Papers by Wenhao Lu
Curriculum-RLAIF: Curriculum Alignment with Reinforcement Learning from AI Feedback (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to align large language models with human preferences are limited in generalizability due to distribution shift, preference label noise, and mismatch of challenging samples with model capacity. |
| Approach: | They propose a framework that constructs preference pairs with varying difficulty levels and then produces a specific curriculum for reward model training. |
| Outcome: | The proposed framework improves generalizability of reward models by a significant margin without incurring additional inference costs compared to existing non-curriculum baselines. |
Document Segmentation Matters for Retrieval-Augmented Generation (2025.findings-acl)
Copied to clipboard
Zhitong Wang, Cheng Gao, Chaojun Xiao, Yufei Huang, Shuzheng Si, Kangyang Luo, Yuzhuo Bai, Wenhao Li, Tangjian Duan, Chuancheng Lv, Guoshan Lu, Gang Chen, Fanchao Qi, Maosong Sun
| Challenge: | Existing rule-based chunking methods lead to suboptimal splits, where overly large chunks introduce irrelevant information and small chunks lack semantic coherence. |
| Approach: | They propose a method that leverages document summaries as pseudo-instructions to guide chunking by computing semantic similarity between sentences and the summary. |
| Outcome: | Experiments on multiple open-domain question-answering benchmarks show that PIC significantly improves retrieval accuracy (Hits@k) and end-to-end QA performance (Exact Match) without any additional training. |
STORM-BORN: A Challenging Mathematical Derivations Dataset Curated via a Human-in-the-Loop Multi-Agent Framework (2025.findings-acl)
Copied to clipboard
Wenhao Liu, Zhenyi Lu, Xinyu Hu, Jerry Zhang, Dailin Li, Jiacheng Cen, Huilin Cao, Haiteng Wang, Yuhan Li, Xie Kun, Dandan Li, Pei Zhang, Chengbo Zhang, Yuxiang Ren, Xiaohong Huang, Yan Ma
| Challenge: | Existing datasets suffer from outdated and insufficient challenging content, neglecting human-like reasoning, and limited reliability due to single-LLM generation. |
| Approach: | They propose a human-in-the-loop, multi-agent data generation framework that integrates reasoning-dense filters, multiagent collaboration, and human mathematicians’ evaluations to ensure the reliability and quality of the dataset. |
| Outcome: | The proposed framework improves accuracy and quality of the 2,000-synthesized datasets by integrating reasoning-dense filters, multi-agent collaboration, and human mathematicians’ evaluations. |
Tell Me What You Don’t Know: Enhancing Refusal Capabilities of Role-Playing Agents via Representation Space Analysis and Editing (2025.findings-acl)
Copied to clipboard
Wenhao Liu, Siyu An, Junru Lu, Muling Wu, Tianlong Li, Xiaohua Wang, Changze Lv, Xiaoqing Zheng, Di Yin, Xing Sun, Xuanjing Huang
| Challenge: | Role-playing Agents (RPAs) struggle to recognize and respond to hard queries that conflict with their role-play knowledge. |
| Approach: | They propose a lightweight representation editing approach that conveniently shifts conflicting requests to the rejection region, thereby enhancing the model’s refusal accuracy. |
| Outcome: | The proposed model improves RPAs’ refusal ability of conflicting requests while maintaining their general role-playing capabilities. |
Teaching According to Talents! Instruction Tuning LLMs with Competence-Aware Curriculum Learning (2025.findings-emnlp)
Copied to clipboard
Yangning Li, Tingwei Lu, Yinghui Li, Yankai Chen, Wei-Chieh Huang, Wenhao Jiang, Hui Wang, Hai-Tao Zheng, Philip S. Yu
| Challenge: | Efficient instruction tuning aims to enhance the ultimate performance of large language models (LLMs) current methods suffer from the curriculum rigidity, resulting in a fixed and potentially sub-optimal learning trajectory. |
| Approach: | a framework for efficient instruction tuning is proposed to address the issue of curriculum rigidity . current methods rely on static heuristic difficulty metrics and fail to adapt to evolving capabilities . |
| Outcome: | Efficient instruction tuning aims to enhance the ultimate performance of large language models . current methods suffer from the curriculum rigidity, resulting in a fixed learning trajectory . |
LLaMAX: Scaling Linguistic Horizons of LLM by Enhancing Translation Capabilities Beyond 100 Languages (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit exceptional translation capabilities in high-resource language tasks, yet their effectiveness in low-resourced languages is suboptimal. |
| Approach: | They conduct extensive multilingual continual pre-training on the LLaMA series models and develop LLiMAX for translation support across more than 100 languages. |
| Outcome: | The proposed model achieves higher translation performance than existing open-source models and performs on-par with specialized translation model on the Flores-101 benchmark. |
