfer method can be used on Chinese to English translators with more proper slot tags . |
Scaling Behaviors of LLM Reinforcement Learning Post-Training: An Empirical Study in Mathematical Reasoning (2026.acl-long)
Copied to clipboard
Zelin Tan, Hejia Geng, Xiaohang Yu, Mulei Zhang, Guancheng Wan, Yifan Zhou, Qiang He, Xiangyuan Xue, Heng Zhou, Yutao Fan, Zhong-Zhi Li, Zaibin Zhang, Guibin Zhang, Chen Zhang, Zhenfei Yin, Philip Torr, Lei Bai
| Challenge: | elucidating scaling laws for large language models (LLMs) during pre-training remains unexplored. |
| Approach: | They characterize how model scale, data, and compute interact during pre-training . they find that large models consistently demonstrate superior compute and data efficiency . |
| Outcome: | The proposed scaling laws offer practical guidance for scaling reasoning capabilities through reinforcement learning post-training. |
MT-Bench-101: A Fine-Grained Benchmark for Evaluating Large Language Models in Multi-Turn Dialogues (2024.acl-long)
Copied to clipboard
Ge Bai, Jie Liu, Xingyuan Bu, Yancheng He, Jiaheng Liu, Zhanhui Zhou, Zhuoran Lin, Wenbo Su, Tiezheng Ge, Bo Zheng, Wanli Ouyang
| Challenge: | Large Language Models (LLMs) have greatly enhanced dialogue systems, but evaluation of their capabilities remains a challenge. |
| Approach: | They propose a model to evaluate the fine-grained abilities of Large Language Models in multi-turn dialogues. |
| Outcome: | The proposed model evaluates 21 popular chatbots based on MT-Bench-101 . it includes 3 overarching abilities and 13 distinct tasks within multi-turn dialogue scenarios. |
GraphReader: Building Graph-based Agent to Enhance Long-Context Abilities of Large Language Models (2024.findings-emnlp)
Copied to clipboard
Shilong Li, Yancheng He, Hangyu Guo, Xingyuan Bu, Ge Bai, Jie Liu, Jiaheng Liu, Xingwei Qu, Yangguang Li, Wanli Ouyang, Wenbo Su, Bo Zheng
| Challenge: | Existing models for long contexts struggle to handle long inputs due to limited context window and memory usage. |
| Approach: | They propose a graph-based agent system that analyzes long texts into a graphical graph . GraphReader consistently outperforms GPT-4-128k across context lengths from 16k to 256k . |
| Outcome: | The proposed model outperforms existing models on four challenging benchmarks. |
LLM-KT: Enhancing Large Language Models with Knowledge Tracing via Multi-Level Plug-and-Play Alignment (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing methods to learn behavioral sequences fail to capture complex behavioral patterns due to a lack of deep reasoning capabilities and world knowledge. |
| Approach: | They propose a framework that integrates the reasoning power of Large Language Models with the sequential modeling strengths of traditional KT methods via multi-level plug-and-play alignment. |
| Outcome: | Extensive experiments on four standard datasets show that the proposed framework outperforms existing methods on state-of-the-art questions. |
CogAtom: From Cognitive Atoms to Olympiad-level Mathematical Reasoning in Large Language Models (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for generating high-quality, multi-step reasoning are limited . we present a new framework for synthesising rigorous, cognitively diverse problems . |
| Approach: | They propose a cognitive atom-based framework for synthesizing mathematically rigorous problems. |
| Outcome: | The proposed framework outperforms existing methods in accuracy, reasoning depth and diversity while exceeding the difficulty of AIME. |