Lego-MT: Learning Detachable Models for Massively Multilingual Machine Translation (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing monolithic models for multilingual neural machine translation encounter parameter interference and inefficient inference for large models. |
| Approach: | They propose a detachable multi-way model that assigns each language to an individual branch . they use data from OPUS to build a translation benchmark covering 433 languages . |
| Outcome: | The proposed model outperforms existing models in OPUS and is faster than existing models. |
Improving Multi-label Emotion Classification by Integrating both General and Domain-specific Knowledge (D19-55)
Copied to clipboard
| Challenge: | Text in domains like social media has its own salient characteristics. |
| Approach: | They propose a method to obtain domain knowledge and integrate it with general knowledge to improve emotion classification. |
| Outcome: | The proposed method improves performance of emotion classification on Twitter data. |
CityCube: Benchmarking Cross-view Spatial Reasoning on Vision-Language Models in Urban Environments (2026.acl-long)
Copied to clipboard
Haotian Xu, Yue Hu, Zhengqiu Zhu, Chen Gao, Ziyou Wang, Junreng Rao, Wenhao Lu, Weishi Li, Quanjun Yin, Yong Li
| Challenge: | Existing benchmarks focus on indoor or street settings, overlooking challenges of open-ended urban spaces. |
| Approach: | They propose a benchmark to probe cross-view spatial reasoning capabilities of current VLMs in urban settings. |
| Outcome: | The citycube benchmark examines the performance of current vision-language models in urban environments. |
QSpec: Speculative Decoding with Complementary Quantization Schemes (2025.emnlp-main)
Copied to clipboard
| Challenge: | Quantization is widely adopted to accelerate inference and reduce memory consumption in large language models. |
| Approach: | They propose a quantization paradigm that decouples efficiency from quality by integrating two complementary schemes via speculative decoding. |
| Outcome: | The proposed approach achieves 1.64x speedup without quality degradation and outperforms state-of-the-art speculative decoding methods by 1.55x in batched settings. |
Reducing Token Redundancy in LVLMs: A Systematic Review of Token Pruning Methods (2026.acl-long)
Copied to clipboard
| Challenge: | Large Vision-Language Models (LVLMs) excel at visual understanding but face severe computational bottlenecks when processing high-resolution images and long videos due to massive visual token counts. |
| Approach: | They propose a taxonomy categorizing methods into vision-side, LLM-side and hybrid paradigms and analyze token selection mechanisms and pruning strategy. |
| Outcome: | The proposed method selectively removes less informative tokens while maintaining performance. |
A Survey of Deep Learning for Mathematical Reasoning (2023.acl-long)
Copied to clipboard
| Challenge: | a survey of deep learning for mathematical reasoning examines the field . a comprehensive reading list is provided to assist readers interested in the field. |
| Approach: | They present a survey of deep learning for mathematical reasoning over the past decade . they outline directions for future research and highlight potential for further exploration . |
| Outcome: | The proposed framework is based on the results of a decade-long survey of deep learning for mathematical reasoning. |
Enhancing Zero-Shot Chain-of-Thought Reasoning in Large Language Models through Logic (2024.lrec-main)
Copied to clipboard
| Challenge: | Experimental evaluations of large language models demonstrate the efficacy of enhanced reasoning by logic. |
| Approach: | They propose a framework that uses symbolic logic to verify and rectify reasoning steps by steps. |
| Outcome: | The proposed framework improves the zero-shot chain-of-thought reasoning ability of large language models by verifying and rectifying the reasoning steps step by step. |
Orthogonal Representation Editing: Decoupling Semantic Entanglement in Batch Knowledge Editing of LLMs (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing knowledge editing methods suffer from performance degradation in batch knowledge editing. |
| Approach: | They propose an orthogonal representation editing method which decouples semantic entanglement from edit vectors and enforcing orthogonals on edit vector. |
| Outcome: | The proposed method outperforms existing methods and achieves superior performance in cross-lingual knowledge editing scenarios. |
RAISE: Reinforced Adaptive Instruction Selection For Large Language Models (2025.findings-emnlp)
Copied to clipboard
Qingsong Lv, Yangning Li, Zihua Lan, Zishan Xu, Jiwei Tang, Tingwei Lu, Yinghui Li, Wenhao Jiang, Hong-Gee Kim, Hai-Tao Zheng, Philip S. Yu
| Challenge: | Existing selection methods rely on static, heuristic quality scores and are executed only once before training. |
| Approach: | They propose a dynamic selection framework that integrates selection into every training step. |
| Outcome: | The proposed framework integrates selection into every training step. |